1 /* SPDX-License-Identifier: GPL-2.0 */
3 * Copyright (C) 2001 Jens Axboe <axboe@suse.de>
8 #include <linux/mempool.h>
9 /* struct bio, bio_vec and BIO_* flags are defined in blk_types.h */
10 #include <linux/blk_types.h>
11 #include <linux/uio.h>
13 #define BIO_MAX_VECS 256U
15 static inline unsigned int bio_max_segs(unsigned int nr_segs)
17 return min(nr_segs, BIO_MAX_VECS);
20 #define bio_prio(bio) (bio)->bi_ioprio
21 #define bio_set_prio(bio, prio) ((bio)->bi_ioprio = prio)
23 #define bio_iter_iovec(bio, iter) \
24 bvec_iter_bvec((bio)->bi_io_vec, (iter))
26 #define bio_iter_page(bio, iter) \
27 bvec_iter_page((bio)->bi_io_vec, (iter))
28 #define bio_iter_len(bio, iter) \
29 bvec_iter_len((bio)->bi_io_vec, (iter))
30 #define bio_iter_offset(bio, iter) \
31 bvec_iter_offset((bio)->bi_io_vec, (iter))
33 #define bio_page(bio) bio_iter_page((bio), (bio)->bi_iter)
34 #define bio_offset(bio) bio_iter_offset((bio), (bio)->bi_iter)
35 #define bio_iovec(bio) bio_iter_iovec((bio), (bio)->bi_iter)
37 #define bvec_iter_sectors(iter) ((iter).bi_size >> 9)
38 #define bvec_iter_end_sector(iter) ((iter).bi_sector + bvec_iter_sectors((iter)))
40 #define bio_sectors(bio) bvec_iter_sectors((bio)->bi_iter)
41 #define bio_end_sector(bio) bvec_iter_end_sector((bio)->bi_iter)
44 * Return the data direction, READ or WRITE.
46 #define bio_data_dir(bio) \
47 (op_is_write(bio_op(bio)) ? WRITE : READ)
50 * Check whether this bio carries any data or not. A NULL bio is allowed.
52 static inline bool bio_has_data(struct bio *bio)
55 bio->bi_iter.bi_size &&
56 bio_op(bio) != REQ_OP_DISCARD &&
57 bio_op(bio) != REQ_OP_SECURE_ERASE &&
58 bio_op(bio) != REQ_OP_WRITE_ZEROES)
64 static inline bool bio_no_advance_iter(const struct bio *bio)
66 return bio_op(bio) == REQ_OP_DISCARD ||
67 bio_op(bio) == REQ_OP_SECURE_ERASE ||
68 bio_op(bio) == REQ_OP_WRITE_SAME ||
69 bio_op(bio) == REQ_OP_WRITE_ZEROES;
72 static inline void *bio_data(struct bio *bio)
74 if (bio_has_data(bio))
75 return page_address(bio_page(bio)) + bio_offset(bio);
80 static inline bool bio_next_segment(const struct bio *bio,
81 struct bvec_iter_all *iter)
83 if (iter->idx >= bio->bi_vcnt)
86 bvec_advance(&bio->bi_io_vec[iter->idx], iter);
91 * drivers should _never_ use the all version - the bio may have been split
92 * before it got to the driver and the driver won't own all of it
94 #define bio_for_each_segment_all(bvl, bio, iter) \
95 for (bvl = bvec_init_iter_all(&iter); bio_next_segment((bio), &iter); )
97 static inline void bio_advance_iter(const struct bio *bio,
98 struct bvec_iter *iter, unsigned int bytes)
100 iter->bi_sector += bytes >> 9;
102 if (bio_no_advance_iter(bio))
103 iter->bi_size -= bytes;
105 bvec_iter_advance(bio->bi_io_vec, iter, bytes);
106 /* TODO: It is reasonable to complete bio with error here. */
109 /* @bytes should be less or equal to bvec[i->bi_idx].bv_len */
110 static inline void bio_advance_iter_single(const struct bio *bio,
111 struct bvec_iter *iter,
114 iter->bi_sector += bytes >> 9;
116 if (bio_no_advance_iter(bio))
117 iter->bi_size -= bytes;
119 bvec_iter_advance_single(bio->bi_io_vec, iter, bytes);
122 void __bio_advance(struct bio *, unsigned bytes);
125 * bio_advance - increment/complete a bio by some number of bytes
126 * @bio: bio to advance
127 * @bytes: number of bytes to complete
129 * This updates bi_sector, bi_size and bi_idx; if the number of bytes to
130 * complete doesn't align with a bvec boundary, then bv_len and bv_offset will
131 * be updated on the last bvec as well.
133 * @bio will then represent the remaining, uncompleted portion of the io.
135 static inline void bio_advance(struct bio *bio, unsigned int nbytes)
137 if (nbytes == bio->bi_iter.bi_size) {
138 bio->bi_iter.bi_size = 0;
141 __bio_advance(bio, nbytes);
144 #define __bio_for_each_segment(bvl, bio, iter, start) \
145 for (iter = (start); \
147 ((bvl = bio_iter_iovec((bio), (iter))), 1); \
148 bio_advance_iter_single((bio), &(iter), (bvl).bv_len))
150 #define bio_for_each_segment(bvl, bio, iter) \
151 __bio_for_each_segment(bvl, bio, iter, (bio)->bi_iter)
153 #define __bio_for_each_bvec(bvl, bio, iter, start) \
154 for (iter = (start); \
156 ((bvl = mp_bvec_iter_bvec((bio)->bi_io_vec, (iter))), 1); \
157 bio_advance_iter_single((bio), &(iter), (bvl).bv_len))
159 /* iterate over multi-page bvec */
160 #define bio_for_each_bvec(bvl, bio, iter) \
161 __bio_for_each_bvec(bvl, bio, iter, (bio)->bi_iter)
164 * Iterate over all multi-page bvecs. Drivers shouldn't use this version for the
165 * same reasons as bio_for_each_segment_all().
167 #define bio_for_each_bvec_all(bvl, bio, i) \
168 for (i = 0, bvl = bio_first_bvec_all(bio); \
169 i < (bio)->bi_vcnt; i++, bvl++) \
171 #define bio_iter_last(bvec, iter) ((iter).bi_size == (bvec).bv_len)
173 static inline unsigned bio_segments(struct bio *bio)
177 struct bvec_iter iter;
180 * We special case discard/write same/write zeroes, because they
181 * interpret bi_size differently:
184 switch (bio_op(bio)) {
186 case REQ_OP_SECURE_ERASE:
187 case REQ_OP_WRITE_ZEROES:
189 case REQ_OP_WRITE_SAME:
195 bio_for_each_segment(bv, bio, iter)
202 * get a reference to a bio, so it won't disappear. the intended use is
206 * submit_bio(rw, bio);
207 * if (bio->bi_flags ...)
211 * without the bio_get(), it could potentially complete I/O before submit_bio
212 * returns. and then bio would be freed memory when if (bio->bi_flags ...)
215 static inline void bio_get(struct bio *bio)
217 bio->bi_flags |= (1 << BIO_REFFED);
218 smp_mb__before_atomic();
219 atomic_inc(&bio->__bi_cnt);
222 static inline void bio_cnt_set(struct bio *bio, unsigned int count)
225 bio->bi_flags |= (1 << BIO_REFFED);
228 atomic_set(&bio->__bi_cnt, count);
231 static inline bool bio_flagged(struct bio *bio, unsigned int bit)
233 return (bio->bi_flags & (1U << bit)) != 0;
236 static inline void bio_set_flag(struct bio *bio, unsigned int bit)
238 bio->bi_flags |= (1U << bit);
241 static inline void bio_clear_flag(struct bio *bio, unsigned int bit)
243 bio->bi_flags &= ~(1U << bit);
246 static inline struct bio_vec *bio_first_bvec_all(struct bio *bio)
248 WARN_ON_ONCE(bio_flagged(bio, BIO_CLONED));
249 return bio->bi_io_vec;
252 static inline struct page *bio_first_page_all(struct bio *bio)
254 return bio_first_bvec_all(bio)->bv_page;
257 static inline struct bio_vec *bio_last_bvec_all(struct bio *bio)
259 WARN_ON_ONCE(bio_flagged(bio, BIO_CLONED));
260 return &bio->bi_io_vec[bio->bi_vcnt - 1];
264 BIP_BLOCK_INTEGRITY = 1 << 0, /* block layer owns integrity data */
265 BIP_MAPPED_INTEGRITY = 1 << 1, /* ref tag has been remapped */
266 BIP_CTRL_NOCHECK = 1 << 2, /* disable HBA integrity checking */
267 BIP_DISK_NOCHECK = 1 << 3, /* disable disk integrity checking */
268 BIP_IP_CHECKSUM = 1 << 4, /* IP checksum */
272 * bio integrity payload
274 struct bio_integrity_payload {
275 struct bio *bip_bio; /* parent bio */
277 struct bvec_iter bip_iter;
279 unsigned short bip_vcnt; /* # of integrity bio_vecs */
280 unsigned short bip_max_vcnt; /* integrity bio_vec slots */
281 unsigned short bip_flags; /* control flags */
283 struct bvec_iter bio_iter; /* for rewinding parent bio */
285 struct work_struct bip_work; /* I/O completion */
287 struct bio_vec *bip_vec;
288 struct bio_vec bip_inline_vecs[];/* embedded bvec array */
291 #if defined(CONFIG_BLK_DEV_INTEGRITY)
293 static inline struct bio_integrity_payload *bio_integrity(struct bio *bio)
295 if (bio->bi_opf & REQ_INTEGRITY)
296 return bio->bi_integrity;
301 static inline bool bio_integrity_flagged(struct bio *bio, enum bip_flags flag)
303 struct bio_integrity_payload *bip = bio_integrity(bio);
306 return bip->bip_flags & flag;
311 static inline sector_t bip_get_seed(struct bio_integrity_payload *bip)
313 return bip->bip_iter.bi_sector;
316 static inline void bip_set_seed(struct bio_integrity_payload *bip,
319 bip->bip_iter.bi_sector = seed;
322 #endif /* CONFIG_BLK_DEV_INTEGRITY */
324 void bio_trim(struct bio *bio, sector_t offset, sector_t size);
325 extern struct bio *bio_split(struct bio *bio, int sectors,
326 gfp_t gfp, struct bio_set *bs);
329 * bio_next_split - get next @sectors from a bio, splitting if necessary
331 * @sectors: number of sectors to split from the front of @bio
333 * @bs: bio set to allocate from
335 * Returns a bio representing the next @sectors of @bio - if the bio is smaller
336 * than @sectors, returns the original bio unchanged.
338 static inline struct bio *bio_next_split(struct bio *bio, int sectors,
339 gfp_t gfp, struct bio_set *bs)
341 if (sectors >= bio_sectors(bio))
344 return bio_split(bio, sectors, gfp, bs);
348 BIOSET_NEED_BVECS = BIT(0),
349 BIOSET_NEED_RESCUER = BIT(1),
350 BIOSET_PERCPU_CACHE = BIT(2),
352 extern int bioset_init(struct bio_set *, unsigned int, unsigned int, int flags);
353 extern void bioset_exit(struct bio_set *);
354 extern int biovec_init_pool(mempool_t *pool, int pool_entries);
355 extern int bioset_init_from_src(struct bio_set *bs, struct bio_set *src);
357 struct bio *bio_alloc_bioset(gfp_t gfp, unsigned short nr_iovecs,
359 struct bio *bio_alloc_kiocb(struct kiocb *kiocb, unsigned short nr_vecs,
361 struct bio *bio_kmalloc(gfp_t gfp_mask, unsigned short nr_iovecs);
362 extern void bio_put(struct bio *);
364 extern void __bio_clone_fast(struct bio *, struct bio *);
365 extern struct bio *bio_clone_fast(struct bio *, gfp_t, struct bio_set *);
367 extern struct bio_set fs_bio_set;
369 static inline struct bio *bio_alloc(gfp_t gfp_mask, unsigned short nr_iovecs)
371 return bio_alloc_bioset(gfp_mask, nr_iovecs, &fs_bio_set);
374 void submit_bio(struct bio *bio);
376 extern void bio_endio(struct bio *);
378 static inline void bio_io_error(struct bio *bio)
380 bio->bi_status = BLK_STS_IOERR;
384 static inline void bio_wouldblock_error(struct bio *bio)
386 bio_set_flag(bio, BIO_QUIET);
387 bio->bi_status = BLK_STS_AGAIN;
392 * Calculate number of bvec segments that should be allocated to fit data
393 * pointed by @iter. If @iter is backed by bvec it's going to be reused
394 * instead of allocating a new one.
396 static inline int bio_iov_vecs_to_alloc(struct iov_iter *iter, int max_segs)
398 if (iov_iter_is_bvec(iter))
400 return iov_iter_npages(iter, max_segs);
403 struct request_queue;
405 extern int submit_bio_wait(struct bio *bio);
406 extern void bio_init(struct bio *bio, struct bio_vec *table,
407 unsigned short max_vecs);
408 extern void bio_uninit(struct bio *);
409 extern void bio_reset(struct bio *);
410 void bio_chain(struct bio *, struct bio *);
412 extern int bio_add_page(struct bio *, struct page *, unsigned int,unsigned int);
413 extern int bio_add_pc_page(struct request_queue *, struct bio *, struct page *,
414 unsigned int, unsigned int);
415 int bio_add_zone_append_page(struct bio *bio, struct page *page,
416 unsigned int len, unsigned int offset);
417 void __bio_add_page(struct bio *bio, struct page *page,
418 unsigned int len, unsigned int off);
419 int bio_iov_iter_get_pages(struct bio *bio, struct iov_iter *iter);
420 void bio_iov_bvec_set(struct bio *bio, struct iov_iter *iter);
421 void __bio_release_pages(struct bio *bio, bool mark_dirty);
422 extern void bio_set_pages_dirty(struct bio *bio);
423 extern void bio_check_pages_dirty(struct bio *bio);
425 extern void bio_copy_data_iter(struct bio *dst, struct bvec_iter *dst_iter,
426 struct bio *src, struct bvec_iter *src_iter);
427 extern void bio_copy_data(struct bio *dst, struct bio *src);
428 extern void bio_free_pages(struct bio *bio);
429 void guard_bio_eod(struct bio *bio);
430 void zero_fill_bio(struct bio *bio);
432 static inline void bio_release_pages(struct bio *bio, bool mark_dirty)
434 if (!bio_flagged(bio, BIO_NO_PAGE_REF))
435 __bio_release_pages(bio, mark_dirty);
438 extern const char *bio_devname(struct bio *bio, char *buffer);
440 #define bio_dev(bio) \
441 disk_devt((bio)->bi_bdev->bd_disk)
443 #ifdef CONFIG_BLK_CGROUP
444 void bio_associate_blkg(struct bio *bio);
445 void bio_associate_blkg_from_css(struct bio *bio,
446 struct cgroup_subsys_state *css);
447 void bio_clone_blkg_association(struct bio *dst, struct bio *src);
448 #else /* CONFIG_BLK_CGROUP */
449 static inline void bio_associate_blkg(struct bio *bio) { }
450 static inline void bio_associate_blkg_from_css(struct bio *bio,
451 struct cgroup_subsys_state *css)
453 static inline void bio_clone_blkg_association(struct bio *dst,
455 #endif /* CONFIG_BLK_CGROUP */
457 static inline void bio_set_dev(struct bio *bio, struct block_device *bdev)
459 bio_clear_flag(bio, BIO_REMAPPED);
460 if (bio->bi_bdev != bdev)
461 bio_clear_flag(bio, BIO_THROTTLED);
463 bio_associate_blkg(bio);
466 static inline void bio_copy_dev(struct bio *dst, struct bio *src)
468 bio_clear_flag(dst, BIO_REMAPPED);
469 dst->bi_bdev = src->bi_bdev;
470 bio_clone_blkg_association(dst, src);
474 * BIO list management for use by remapping drivers (e.g. DM or MD) and loop.
476 * A bio_list anchors a singly-linked list of bios chained through the bi_next
477 * member of the bio. The bio_list also caches the last list member to allow
478 * fast access to the tail.
485 static inline int bio_list_empty(const struct bio_list *bl)
487 return bl->head == NULL;
490 static inline void bio_list_init(struct bio_list *bl)
492 bl->head = bl->tail = NULL;
495 #define BIO_EMPTY_LIST { NULL, NULL }
497 #define bio_list_for_each(bio, bl) \
498 for (bio = (bl)->head; bio; bio = bio->bi_next)
500 static inline unsigned bio_list_size(const struct bio_list *bl)
505 bio_list_for_each(bio, bl)
511 static inline void bio_list_add(struct bio_list *bl, struct bio *bio)
516 bl->tail->bi_next = bio;
523 static inline void bio_list_add_head(struct bio_list *bl, struct bio *bio)
525 bio->bi_next = bl->head;
533 static inline void bio_list_merge(struct bio_list *bl, struct bio_list *bl2)
539 bl->tail->bi_next = bl2->head;
541 bl->head = bl2->head;
543 bl->tail = bl2->tail;
546 static inline void bio_list_merge_head(struct bio_list *bl,
547 struct bio_list *bl2)
553 bl2->tail->bi_next = bl->head;
555 bl->tail = bl2->tail;
557 bl->head = bl2->head;
560 static inline struct bio *bio_list_peek(struct bio_list *bl)
565 static inline struct bio *bio_list_pop(struct bio_list *bl)
567 struct bio *bio = bl->head;
570 bl->head = bl->head->bi_next;
580 static inline struct bio *bio_list_get(struct bio_list *bl)
582 struct bio *bio = bl->head;
584 bl->head = bl->tail = NULL;
590 * Increment chain count for the bio. Make sure the CHAIN flag update
591 * is visible before the raised count.
593 static inline void bio_inc_remaining(struct bio *bio)
595 bio_set_flag(bio, BIO_CHAIN);
596 smp_mb__before_atomic();
597 atomic_inc(&bio->__bi_remaining);
601 * bio_set is used to allow other portions of the IO system to
602 * allocate their own private memory pools for bio and iovec structures.
603 * These memory pools in turn all allocate from the bio_slab
604 * and the bvec_slabs[].
606 #define BIO_POOL_SIZE 2
609 struct kmem_cache *bio_slab;
610 unsigned int front_pad;
613 * per-cpu bio alloc cache
615 struct bio_alloc_cache __percpu *cache;
619 #if defined(CONFIG_BLK_DEV_INTEGRITY)
620 mempool_t bio_integrity_pool;
621 mempool_t bvec_integrity_pool;
624 unsigned int back_pad;
626 * Deadlock avoidance for stacking block drivers: see comments in
627 * bio_alloc_bioset() for details
629 spinlock_t rescue_lock;
630 struct bio_list rescue_list;
631 struct work_struct rescue_work;
632 struct workqueue_struct *rescue_workqueue;
635 * Hot un-plug notifier for the per-cpu cache, if used
637 struct hlist_node cpuhp_dead;
640 static inline bool bioset_initialized(struct bio_set *bs)
642 return bs->bio_slab != NULL;
645 #if defined(CONFIG_BLK_DEV_INTEGRITY)
647 #define bip_for_each_vec(bvl, bip, iter) \
648 for_each_bvec(bvl, (bip)->bip_vec, iter, (bip)->bip_iter)
650 #define bio_for_each_integrity_vec(_bvl, _bio, _iter) \
652 bip_for_each_vec(_bvl, _bio->bi_integrity, _iter)
654 extern struct bio_integrity_payload *bio_integrity_alloc(struct bio *, gfp_t, unsigned int);
655 extern int bio_integrity_add_page(struct bio *, struct page *, unsigned int, unsigned int);
656 extern bool bio_integrity_prep(struct bio *);
657 extern void bio_integrity_advance(struct bio *, unsigned int);
658 extern void bio_integrity_trim(struct bio *);
659 extern int bio_integrity_clone(struct bio *, struct bio *, gfp_t);
660 extern int bioset_integrity_create(struct bio_set *, int);
661 extern void bioset_integrity_free(struct bio_set *);
662 extern void bio_integrity_init(void);
664 #else /* CONFIG_BLK_DEV_INTEGRITY */
666 static inline void *bio_integrity(struct bio *bio)
671 static inline int bioset_integrity_create(struct bio_set *bs, int pool_size)
676 static inline void bioset_integrity_free (struct bio_set *bs)
681 static inline bool bio_integrity_prep(struct bio *bio)
686 static inline int bio_integrity_clone(struct bio *bio, struct bio *bio_src,
692 static inline void bio_integrity_advance(struct bio *bio,
693 unsigned int bytes_done)
698 static inline void bio_integrity_trim(struct bio *bio)
703 static inline void bio_integrity_init(void)
708 static inline bool bio_integrity_flagged(struct bio *bio, enum bip_flags flag)
713 static inline void *bio_integrity_alloc(struct bio * bio, gfp_t gfp,
716 return ERR_PTR(-EINVAL);
719 static inline int bio_integrity_add_page(struct bio *bio, struct page *page,
720 unsigned int len, unsigned int offset)
725 #endif /* CONFIG_BLK_DEV_INTEGRITY */
728 * Mark a bio as polled. Note that for async polled IO, the caller must
729 * expect -EWOULDBLOCK if we cannot allocate a request (or other resources).
730 * We cannot block waiting for requests on polled IO, as those completions
731 * must be found by the caller. This is different than IRQ driven IO, where
732 * it's safe to wait for IO to complete.
734 static inline void bio_set_polled(struct bio *bio, struct kiocb *kiocb)
736 bio->bi_opf |= REQ_POLLED;
737 if (!is_sync_kiocb(kiocb))
738 bio->bi_opf |= REQ_NOWAIT;
741 struct bio *blk_next_bio(struct bio *bio, unsigned int nr_pages, gfp_t gfp);
743 #endif /* __LINUX_BIO_H */