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 logical_block_size;
332 unsigned int alignment_offset;
335 unsigned int max_discard_sectors;
336 unsigned int max_hw_discard_sectors;
337 unsigned int max_write_same_sectors;
338 unsigned int max_write_zeroes_sectors;
339 unsigned int discard_granularity;
340 unsigned int discard_alignment;
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);
360 unsigned int blkdev_nr_zones(struct gendisk *disk);
361 extern int blkdev_zone_mgmt(struct block_device *bdev, enum req_opf op,
362 sector_t sectors, sector_t nr_sectors,
364 extern int blk_revalidate_disk_zones(struct gendisk *disk);
366 extern int blkdev_report_zones_ioctl(struct block_device *bdev, fmode_t mode,
367 unsigned int cmd, unsigned long arg);
368 extern int blkdev_zone_mgmt_ioctl(struct block_device *bdev, fmode_t mode,
369 unsigned int cmd, unsigned long arg);
371 #else /* CONFIG_BLK_DEV_ZONED */
373 static inline unsigned int blkdev_nr_zones(struct gendisk *disk)
378 static inline int blkdev_report_zones_ioctl(struct block_device *bdev,
379 fmode_t mode, unsigned int cmd,
385 static inline int blkdev_zone_mgmt_ioctl(struct block_device *bdev,
386 fmode_t mode, unsigned int cmd,
392 #endif /* CONFIG_BLK_DEV_ZONED */
394 struct request_queue {
395 struct request *last_merge;
396 struct elevator_queue *elevator;
398 struct blk_queue_stats *stats;
399 struct rq_qos *rq_qos;
401 make_request_fn *make_request_fn;
402 dma_drain_needed_fn *dma_drain_needed;
404 const struct blk_mq_ops *mq_ops;
407 struct blk_mq_ctx __percpu *queue_ctx;
409 unsigned int queue_depth;
411 /* hw dispatch queues */
412 struct blk_mq_hw_ctx **queue_hw_ctx;
413 unsigned int nr_hw_queues;
415 struct backing_dev_info *backing_dev_info;
418 * The queue owner gets to use this for whatever they like.
419 * ll_rw_blk doesn't touch it.
424 * various queue flags, see QUEUE_* below
426 unsigned long queue_flags;
428 * Number of contexts that have called blk_set_pm_only(). If this
429 * counter is above zero then only RQF_PM and RQF_PREEMPT requests are
435 * ida allocated id for this queue. Used to index queues from
441 * queue needs bounce pages for pages above this limit
445 spinlock_t queue_lock;
455 struct kobject *mq_kobj;
457 #ifdef CONFIG_BLK_DEV_INTEGRITY
458 struct blk_integrity integrity;
459 #endif /* CONFIG_BLK_DEV_INTEGRITY */
464 unsigned int nr_pending;
470 unsigned long nr_requests; /* Max # of requests */
472 unsigned int dma_drain_size;
473 void *dma_drain_buffer;
474 unsigned int dma_pad_mask;
475 unsigned int dma_alignment;
477 unsigned int rq_timeout;
480 struct blk_stat_callback *poll_cb;
481 struct blk_rq_stat poll_stat[BLK_MQ_POLL_STATS_BKTS];
483 struct timer_list timeout;
484 struct work_struct timeout_work;
486 struct list_head icq_list;
487 #ifdef CONFIG_BLK_CGROUP
488 DECLARE_BITMAP (blkcg_pols, BLKCG_MAX_POLS);
489 struct blkcg_gq *root_blkg;
490 struct list_head blkg_list;
493 struct queue_limits limits;
495 unsigned int required_elevator_features;
497 #ifdef CONFIG_BLK_DEV_ZONED
499 * Zoned block device information for request dispatch control.
500 * nr_zones is the total number of zones of the device. This is always
501 * 0 for regular block devices. conv_zones_bitmap is a bitmap of nr_zones
502 * bits which indicates if a zone is conventional (bit set) or
503 * sequential (bit clear). seq_zones_wlock is a bitmap of nr_zones
504 * bits which indicates if a zone is write locked, that is, if a write
505 * request targeting the zone was dispatched. All three fields are
506 * initialized by the low level device driver (e.g. scsi/sd.c).
507 * Stacking drivers (device mappers) may or may not initialize
510 * Reads of this information must be protected with blk_queue_enter() /
511 * blk_queue_exit(). Modifying this information is only allowed while
512 * no requests are being processed. See also blk_mq_freeze_queue() and
513 * blk_mq_unfreeze_queue().
515 unsigned int nr_zones;
516 unsigned long *conv_zones_bitmap;
517 unsigned long *seq_zones_wlock;
518 #endif /* CONFIG_BLK_DEV_ZONED */
523 unsigned int sg_timeout;
524 unsigned int sg_reserved_size;
526 #ifdef CONFIG_BLK_DEV_IO_TRACE
527 struct blk_trace __rcu *blk_trace;
528 struct mutex blk_trace_mutex;
531 * for flush operations
533 struct blk_flush_queue *fq;
535 struct list_head requeue_list;
536 spinlock_t requeue_lock;
537 struct delayed_work requeue_work;
539 struct mutex sysfs_lock;
540 struct mutex sysfs_dir_lock;
543 * for reusing dead hctx instance in case of updating
546 struct list_head unused_hctx_list;
547 spinlock_t unused_hctx_lock;
551 #if defined(CONFIG_BLK_DEV_BSG)
552 struct bsg_class_device bsg_dev;
555 #ifdef CONFIG_BLK_DEV_THROTTLING
557 struct throtl_data *td;
559 struct rcu_head rcu_head;
560 wait_queue_head_t mq_freeze_wq;
562 * Protect concurrent access to q_usage_counter by
563 * percpu_ref_kill() and percpu_ref_reinit().
565 struct mutex mq_freeze_lock;
566 struct percpu_ref q_usage_counter;
568 struct blk_mq_tag_set *tag_set;
569 struct list_head tag_set_list;
570 struct bio_set bio_split;
572 #ifdef CONFIG_BLK_DEBUG_FS
573 struct dentry *debugfs_dir;
574 struct dentry *sched_debugfs_dir;
575 struct dentry *rqos_debugfs_dir;
578 bool mq_sysfs_init_done;
582 struct work_struct release_work;
584 #define BLK_MAX_WRITE_HINTS 5
585 u64 write_hints[BLK_MAX_WRITE_HINTS];
588 #define QUEUE_FLAG_STOPPED 0 /* queue is stopped */
589 #define QUEUE_FLAG_DYING 1 /* queue being torn down */
590 #define QUEUE_FLAG_NOMERGES 3 /* disable merge attempts */
591 #define QUEUE_FLAG_SAME_COMP 4 /* complete on same CPU-group */
592 #define QUEUE_FLAG_FAIL_IO 5 /* fake timeout */
593 #define QUEUE_FLAG_NONROT 6 /* non-rotational device (SSD) */
594 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
595 #define QUEUE_FLAG_IO_STAT 7 /* do disk/partitions IO accounting */
596 #define QUEUE_FLAG_DISCARD 8 /* supports DISCARD */
597 #define QUEUE_FLAG_NOXMERGES 9 /* No extended merges */
598 #define QUEUE_FLAG_ADD_RANDOM 10 /* Contributes to random pool */
599 #define QUEUE_FLAG_SECERASE 11 /* supports secure erase */
600 #define QUEUE_FLAG_SAME_FORCE 12 /* force complete on same CPU */
601 #define QUEUE_FLAG_DEAD 13 /* queue tear-down finished */
602 #define QUEUE_FLAG_INIT_DONE 14 /* queue is initialized */
603 #define QUEUE_FLAG_POLL 16 /* IO polling enabled if set */
604 #define QUEUE_FLAG_WC 17 /* Write back caching */
605 #define QUEUE_FLAG_FUA 18 /* device supports FUA writes */
606 #define QUEUE_FLAG_DAX 19 /* device supports DAX */
607 #define QUEUE_FLAG_STATS 20 /* track IO start and completion times */
608 #define QUEUE_FLAG_POLL_STATS 21 /* collecting stats for hybrid polling */
609 #define QUEUE_FLAG_REGISTERED 22 /* queue has been registered to a disk */
610 #define QUEUE_FLAG_SCSI_PASSTHROUGH 23 /* queue supports SCSI commands */
611 #define QUEUE_FLAG_QUIESCED 24 /* queue has been quiesced */
612 #define QUEUE_FLAG_PCI_P2PDMA 25 /* device supports PCI p2p requests */
613 #define QUEUE_FLAG_ZONE_RESETALL 26 /* supports Zone Reset All */
614 #define QUEUE_FLAG_RQ_ALLOC_TIME 27 /* record rq->alloc_time_ns */
616 #define QUEUE_FLAG_MQ_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
617 (1 << QUEUE_FLAG_SAME_COMP))
619 void blk_queue_flag_set(unsigned int flag, struct request_queue *q);
620 void blk_queue_flag_clear(unsigned int flag, struct request_queue *q);
621 bool blk_queue_flag_test_and_set(unsigned int flag, struct request_queue *q);
623 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
624 #define blk_queue_dying(q) test_bit(QUEUE_FLAG_DYING, &(q)->queue_flags)
625 #define blk_queue_dead(q) test_bit(QUEUE_FLAG_DEAD, &(q)->queue_flags)
626 #define blk_queue_init_done(q) test_bit(QUEUE_FLAG_INIT_DONE, &(q)->queue_flags)
627 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
628 #define blk_queue_noxmerges(q) \
629 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
630 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
631 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
632 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
633 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
634 #define blk_queue_zone_resetall(q) \
635 test_bit(QUEUE_FLAG_ZONE_RESETALL, &(q)->queue_flags)
636 #define blk_queue_secure_erase(q) \
637 (test_bit(QUEUE_FLAG_SECERASE, &(q)->queue_flags))
638 #define blk_queue_dax(q) test_bit(QUEUE_FLAG_DAX, &(q)->queue_flags)
639 #define blk_queue_scsi_passthrough(q) \
640 test_bit(QUEUE_FLAG_SCSI_PASSTHROUGH, &(q)->queue_flags)
641 #define blk_queue_pci_p2pdma(q) \
642 test_bit(QUEUE_FLAG_PCI_P2PDMA, &(q)->queue_flags)
643 #ifdef CONFIG_BLK_RQ_ALLOC_TIME
644 #define blk_queue_rq_alloc_time(q) \
645 test_bit(QUEUE_FLAG_RQ_ALLOC_TIME, &(q)->queue_flags)
647 #define blk_queue_rq_alloc_time(q) false
650 #define blk_noretry_request(rq) \
651 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
652 REQ_FAILFAST_DRIVER))
653 #define blk_queue_quiesced(q) test_bit(QUEUE_FLAG_QUIESCED, &(q)->queue_flags)
654 #define blk_queue_pm_only(q) atomic_read(&(q)->pm_only)
655 #define blk_queue_fua(q) test_bit(QUEUE_FLAG_FUA, &(q)->queue_flags)
656 #define blk_queue_registered(q) test_bit(QUEUE_FLAG_REGISTERED, &(q)->queue_flags)
658 extern void blk_set_pm_only(struct request_queue *q);
659 extern void blk_clear_pm_only(struct request_queue *q);
661 static inline bool blk_account_rq(struct request *rq)
663 return (rq->rq_flags & RQF_STARTED) && !blk_rq_is_passthrough(rq);
666 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
668 #define rq_data_dir(rq) (op_is_write(req_op(rq)) ? WRITE : READ)
670 #define rq_dma_dir(rq) \
671 (op_is_write(req_op(rq)) ? DMA_TO_DEVICE : DMA_FROM_DEVICE)
673 #define dma_map_bvec(dev, bv, dir, attrs) \
674 dma_map_page_attrs(dev, (bv)->bv_page, (bv)->bv_offset, (bv)->bv_len, \
677 static inline bool queue_is_mq(struct request_queue *q)
682 static inline enum blk_zoned_model
683 blk_queue_zoned_model(struct request_queue *q)
685 return q->limits.zoned;
688 static inline bool blk_queue_is_zoned(struct request_queue *q)
690 switch (blk_queue_zoned_model(q)) {
699 static inline sector_t blk_queue_zone_sectors(struct request_queue *q)
701 return blk_queue_is_zoned(q) ? q->limits.chunk_sectors : 0;
704 #ifdef CONFIG_BLK_DEV_ZONED
705 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
707 return blk_queue_is_zoned(q) ? q->nr_zones : 0;
710 static inline unsigned int blk_queue_zone_no(struct request_queue *q,
713 if (!blk_queue_is_zoned(q))
715 return sector >> ilog2(q->limits.chunk_sectors);
718 static inline bool blk_queue_zone_is_seq(struct request_queue *q,
721 if (!blk_queue_is_zoned(q))
723 if (!q->conv_zones_bitmap)
725 return !test_bit(blk_queue_zone_no(q, sector), q->conv_zones_bitmap);
727 #else /* CONFIG_BLK_DEV_ZONED */
728 static inline unsigned int blk_queue_nr_zones(struct request_queue *q)
732 #endif /* CONFIG_BLK_DEV_ZONED */
734 static inline bool rq_is_sync(struct request *rq)
736 return op_is_sync(rq->cmd_flags);
739 static inline bool rq_mergeable(struct request *rq)
741 if (blk_rq_is_passthrough(rq))
744 if (req_op(rq) == REQ_OP_FLUSH)
747 if (req_op(rq) == REQ_OP_WRITE_ZEROES)
750 if (rq->cmd_flags & REQ_NOMERGE_FLAGS)
752 if (rq->rq_flags & RQF_NOMERGE_FLAGS)
758 static inline bool blk_write_same_mergeable(struct bio *a, struct bio *b)
760 if (bio_page(a) == bio_page(b) &&
761 bio_offset(a) == bio_offset(b))
767 static inline unsigned int blk_queue_depth(struct request_queue *q)
770 return q->queue_depth;
772 return q->nr_requests;
775 extern unsigned long blk_max_low_pfn, blk_max_pfn;
778 * standard bounce addresses:
780 * BLK_BOUNCE_HIGH : bounce all highmem pages
781 * BLK_BOUNCE_ANY : don't bounce anything
782 * BLK_BOUNCE_ISA : bounce pages above ISA DMA boundary
785 #if BITS_PER_LONG == 32
786 #define BLK_BOUNCE_HIGH ((u64)blk_max_low_pfn << PAGE_SHIFT)
788 #define BLK_BOUNCE_HIGH -1ULL
790 #define BLK_BOUNCE_ANY (-1ULL)
791 #define BLK_BOUNCE_ISA (DMA_BIT_MASK(24))
794 * default timeout for SG_IO if none specified
796 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
797 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
803 unsigned long offset;
808 struct req_iterator {
809 struct bvec_iter iter;
813 /* This should not be used directly - use rq_for_each_segment */
814 #define for_each_bio(_bio) \
815 for (; _bio; _bio = _bio->bi_next)
816 #define __rq_for_each_bio(_bio, rq) \
818 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
820 #define rq_for_each_segment(bvl, _rq, _iter) \
821 __rq_for_each_bio(_iter.bio, _rq) \
822 bio_for_each_segment(bvl, _iter.bio, _iter.iter)
824 #define rq_for_each_bvec(bvl, _rq, _iter) \
825 __rq_for_each_bio(_iter.bio, _rq) \
826 bio_for_each_bvec(bvl, _iter.bio, _iter.iter)
828 #define rq_iter_last(bvec, _iter) \
829 (_iter.bio->bi_next == NULL && \
830 bio_iter_last(bvec, _iter.iter))
832 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
833 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
835 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
836 extern void rq_flush_dcache_pages(struct request *rq);
838 static inline void rq_flush_dcache_pages(struct request *rq)
843 extern int blk_register_queue(struct gendisk *disk);
844 extern void blk_unregister_queue(struct gendisk *disk);
845 extern blk_qc_t generic_make_request(struct bio *bio);
846 extern blk_qc_t direct_make_request(struct bio *bio);
847 extern void blk_rq_init(struct request_queue *q, struct request *rq);
848 extern void blk_put_request(struct request *);
849 extern struct request *blk_get_request(struct request_queue *, unsigned int op,
850 blk_mq_req_flags_t flags);
851 extern int blk_lld_busy(struct request_queue *q);
852 extern int blk_rq_prep_clone(struct request *rq, struct request *rq_src,
853 struct bio_set *bs, gfp_t gfp_mask,
854 int (*bio_ctr)(struct bio *, struct bio *, void *),
856 extern void blk_rq_unprep_clone(struct request *rq);
857 extern blk_status_t blk_insert_cloned_request(struct request_queue *q,
859 extern int blk_rq_append_bio(struct request *rq, struct bio **bio);
860 extern void blk_queue_split(struct request_queue *, struct bio **);
861 extern int scsi_verify_blk_ioctl(struct block_device *, unsigned int);
862 extern int scsi_cmd_blk_ioctl(struct block_device *, fmode_t,
863 unsigned int, void __user *);
864 extern int scsi_cmd_ioctl(struct request_queue *, struct gendisk *, fmode_t,
865 unsigned int, void __user *);
866 extern int sg_scsi_ioctl(struct request_queue *, struct gendisk *, fmode_t,
867 struct scsi_ioctl_command __user *);
868 extern int get_sg_io_hdr(struct sg_io_hdr *hdr, const void __user *argp);
869 extern int put_sg_io_hdr(const struct sg_io_hdr *hdr, void __user *argp);
871 extern int blk_queue_enter(struct request_queue *q, blk_mq_req_flags_t flags);
872 extern void blk_queue_exit(struct request_queue *q);
873 extern void blk_sync_queue(struct request_queue *q);
874 extern int blk_rq_map_user(struct request_queue *, struct request *,
875 struct rq_map_data *, void __user *, unsigned long,
877 extern int blk_rq_unmap_user(struct bio *);
878 extern int blk_rq_map_kern(struct request_queue *, struct request *, void *, unsigned int, gfp_t);
879 extern int blk_rq_map_user_iov(struct request_queue *, struct request *,
880 struct rq_map_data *, const struct iov_iter *,
882 extern void blk_execute_rq(struct request_queue *, struct gendisk *,
883 struct request *, int);
884 extern void blk_execute_rq_nowait(struct request_queue *, struct gendisk *,
885 struct request *, int, rq_end_io_fn *);
887 /* Helper to convert REQ_OP_XXX to its string format XXX */
888 extern const char *blk_op_str(unsigned int op);
890 int blk_status_to_errno(blk_status_t status);
891 blk_status_t errno_to_blk_status(int errno);
893 int blk_poll(struct request_queue *q, blk_qc_t cookie, bool spin);
895 static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
897 return bdev->bd_disk->queue; /* this is never NULL */
901 * The basic unit of block I/O is a sector. It is used in a number of contexts
902 * in Linux (blk, bio, genhd). The size of one sector is 512 = 2**9
903 * bytes. Variables of type sector_t represent an offset or size that is a
904 * multiple of 512 bytes. Hence these two constants.
907 #define SECTOR_SHIFT 9
910 #define SECTOR_SIZE (1 << SECTOR_SHIFT)
914 * blk_rq_pos() : the current sector
915 * blk_rq_bytes() : bytes left in the entire request
916 * blk_rq_cur_bytes() : bytes left in the current segment
917 * blk_rq_err_bytes() : bytes left till the next error boundary
918 * blk_rq_sectors() : sectors left in the entire request
919 * blk_rq_cur_sectors() : sectors left in the current segment
920 * blk_rq_stats_sectors() : sectors of the entire request used for stats
922 static inline sector_t blk_rq_pos(const struct request *rq)
927 static inline unsigned int blk_rq_bytes(const struct request *rq)
929 return rq->__data_len;
932 static inline int blk_rq_cur_bytes(const struct request *rq)
934 return rq->bio ? bio_cur_bytes(rq->bio) : 0;
937 extern unsigned int blk_rq_err_bytes(const struct request *rq);
939 static inline unsigned int blk_rq_sectors(const struct request *rq)
941 return blk_rq_bytes(rq) >> SECTOR_SHIFT;
944 static inline unsigned int blk_rq_cur_sectors(const struct request *rq)
946 return blk_rq_cur_bytes(rq) >> SECTOR_SHIFT;
949 static inline unsigned int blk_rq_stats_sectors(const struct request *rq)
951 return rq->stats_sectors;
954 #ifdef CONFIG_BLK_DEV_ZONED
955 static inline unsigned int blk_rq_zone_no(struct request *rq)
957 return blk_queue_zone_no(rq->q, blk_rq_pos(rq));
960 static inline unsigned int blk_rq_zone_is_seq(struct request *rq)
962 return blk_queue_zone_is_seq(rq->q, blk_rq_pos(rq));
964 #endif /* CONFIG_BLK_DEV_ZONED */
967 * Some commands like WRITE SAME have a payload or data transfer size which
968 * is different from the size of the request. Any driver that supports such
969 * commands using the RQF_SPECIAL_PAYLOAD flag needs to use this helper to
970 * calculate the data transfer size.
972 static inline unsigned int blk_rq_payload_bytes(struct request *rq)
974 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
975 return rq->special_vec.bv_len;
976 return blk_rq_bytes(rq);
980 * Return the first full biovec in the request. The caller needs to check that
981 * there are any bvecs before calling this helper.
983 static inline struct bio_vec req_bvec(struct request *rq)
985 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
986 return rq->special_vec;
987 return mp_bvec_iter_bvec(rq->bio->bi_io_vec, rq->bio->bi_iter);
990 static inline unsigned int blk_queue_get_max_sectors(struct request_queue *q,
993 if (unlikely(op == REQ_OP_DISCARD || op == REQ_OP_SECURE_ERASE))
994 return min(q->limits.max_discard_sectors,
995 UINT_MAX >> SECTOR_SHIFT);
997 if (unlikely(op == REQ_OP_WRITE_SAME))
998 return q->limits.max_write_same_sectors;
1000 if (unlikely(op == REQ_OP_WRITE_ZEROES))
1001 return q->limits.max_write_zeroes_sectors;
1003 return q->limits.max_sectors;
1007 * Return maximum size of a request at given offset. Only valid for
1008 * file system requests.
1010 static inline unsigned int blk_max_size_offset(struct request_queue *q,
1013 if (!q->limits.chunk_sectors)
1014 return q->limits.max_sectors;
1016 return min(q->limits.max_sectors, (unsigned int)(q->limits.chunk_sectors -
1017 (offset & (q->limits.chunk_sectors - 1))));
1020 static inline unsigned int blk_rq_get_max_sectors(struct request *rq,
1023 struct request_queue *q = rq->q;
1025 if (blk_rq_is_passthrough(rq))
1026 return q->limits.max_hw_sectors;
1028 if (!q->limits.chunk_sectors ||
1029 req_op(rq) == REQ_OP_DISCARD ||
1030 req_op(rq) == REQ_OP_SECURE_ERASE)
1031 return blk_queue_get_max_sectors(q, req_op(rq));
1033 return min(blk_max_size_offset(q, offset),
1034 blk_queue_get_max_sectors(q, req_op(rq)));
1037 static inline unsigned int blk_rq_count_bios(struct request *rq)
1039 unsigned int nr_bios = 0;
1042 __rq_for_each_bio(bio, rq)
1048 void blk_steal_bios(struct bio_list *list, struct request *rq);
1051 * Request completion related functions.
1053 * blk_update_request() completes given number of bytes and updates
1054 * the request without completing it.
1056 extern bool blk_update_request(struct request *rq, blk_status_t error,
1057 unsigned int nr_bytes);
1059 extern void __blk_complete_request(struct request *);
1060 extern void blk_abort_request(struct request *);
1063 * Access functions for manipulating queue properties
1065 extern void blk_cleanup_queue(struct request_queue *);
1066 extern void blk_queue_make_request(struct request_queue *, make_request_fn *);
1067 extern void blk_queue_bounce_limit(struct request_queue *, u64);
1068 extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int);
1069 extern void blk_queue_chunk_sectors(struct request_queue *, unsigned int);
1070 extern void blk_queue_max_segments(struct request_queue *, unsigned short);
1071 extern void blk_queue_max_discard_segments(struct request_queue *,
1073 extern void blk_queue_max_segment_size(struct request_queue *, unsigned int);
1074 extern void blk_queue_max_discard_sectors(struct request_queue *q,
1075 unsigned int max_discard_sectors);
1076 extern void blk_queue_max_write_same_sectors(struct request_queue *q,
1077 unsigned int max_write_same_sectors);
1078 extern void blk_queue_max_write_zeroes_sectors(struct request_queue *q,
1079 unsigned int max_write_same_sectors);
1080 extern void blk_queue_logical_block_size(struct request_queue *, unsigned int);
1081 extern void blk_queue_physical_block_size(struct request_queue *, unsigned int);
1082 extern void blk_queue_alignment_offset(struct request_queue *q,
1083 unsigned int alignment);
1084 extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min);
1085 extern void blk_queue_io_min(struct request_queue *q, unsigned int min);
1086 extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt);
1087 extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt);
1088 extern void blk_set_queue_depth(struct request_queue *q, unsigned int depth);
1089 extern void blk_set_default_limits(struct queue_limits *lim);
1090 extern void blk_set_stacking_limits(struct queue_limits *lim);
1091 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
1093 extern int bdev_stack_limits(struct queue_limits *t, struct block_device *bdev,
1095 extern void disk_stack_limits(struct gendisk *disk, struct block_device *bdev,
1097 extern void blk_queue_stack_limits(struct request_queue *t, struct request_queue *b);
1098 extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int);
1099 extern int blk_queue_dma_drain(struct request_queue *q,
1100 dma_drain_needed_fn *dma_drain_needed,
1101 void *buf, unsigned int size);
1102 extern void blk_queue_segment_boundary(struct request_queue *, unsigned long);
1103 extern void blk_queue_virt_boundary(struct request_queue *, unsigned long);
1104 extern void blk_queue_dma_alignment(struct request_queue *, int);
1105 extern void blk_queue_update_dma_alignment(struct request_queue *, int);
1106 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
1107 extern void blk_queue_write_cache(struct request_queue *q, bool enabled, bool fua);
1108 extern void blk_queue_required_elevator_features(struct request_queue *q,
1109 unsigned int features);
1110 extern bool blk_queue_can_use_dma_map_merging(struct request_queue *q,
1111 struct device *dev);
1114 * Number of physical segments as sent to the device.
1116 * Normally this is the number of discontiguous data segments sent by the
1117 * submitter. But for data-less command like discard we might have no
1118 * actual data segments submitted, but the driver might have to add it's
1119 * own special payload. In that case we still return 1 here so that this
1120 * special payload will be mapped.
1122 static inline unsigned short blk_rq_nr_phys_segments(struct request *rq)
1124 if (rq->rq_flags & RQF_SPECIAL_PAYLOAD)
1126 return rq->nr_phys_segments;
1130 * Number of discard segments (or ranges) the driver needs to fill in.
1131 * Each discard bio merged into a request is counted as one segment.
1133 static inline unsigned short blk_rq_nr_discard_segments(struct request *rq)
1135 return max_t(unsigned short, rq->nr_phys_segments, 1);
1138 extern int blk_rq_map_sg(struct request_queue *, struct request *, struct scatterlist *);
1139 extern void blk_dump_rq_flags(struct request *, char *);
1140 extern long nr_blockdev_pages(void);
1142 bool __must_check blk_get_queue(struct request_queue *);
1143 struct request_queue *blk_alloc_queue(gfp_t);
1144 struct request_queue *blk_alloc_queue_node(gfp_t gfp_mask, int node_id);
1145 extern void blk_put_queue(struct request_queue *);
1146 extern void blk_set_queue_dying(struct request_queue *);
1149 * blk_plug permits building a queue of related requests by holding the I/O
1150 * fragments for a short period. This allows merging of sequential requests
1151 * into single larger request. As the requests are moved from a per-task list to
1152 * the device's request_queue in a batch, this results in improved scalability
1153 * as the lock contention for request_queue lock is reduced.
1155 * It is ok not to disable preemption when adding the request to the plug list
1156 * or when attempting a merge, because blk_schedule_flush_list() will only flush
1157 * the plug list when the task sleeps by itself. For details, please see
1158 * schedule() where blk_schedule_flush_plug() is called.
1161 struct list_head mq_list; /* blk-mq requests */
1162 struct list_head cb_list; /* md requires an unplug callback */
1163 unsigned short rq_count;
1164 bool multiple_queues;
1166 #define BLK_MAX_REQUEST_COUNT 16
1167 #define BLK_PLUG_FLUSH_SIZE (128 * 1024)
1170 typedef void (*blk_plug_cb_fn)(struct blk_plug_cb *, bool);
1171 struct blk_plug_cb {
1172 struct list_head list;
1173 blk_plug_cb_fn callback;
1176 extern struct blk_plug_cb *blk_check_plugged(blk_plug_cb_fn unplug,
1177 void *data, int size);
1178 extern void blk_start_plug(struct blk_plug *);
1179 extern void blk_finish_plug(struct blk_plug *);
1180 extern void blk_flush_plug_list(struct blk_plug *, bool);
1182 static inline void blk_flush_plug(struct task_struct *tsk)
1184 struct blk_plug *plug = tsk->plug;
1187 blk_flush_plug_list(plug, false);
1190 static inline void blk_schedule_flush_plug(struct task_struct *tsk)
1192 struct blk_plug *plug = tsk->plug;
1195 blk_flush_plug_list(plug, true);
1198 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1200 struct blk_plug *plug = tsk->plug;
1203 (!list_empty(&plug->mq_list) ||
1204 !list_empty(&plug->cb_list));
1207 extern int blkdev_issue_flush(struct block_device *, gfp_t, sector_t *);
1208 extern int blkdev_issue_write_same(struct block_device *bdev, sector_t sector,
1209 sector_t nr_sects, gfp_t gfp_mask, struct page *page);
1211 #define BLKDEV_DISCARD_SECURE (1 << 0) /* issue a secure erase */
1213 extern int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1214 sector_t nr_sects, gfp_t gfp_mask, unsigned long flags);
1215 extern int __blkdev_issue_discard(struct block_device *bdev, sector_t sector,
1216 sector_t nr_sects, gfp_t gfp_mask, int flags,
1219 #define BLKDEV_ZERO_NOUNMAP (1 << 0) /* do not free blocks */
1220 #define BLKDEV_ZERO_NOFALLBACK (1 << 1) /* don't write explicit zeroes */
1222 extern int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1223 sector_t nr_sects, gfp_t gfp_mask, struct bio **biop,
1225 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
1226 sector_t nr_sects, gfp_t gfp_mask, unsigned flags);
1228 static inline int sb_issue_discard(struct super_block *sb, sector_t block,
1229 sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags)
1231 return blkdev_issue_discard(sb->s_bdev,
1232 block << (sb->s_blocksize_bits -
1234 nr_blocks << (sb->s_blocksize_bits -
1238 static inline int sb_issue_zeroout(struct super_block *sb, sector_t block,
1239 sector_t nr_blocks, gfp_t gfp_mask)
1241 return blkdev_issue_zeroout(sb->s_bdev,
1242 block << (sb->s_blocksize_bits -
1244 nr_blocks << (sb->s_blocksize_bits -
1249 extern int blk_verify_command(unsigned char *cmd, fmode_t mode);
1251 enum blk_default_limits {
1252 BLK_MAX_SEGMENTS = 128,
1253 BLK_SAFE_MAX_SECTORS = 255,
1254 BLK_DEF_MAX_SECTORS = 2560,
1255 BLK_MAX_SEGMENT_SIZE = 65536,
1256 BLK_SEG_BOUNDARY_MASK = 0xFFFFFFFFUL,
1259 static inline unsigned long queue_segment_boundary(const struct request_queue *q)
1261 return q->limits.seg_boundary_mask;
1264 static inline unsigned long queue_virt_boundary(const struct request_queue *q)
1266 return q->limits.virt_boundary_mask;
1269 static inline unsigned int queue_max_sectors(const struct request_queue *q)
1271 return q->limits.max_sectors;
1274 static inline unsigned int queue_max_hw_sectors(const struct request_queue *q)
1276 return q->limits.max_hw_sectors;
1279 static inline unsigned short queue_max_segments(const struct request_queue *q)
1281 return q->limits.max_segments;
1284 static inline unsigned short queue_max_discard_segments(const struct request_queue *q)
1286 return q->limits.max_discard_segments;
1289 static inline unsigned int queue_max_segment_size(const struct request_queue *q)
1291 return q->limits.max_segment_size;
1294 static inline unsigned queue_logical_block_size(const struct request_queue *q)
1298 if (q && q->limits.logical_block_size)
1299 retval = q->limits.logical_block_size;
1304 static inline unsigned int bdev_logical_block_size(struct block_device *bdev)
1306 return queue_logical_block_size(bdev_get_queue(bdev));
1309 static inline unsigned int queue_physical_block_size(const struct request_queue *q)
1311 return q->limits.physical_block_size;
1314 static inline unsigned int bdev_physical_block_size(struct block_device *bdev)
1316 return queue_physical_block_size(bdev_get_queue(bdev));
1319 static inline unsigned int queue_io_min(const struct request_queue *q)
1321 return q->limits.io_min;
1324 static inline int bdev_io_min(struct block_device *bdev)
1326 return queue_io_min(bdev_get_queue(bdev));
1329 static inline unsigned int queue_io_opt(const struct request_queue *q)
1331 return q->limits.io_opt;
1334 static inline int bdev_io_opt(struct block_device *bdev)
1336 return queue_io_opt(bdev_get_queue(bdev));
1339 static inline int queue_alignment_offset(const struct request_queue *q)
1341 if (q->limits.misaligned)
1344 return q->limits.alignment_offset;
1347 static inline int queue_limit_alignment_offset(struct queue_limits *lim, sector_t sector)
1349 unsigned int granularity = max(lim->physical_block_size, lim->io_min);
1350 unsigned int alignment = sector_div(sector, granularity >> SECTOR_SHIFT)
1353 return (granularity + lim->alignment_offset - alignment) % granularity;
1356 static inline int bdev_alignment_offset(struct block_device *bdev)
1358 struct request_queue *q = bdev_get_queue(bdev);
1360 if (q->limits.misaligned)
1363 if (bdev != bdev->bd_contains)
1364 return bdev->bd_part->alignment_offset;
1366 return q->limits.alignment_offset;
1369 static inline int queue_discard_alignment(const struct request_queue *q)
1371 if (q->limits.discard_misaligned)
1374 return q->limits.discard_alignment;
1377 static inline int queue_limit_discard_alignment(struct queue_limits *lim, sector_t sector)
1379 unsigned int alignment, granularity, offset;
1381 if (!lim->max_discard_sectors)
1384 /* Why are these in bytes, not sectors? */
1385 alignment = lim->discard_alignment >> SECTOR_SHIFT;
1386 granularity = lim->discard_granularity >> SECTOR_SHIFT;
1390 /* Offset of the partition start in 'granularity' sectors */
1391 offset = sector_div(sector, granularity);
1393 /* And why do we do this modulus *again* in blkdev_issue_discard()? */
1394 offset = (granularity + alignment - offset) % granularity;
1396 /* Turn it back into bytes, gaah */
1397 return offset << SECTOR_SHIFT;
1400 static inline int bdev_discard_alignment(struct block_device *bdev)
1402 struct request_queue *q = bdev_get_queue(bdev);
1404 if (bdev != bdev->bd_contains)
1405 return bdev->bd_part->discard_alignment;
1407 return q->limits.discard_alignment;
1410 static inline unsigned int bdev_write_same(struct block_device *bdev)
1412 struct request_queue *q = bdev_get_queue(bdev);
1415 return q->limits.max_write_same_sectors;
1420 static inline unsigned int bdev_write_zeroes_sectors(struct block_device *bdev)
1422 struct request_queue *q = bdev_get_queue(bdev);
1425 return q->limits.max_write_zeroes_sectors;
1430 static inline enum blk_zoned_model bdev_zoned_model(struct block_device *bdev)
1432 struct request_queue *q = bdev_get_queue(bdev);
1435 return blk_queue_zoned_model(q);
1437 return BLK_ZONED_NONE;
1440 static inline bool bdev_is_zoned(struct block_device *bdev)
1442 struct request_queue *q = bdev_get_queue(bdev);
1445 return blk_queue_is_zoned(q);
1450 static inline sector_t bdev_zone_sectors(struct block_device *bdev)
1452 struct request_queue *q = bdev_get_queue(bdev);
1455 return blk_queue_zone_sectors(q);
1459 static inline int queue_dma_alignment(const struct request_queue *q)
1461 return q ? q->dma_alignment : 511;
1464 static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr,
1467 unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask;
1468 return !(addr & alignment) && !(len & alignment);
1471 /* assumes size > 256 */
1472 static inline unsigned int blksize_bits(unsigned int size)
1474 unsigned int bits = 8;
1478 } while (size > 256);
1482 static inline unsigned int block_size(struct block_device *bdev)
1484 return bdev->bd_block_size;
1487 typedef struct {struct page *v;} Sector;
1489 unsigned char *read_dev_sector(struct block_device *, sector_t, Sector *);
1491 static inline void put_dev_sector(Sector p)
1496 int kblockd_schedule_work(struct work_struct *work);
1497 int kblockd_mod_delayed_work_on(int cpu, struct delayed_work *dwork, unsigned long delay);
1499 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1500 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1501 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1502 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1504 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1506 enum blk_integrity_flags {
1507 BLK_INTEGRITY_VERIFY = 1 << 0,
1508 BLK_INTEGRITY_GENERATE = 1 << 1,
1509 BLK_INTEGRITY_DEVICE_CAPABLE = 1 << 2,
1510 BLK_INTEGRITY_IP_CHECKSUM = 1 << 3,
1513 struct blk_integrity_iter {
1517 unsigned int data_size;
1518 unsigned short interval;
1519 const char *disk_name;
1522 typedef blk_status_t (integrity_processing_fn) (struct blk_integrity_iter *);
1523 typedef void (integrity_prepare_fn) (struct request *);
1524 typedef void (integrity_complete_fn) (struct request *, unsigned int);
1526 struct blk_integrity_profile {
1527 integrity_processing_fn *generate_fn;
1528 integrity_processing_fn *verify_fn;
1529 integrity_prepare_fn *prepare_fn;
1530 integrity_complete_fn *complete_fn;
1534 extern void blk_integrity_register(struct gendisk *, struct blk_integrity *);
1535 extern void blk_integrity_unregister(struct gendisk *);
1536 extern int blk_integrity_compare(struct gendisk *, struct gendisk *);
1537 extern int blk_rq_map_integrity_sg(struct request_queue *, struct bio *,
1538 struct scatterlist *);
1539 extern int blk_rq_count_integrity_sg(struct request_queue *, struct bio *);
1540 extern bool blk_integrity_merge_rq(struct request_queue *, struct request *,
1542 extern bool blk_integrity_merge_bio(struct request_queue *, struct request *,
1545 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1547 struct blk_integrity *bi = &disk->queue->integrity;
1556 struct blk_integrity *bdev_get_integrity(struct block_device *bdev)
1558 return blk_get_integrity(bdev->bd_disk);
1561 static inline bool blk_integrity_rq(struct request *rq)
1563 return rq->cmd_flags & REQ_INTEGRITY;
1566 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1569 q->limits.max_integrity_segments = segs;
1572 static inline unsigned short
1573 queue_max_integrity_segments(const struct request_queue *q)
1575 return q->limits.max_integrity_segments;
1579 * bio_integrity_intervals - Return number of integrity intervals for a bio
1580 * @bi: blk_integrity profile for device
1581 * @sectors: Size of the bio in 512-byte sectors
1583 * Description: The block layer calculates everything in 512 byte
1584 * sectors but integrity metadata is done in terms of the data integrity
1585 * interval size of the storage device. Convert the block layer sectors
1586 * to the appropriate number of integrity intervals.
1588 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1589 unsigned int sectors)
1591 return sectors >> (bi->interval_exp - 9);
1594 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1595 unsigned int sectors)
1597 return bio_integrity_intervals(bi, sectors) * bi->tuple_size;
1601 * Return the first bvec that contains integrity data. Only drivers that are
1602 * limited to a single integrity segment should use this helper.
1604 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1606 if (WARN_ON_ONCE(queue_max_integrity_segments(rq->q) > 1))
1608 return rq->bio->bi_integrity->bip_vec;
1611 #else /* CONFIG_BLK_DEV_INTEGRITY */
1614 struct block_device;
1616 struct blk_integrity;
1618 static inline int blk_integrity_rq(struct request *rq)
1622 static inline int blk_rq_count_integrity_sg(struct request_queue *q,
1627 static inline int blk_rq_map_integrity_sg(struct request_queue *q,
1629 struct scatterlist *s)
1633 static inline struct blk_integrity *bdev_get_integrity(struct block_device *b)
1637 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1641 static inline int blk_integrity_compare(struct gendisk *a, struct gendisk *b)
1645 static inline void blk_integrity_register(struct gendisk *d,
1646 struct blk_integrity *b)
1649 static inline void blk_integrity_unregister(struct gendisk *d)
1652 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1656 static inline unsigned short queue_max_integrity_segments(const struct request_queue *q)
1660 static inline bool blk_integrity_merge_rq(struct request_queue *rq,
1666 static inline bool blk_integrity_merge_bio(struct request_queue *rq,
1673 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
1674 unsigned int sectors)
1679 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
1680 unsigned int sectors)
1685 static inline struct bio_vec *rq_integrity_vec(struct request *rq)
1690 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1692 struct block_device_operations {
1693 int (*open) (struct block_device *, fmode_t);
1694 void (*release) (struct gendisk *, fmode_t);
1695 int (*rw_page)(struct block_device *, sector_t, struct page *, unsigned int);
1696 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1697 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1698 unsigned int (*check_events) (struct gendisk *disk,
1699 unsigned int clearing);
1700 /* ->media_changed() is DEPRECATED, use ->check_events() instead */
1701 int (*media_changed) (struct gendisk *);
1702 void (*unlock_native_capacity) (struct gendisk *);
1703 int (*revalidate_disk) (struct gendisk *);
1704 int (*getgeo)(struct block_device *, struct hd_geometry *);
1705 /* this callback is with swap_lock and sometimes page table lock held */
1706 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1707 int (*report_zones)(struct gendisk *, sector_t sector,
1708 unsigned int nr_zones, report_zones_cb cb, void *data);
1709 struct module *owner;
1710 const struct pr_ops *pr_ops;
1713 #ifdef CONFIG_COMPAT
1714 extern int blkdev_compat_ptr_ioctl(struct block_device *, fmode_t,
1715 unsigned int, unsigned long);
1717 #define blkdev_compat_ptr_ioctl NULL
1720 extern int __blkdev_driver_ioctl(struct block_device *, fmode_t, unsigned int,
1722 extern int bdev_read_page(struct block_device *, sector_t, struct page *);
1723 extern int bdev_write_page(struct block_device *, sector_t, struct page *,
1724 struct writeback_control *);
1726 #ifdef CONFIG_BLK_DEV_ZONED
1727 bool blk_req_needs_zone_write_lock(struct request *rq);
1728 void __blk_req_zone_write_lock(struct request *rq);
1729 void __blk_req_zone_write_unlock(struct request *rq);
1731 static inline void blk_req_zone_write_lock(struct request *rq)
1733 if (blk_req_needs_zone_write_lock(rq))
1734 __blk_req_zone_write_lock(rq);
1737 static inline void blk_req_zone_write_unlock(struct request *rq)
1739 if (rq->rq_flags & RQF_ZONE_WRITE_LOCKED)
1740 __blk_req_zone_write_unlock(rq);
1743 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1745 return rq->q->seq_zones_wlock &&
1746 test_bit(blk_rq_zone_no(rq), rq->q->seq_zones_wlock);
1749 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1751 if (!blk_req_needs_zone_write_lock(rq))
1753 return !blk_req_zone_is_write_locked(rq);
1756 static inline bool blk_req_needs_zone_write_lock(struct request *rq)
1761 static inline void blk_req_zone_write_lock(struct request *rq)
1765 static inline void blk_req_zone_write_unlock(struct request *rq)
1768 static inline bool blk_req_zone_is_write_locked(struct request *rq)
1773 static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
1777 #endif /* CONFIG_BLK_DEV_ZONED */
1779 #else /* CONFIG_BLOCK */
1781 struct block_device;
1784 * stubs for when the block layer is configured out
1786 #define buffer_heads_over_limit 0
1788 static inline long nr_blockdev_pages(void)
1796 static inline void blk_start_plug(struct blk_plug *plug)
1800 static inline void blk_finish_plug(struct blk_plug *plug)
1804 static inline void blk_flush_plug(struct task_struct *task)
1808 static inline void blk_schedule_flush_plug(struct task_struct *task)
1813 static inline bool blk_needs_flush_plug(struct task_struct *tsk)
1818 static inline int blkdev_issue_flush(struct block_device *bdev, gfp_t gfp_mask,
1819 sector_t *error_sector)
1824 #endif /* CONFIG_BLOCK */
1826 static inline void blk_wake_io_task(struct task_struct *waiter)
1829 * If we're polling, the task itself is doing the completions. For
1830 * that case, we don't need to signal a wakeup, it's enough to just
1831 * mark us as RUNNING.
1833 if (waiter == current)
1834 __set_current_state(TASK_RUNNING);
1836 wake_up_process(waiter);