1 /* SPDX-License-Identifier: GPL-2.0 */
3 * Copyright (C) 2001 Jens Axboe <axboe@suse.de>
8 #include <linux/highmem.h>
9 #include <linux/mempool.h>
10 #include <linux/ioprio.h>
11 /* struct bio, bio_vec and BIO_* flags are defined in blk_types.h */
12 #include <linux/blk_types.h>
13 #include <linux/uio.h>
18 #define BIO_BUG_ON BUG_ON
23 #define BIO_MAX_PAGES 256
25 #define bio_prio(bio) (bio)->bi_ioprio
26 #define bio_set_prio(bio, prio) ((bio)->bi_ioprio = prio)
28 #define bio_iter_iovec(bio, iter) \
29 bvec_iter_bvec((bio)->bi_io_vec, (iter))
31 #define bio_iter_page(bio, iter) \
32 bvec_iter_page((bio)->bi_io_vec, (iter))
33 #define bio_iter_len(bio, iter) \
34 bvec_iter_len((bio)->bi_io_vec, (iter))
35 #define bio_iter_offset(bio, iter) \
36 bvec_iter_offset((bio)->bi_io_vec, (iter))
38 #define bio_page(bio) bio_iter_page((bio), (bio)->bi_iter)
39 #define bio_offset(bio) bio_iter_offset((bio), (bio)->bi_iter)
40 #define bio_iovec(bio) bio_iter_iovec((bio), (bio)->bi_iter)
42 #define bio_multiple_segments(bio) \
43 ((bio)->bi_iter.bi_size != bio_iovec(bio).bv_len)
45 #define bvec_iter_sectors(iter) ((iter).bi_size >> 9)
46 #define bvec_iter_end_sector(iter) ((iter).bi_sector + bvec_iter_sectors((iter)))
48 #define bio_sectors(bio) bvec_iter_sectors((bio)->bi_iter)
49 #define bio_end_sector(bio) bvec_iter_end_sector((bio)->bi_iter)
52 * Return the data direction, READ or WRITE.
54 #define bio_data_dir(bio) \
55 (op_is_write(bio_op(bio)) ? WRITE : READ)
58 * Check whether this bio carries any data or not. A NULL bio is allowed.
60 static inline bool bio_has_data(struct bio *bio)
63 bio->bi_iter.bi_size &&
64 bio_op(bio) != REQ_OP_DISCARD &&
65 bio_op(bio) != REQ_OP_SECURE_ERASE &&
66 bio_op(bio) != REQ_OP_WRITE_ZEROES)
72 static inline bool bio_no_advance_iter(const struct bio *bio)
74 return bio_op(bio) == REQ_OP_DISCARD ||
75 bio_op(bio) == REQ_OP_SECURE_ERASE ||
76 bio_op(bio) == REQ_OP_WRITE_SAME ||
77 bio_op(bio) == REQ_OP_WRITE_ZEROES;
80 static inline bool bio_mergeable(struct bio *bio)
82 if (bio->bi_opf & REQ_NOMERGE_FLAGS)
88 static inline unsigned int bio_cur_bytes(struct bio *bio)
90 if (bio_has_data(bio))
91 return bio_iovec(bio).bv_len;
92 else /* dataless requests such as discard */
93 return bio->bi_iter.bi_size;
96 static inline void *bio_data(struct bio *bio)
98 if (bio_has_data(bio))
99 return page_address(bio_page(bio)) + bio_offset(bio);
105 * bio_full - check if the bio is full
107 * @len: length of one segment to be added
109 * Return true if @bio is full and one segment with @len bytes can't be
110 * added to the bio, otherwise return false
112 static inline bool bio_full(struct bio *bio, unsigned len)
114 if (bio->bi_vcnt >= bio->bi_max_vecs)
117 if (bio->bi_iter.bi_size > UINT_MAX - len)
123 static inline bool bio_next_segment(const struct bio *bio,
124 struct bvec_iter_all *iter)
126 if (iter->idx >= bio->bi_vcnt)
129 bvec_advance(&bio->bi_io_vec[iter->idx], iter);
134 * drivers should _never_ use the all version - the bio may have been split
135 * before it got to the driver and the driver won't own all of it
137 #define bio_for_each_segment_all(bvl, bio, iter) \
138 for (bvl = bvec_init_iter_all(&iter); bio_next_segment((bio), &iter); )
140 static inline void bio_advance_iter(const struct bio *bio,
141 struct bvec_iter *iter, unsigned int bytes)
143 iter->bi_sector += bytes >> 9;
145 if (bio_no_advance_iter(bio))
146 iter->bi_size -= bytes;
148 bvec_iter_advance(bio->bi_io_vec, iter, bytes);
149 /* TODO: It is reasonable to complete bio with error here. */
152 /* @bytes should be less or equal to bvec[i->bi_idx].bv_len */
153 static inline void bio_advance_iter_single(const struct bio *bio,
154 struct bvec_iter *iter,
157 iter->bi_sector += bytes >> 9;
159 if (bio_no_advance_iter(bio))
160 iter->bi_size -= bytes;
162 bvec_iter_advance_single(bio->bi_io_vec, iter, bytes);
165 #define __bio_for_each_segment(bvl, bio, iter, start) \
166 for (iter = (start); \
168 ((bvl = bio_iter_iovec((bio), (iter))), 1); \
169 bio_advance_iter_single((bio), &(iter), (bvl).bv_len))
171 #define bio_for_each_segment(bvl, bio, iter) \
172 __bio_for_each_segment(bvl, bio, iter, (bio)->bi_iter)
174 #define __bio_for_each_bvec(bvl, bio, iter, start) \
175 for (iter = (start); \
177 ((bvl = mp_bvec_iter_bvec((bio)->bi_io_vec, (iter))), 1); \
178 bio_advance_iter_single((bio), &(iter), (bvl).bv_len))
180 /* iterate over multi-page bvec */
181 #define bio_for_each_bvec(bvl, bio, iter) \
182 __bio_for_each_bvec(bvl, bio, iter, (bio)->bi_iter)
185 * Iterate over all multi-page bvecs. Drivers shouldn't use this version for the
186 * same reasons as bio_for_each_segment_all().
188 #define bio_for_each_bvec_all(bvl, bio, i) \
189 for (i = 0, bvl = bio_first_bvec_all(bio); \
190 i < (bio)->bi_vcnt; i++, bvl++) \
192 #define bio_iter_last(bvec, iter) ((iter).bi_size == (bvec).bv_len)
194 static inline unsigned bio_segments(struct bio *bio)
198 struct bvec_iter iter;
201 * We special case discard/write same/write zeroes, because they
202 * interpret bi_size differently:
205 switch (bio_op(bio)) {
207 case REQ_OP_SECURE_ERASE:
208 case REQ_OP_WRITE_ZEROES:
210 case REQ_OP_WRITE_SAME:
216 bio_for_each_segment(bv, bio, iter)
223 * get a reference to a bio, so it won't disappear. the intended use is
227 * submit_bio(rw, bio);
228 * if (bio->bi_flags ...)
232 * without the bio_get(), it could potentially complete I/O before submit_bio
233 * returns. and then bio would be freed memory when if (bio->bi_flags ...)
236 static inline void bio_get(struct bio *bio)
238 bio->bi_flags |= (1 << BIO_REFFED);
239 smp_mb__before_atomic();
240 atomic_inc(&bio->__bi_cnt);
243 static inline void bio_cnt_set(struct bio *bio, unsigned int count)
246 bio->bi_flags |= (1 << BIO_REFFED);
249 atomic_set(&bio->__bi_cnt, count);
252 static inline bool bio_flagged(struct bio *bio, unsigned int bit)
254 return (bio->bi_flags & (1U << bit)) != 0;
257 static inline void bio_set_flag(struct bio *bio, unsigned int bit)
259 bio->bi_flags |= (1U << bit);
262 static inline void bio_clear_flag(struct bio *bio, unsigned int bit)
264 bio->bi_flags &= ~(1U << bit);
267 static inline void bio_get_first_bvec(struct bio *bio, struct bio_vec *bv)
269 *bv = bio_iovec(bio);
272 static inline void bio_get_last_bvec(struct bio *bio, struct bio_vec *bv)
274 struct bvec_iter iter = bio->bi_iter;
277 if (unlikely(!bio_multiple_segments(bio))) {
278 *bv = bio_iovec(bio);
282 bio_advance_iter(bio, &iter, iter.bi_size);
284 if (!iter.bi_bvec_done)
285 idx = iter.bi_idx - 1;
286 else /* in the middle of bvec */
289 *bv = bio->bi_io_vec[idx];
292 * iter.bi_bvec_done records actual length of the last bvec
293 * if this bio ends in the middle of one io vector
295 if (iter.bi_bvec_done)
296 bv->bv_len = iter.bi_bvec_done;
299 static inline struct bio_vec *bio_first_bvec_all(struct bio *bio)
301 WARN_ON_ONCE(bio_flagged(bio, BIO_CLONED));
302 return bio->bi_io_vec;
305 static inline struct page *bio_first_page_all(struct bio *bio)
307 return bio_first_bvec_all(bio)->bv_page;
310 static inline struct bio_vec *bio_last_bvec_all(struct bio *bio)
312 WARN_ON_ONCE(bio_flagged(bio, BIO_CLONED));
313 return &bio->bi_io_vec[bio->bi_vcnt - 1];
317 BIP_BLOCK_INTEGRITY = 1 << 0, /* block layer owns integrity data */
318 BIP_MAPPED_INTEGRITY = 1 << 1, /* ref tag has been remapped */
319 BIP_CTRL_NOCHECK = 1 << 2, /* disable HBA integrity checking */
320 BIP_DISK_NOCHECK = 1 << 3, /* disable disk integrity checking */
321 BIP_IP_CHECKSUM = 1 << 4, /* IP checksum */
325 * bio integrity payload
327 struct bio_integrity_payload {
328 struct bio *bip_bio; /* parent bio */
330 struct bvec_iter bip_iter;
332 unsigned short bip_vcnt; /* # of integrity bio_vecs */
333 unsigned short bip_max_vcnt; /* integrity bio_vec slots */
334 unsigned short bip_flags; /* control flags */
336 struct bvec_iter bio_iter; /* for rewinding parent bio */
338 struct work_struct bip_work; /* I/O completion */
340 struct bio_vec *bip_vec;
341 struct bio_vec bip_inline_vecs[];/* embedded bvec array */
344 #if defined(CONFIG_BLK_DEV_INTEGRITY)
346 static inline struct bio_integrity_payload *bio_integrity(struct bio *bio)
348 if (bio->bi_opf & REQ_INTEGRITY)
349 return bio->bi_integrity;
354 static inline bool bio_integrity_flagged(struct bio *bio, enum bip_flags flag)
356 struct bio_integrity_payload *bip = bio_integrity(bio);
359 return bip->bip_flags & flag;
364 static inline sector_t bip_get_seed(struct bio_integrity_payload *bip)
366 return bip->bip_iter.bi_sector;
369 static inline void bip_set_seed(struct bio_integrity_payload *bip,
372 bip->bip_iter.bi_sector = seed;
375 #endif /* CONFIG_BLK_DEV_INTEGRITY */
377 extern void bio_trim(struct bio *bio, int offset, int size);
378 extern struct bio *bio_split(struct bio *bio, int sectors,
379 gfp_t gfp, struct bio_set *bs);
382 * bio_next_split - get next @sectors from a bio, splitting if necessary
384 * @sectors: number of sectors to split from the front of @bio
386 * @bs: bio set to allocate from
388 * Returns a bio representing the next @sectors of @bio - if the bio is smaller
389 * than @sectors, returns the original bio unchanged.
391 static inline struct bio *bio_next_split(struct bio *bio, int sectors,
392 gfp_t gfp, struct bio_set *bs)
394 if (sectors >= bio_sectors(bio))
397 return bio_split(bio, sectors, gfp, bs);
401 BIOSET_NEED_BVECS = BIT(0),
402 BIOSET_NEED_RESCUER = BIT(1),
404 extern int bioset_init(struct bio_set *, unsigned int, unsigned int, int flags);
405 extern void bioset_exit(struct bio_set *);
406 extern int biovec_init_pool(mempool_t *pool, int pool_entries);
407 extern int bioset_init_from_src(struct bio_set *bs, struct bio_set *src);
409 struct bio *bio_alloc_bioset(gfp_t gfp, unsigned short nr_iovecs,
411 struct bio *bio_kmalloc(gfp_t gfp_mask, unsigned short nr_iovecs);
412 extern void bio_put(struct bio *);
414 extern void __bio_clone_fast(struct bio *, struct bio *);
415 extern struct bio *bio_clone_fast(struct bio *, gfp_t, struct bio_set *);
417 extern struct bio_set fs_bio_set;
419 static inline struct bio *bio_alloc(gfp_t gfp_mask, unsigned short nr_iovecs)
421 return bio_alloc_bioset(gfp_mask, nr_iovecs, &fs_bio_set);
424 extern blk_qc_t submit_bio(struct bio *);
426 extern void bio_endio(struct bio *);
428 static inline void bio_io_error(struct bio *bio)
430 bio->bi_status = BLK_STS_IOERR;
434 static inline void bio_wouldblock_error(struct bio *bio)
436 bio_set_flag(bio, BIO_QUIET);
437 bio->bi_status = BLK_STS_AGAIN;
442 * Calculate number of bvec segments that should be allocated to fit data
443 * pointed by @iter. If @iter is backed by bvec it's going to be reused
444 * instead of allocating a new one.
446 static inline int bio_iov_vecs_to_alloc(struct iov_iter *iter, int max_segs)
448 if (iov_iter_is_bvec(iter))
450 return iov_iter_npages(iter, max_segs);
453 struct request_queue;
455 extern int submit_bio_wait(struct bio *bio);
456 extern void bio_advance(struct bio *, unsigned);
458 extern void bio_init(struct bio *bio, struct bio_vec *table,
459 unsigned short max_vecs);
460 extern void bio_uninit(struct bio *);
461 extern void bio_reset(struct bio *);
462 void bio_chain(struct bio *, struct bio *);
464 extern int bio_add_page(struct bio *, struct page *, unsigned int,unsigned int);
465 extern int bio_add_pc_page(struct request_queue *, struct bio *, struct page *,
466 unsigned int, unsigned int);
467 int bio_add_zone_append_page(struct bio *bio, struct page *page,
468 unsigned int len, unsigned int offset);
469 bool __bio_try_merge_page(struct bio *bio, struct page *page,
470 unsigned int len, unsigned int off, bool *same_page);
471 void __bio_add_page(struct bio *bio, struct page *page,
472 unsigned int len, unsigned int off);
473 int bio_iov_iter_get_pages(struct bio *bio, struct iov_iter *iter);
474 void bio_release_pages(struct bio *bio, bool mark_dirty);
475 extern void bio_set_pages_dirty(struct bio *bio);
476 extern void bio_check_pages_dirty(struct bio *bio);
478 extern void bio_copy_data_iter(struct bio *dst, struct bvec_iter *dst_iter,
479 struct bio *src, struct bvec_iter *src_iter);
480 extern void bio_copy_data(struct bio *dst, struct bio *src);
481 extern void bio_list_copy_data(struct bio *dst, struct bio *src);
482 extern void bio_free_pages(struct bio *bio);
483 void zero_fill_bio_iter(struct bio *bio, struct bvec_iter iter);
484 void bio_truncate(struct bio *bio, unsigned new_size);
485 void guard_bio_eod(struct bio *bio);
487 static inline void zero_fill_bio(struct bio *bio)
489 zero_fill_bio_iter(bio, bio->bi_iter);
492 extern const char *bio_devname(struct bio *bio, char *buffer);
494 #define bio_set_dev(bio, bdev) \
496 bio_clear_flag(bio, BIO_REMAPPED); \
497 if ((bio)->bi_bdev != (bdev)) \
498 bio_clear_flag(bio, BIO_THROTTLED); \
499 (bio)->bi_bdev = (bdev); \
500 bio_associate_blkg(bio); \
503 #define bio_copy_dev(dst, src) \
505 bio_clear_flag(dst, BIO_REMAPPED); \
506 (dst)->bi_bdev = (src)->bi_bdev; \
507 bio_clone_blkg_association(dst, src); \
510 #define bio_dev(bio) \
511 disk_devt((bio)->bi_bdev->bd_disk)
513 #ifdef CONFIG_BLK_CGROUP
514 void bio_associate_blkg(struct bio *bio);
515 void bio_associate_blkg_from_css(struct bio *bio,
516 struct cgroup_subsys_state *css);
517 void bio_clone_blkg_association(struct bio *dst, struct bio *src);
518 #else /* CONFIG_BLK_CGROUP */
519 static inline void bio_associate_blkg(struct bio *bio) { }
520 static inline void bio_associate_blkg_from_css(struct bio *bio,
521 struct cgroup_subsys_state *css)
523 static inline void bio_clone_blkg_association(struct bio *dst,
525 #endif /* CONFIG_BLK_CGROUP */
527 #ifdef CONFIG_HIGHMEM
529 * remember never ever reenable interrupts between a bvec_kmap_irq and
532 static inline char *bvec_kmap_irq(struct bio_vec *bvec, unsigned long *flags)
537 * might not be a highmem page, but the preempt/irq count
538 * balancing is a lot nicer this way
540 local_irq_save(*flags);
541 addr = (unsigned long) kmap_atomic(bvec->bv_page);
543 BUG_ON(addr & ~PAGE_MASK);
545 return (char *) addr + bvec->bv_offset;
548 static inline void bvec_kunmap_irq(char *buffer, unsigned long *flags)
550 unsigned long ptr = (unsigned long) buffer & PAGE_MASK;
552 kunmap_atomic((void *) ptr);
553 local_irq_restore(*flags);
557 static inline char *bvec_kmap_irq(struct bio_vec *bvec, unsigned long *flags)
559 return page_address(bvec->bv_page) + bvec->bv_offset;
562 static inline void bvec_kunmap_irq(char *buffer, unsigned long *flags)
569 * BIO list management for use by remapping drivers (e.g. DM or MD) and loop.
571 * A bio_list anchors a singly-linked list of bios chained through the bi_next
572 * member of the bio. The bio_list also caches the last list member to allow
573 * fast access to the tail.
580 static inline int bio_list_empty(const struct bio_list *bl)
582 return bl->head == NULL;
585 static inline void bio_list_init(struct bio_list *bl)
587 bl->head = bl->tail = NULL;
590 #define BIO_EMPTY_LIST { NULL, NULL }
592 #define bio_list_for_each(bio, bl) \
593 for (bio = (bl)->head; bio; bio = bio->bi_next)
595 static inline unsigned bio_list_size(const struct bio_list *bl)
600 bio_list_for_each(bio, bl)
606 static inline void bio_list_add(struct bio_list *bl, struct bio *bio)
611 bl->tail->bi_next = bio;
618 static inline void bio_list_add_head(struct bio_list *bl, struct bio *bio)
620 bio->bi_next = bl->head;
628 static inline void bio_list_merge(struct bio_list *bl, struct bio_list *bl2)
634 bl->tail->bi_next = bl2->head;
636 bl->head = bl2->head;
638 bl->tail = bl2->tail;
641 static inline void bio_list_merge_head(struct bio_list *bl,
642 struct bio_list *bl2)
648 bl2->tail->bi_next = bl->head;
650 bl->tail = bl2->tail;
652 bl->head = bl2->head;
655 static inline struct bio *bio_list_peek(struct bio_list *bl)
660 static inline struct bio *bio_list_pop(struct bio_list *bl)
662 struct bio *bio = bl->head;
665 bl->head = bl->head->bi_next;
675 static inline struct bio *bio_list_get(struct bio_list *bl)
677 struct bio *bio = bl->head;
679 bl->head = bl->tail = NULL;
685 * Increment chain count for the bio. Make sure the CHAIN flag update
686 * is visible before the raised count.
688 static inline void bio_inc_remaining(struct bio *bio)
690 bio_set_flag(bio, BIO_CHAIN);
691 smp_mb__before_atomic();
692 atomic_inc(&bio->__bi_remaining);
696 * bio_set is used to allow other portions of the IO system to
697 * allocate their own private memory pools for bio and iovec structures.
698 * These memory pools in turn all allocate from the bio_slab
699 * and the bvec_slabs[].
701 #define BIO_POOL_SIZE 2
704 struct kmem_cache *bio_slab;
705 unsigned int front_pad;
709 #if defined(CONFIG_BLK_DEV_INTEGRITY)
710 mempool_t bio_integrity_pool;
711 mempool_t bvec_integrity_pool;
714 unsigned int back_pad;
716 * Deadlock avoidance for stacking block drivers: see comments in
717 * bio_alloc_bioset() for details
719 spinlock_t rescue_lock;
720 struct bio_list rescue_list;
721 struct work_struct rescue_work;
722 struct workqueue_struct *rescue_workqueue;
725 static inline bool bioset_initialized(struct bio_set *bs)
727 return bs->bio_slab != NULL;
730 #if defined(CONFIG_BLK_DEV_INTEGRITY)
732 #define bip_for_each_vec(bvl, bip, iter) \
733 for_each_bvec(bvl, (bip)->bip_vec, iter, (bip)->bip_iter)
735 #define bio_for_each_integrity_vec(_bvl, _bio, _iter) \
737 bip_for_each_vec(_bvl, _bio->bi_integrity, _iter)
739 extern struct bio_integrity_payload *bio_integrity_alloc(struct bio *, gfp_t, unsigned int);
740 extern int bio_integrity_add_page(struct bio *, struct page *, unsigned int, unsigned int);
741 extern bool bio_integrity_prep(struct bio *);
742 extern void bio_integrity_advance(struct bio *, unsigned int);
743 extern void bio_integrity_trim(struct bio *);
744 extern int bio_integrity_clone(struct bio *, struct bio *, gfp_t);
745 extern int bioset_integrity_create(struct bio_set *, int);
746 extern void bioset_integrity_free(struct bio_set *);
747 extern void bio_integrity_init(void);
749 #else /* CONFIG_BLK_DEV_INTEGRITY */
751 static inline void *bio_integrity(struct bio *bio)
756 static inline int bioset_integrity_create(struct bio_set *bs, int pool_size)
761 static inline void bioset_integrity_free (struct bio_set *bs)
766 static inline bool bio_integrity_prep(struct bio *bio)
771 static inline int bio_integrity_clone(struct bio *bio, struct bio *bio_src,
777 static inline void bio_integrity_advance(struct bio *bio,
778 unsigned int bytes_done)
783 static inline void bio_integrity_trim(struct bio *bio)
788 static inline void bio_integrity_init(void)
793 static inline bool bio_integrity_flagged(struct bio *bio, enum bip_flags flag)
798 static inline void *bio_integrity_alloc(struct bio * bio, gfp_t gfp,
801 return ERR_PTR(-EINVAL);
804 static inline int bio_integrity_add_page(struct bio *bio, struct page *page,
805 unsigned int len, unsigned int offset)
810 #endif /* CONFIG_BLK_DEV_INTEGRITY */
813 * Mark a bio as polled. Note that for async polled IO, the caller must
814 * expect -EWOULDBLOCK if we cannot allocate a request (or other resources).
815 * We cannot block waiting for requests on polled IO, as those completions
816 * must be found by the caller. This is different than IRQ driven IO, where
817 * it's safe to wait for IO to complete.
819 static inline void bio_set_polled(struct bio *bio, struct kiocb *kiocb)
821 bio->bi_opf |= REQ_HIPRI;
822 if (!is_sync_kiocb(kiocb))
823 bio->bi_opf |= REQ_NOWAIT;
826 #endif /* __LINUX_BIO_H */