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>
17 #define BIO_BUG_ON BUG_ON
22 #define BIO_MAX_PAGES 256
24 #define bio_prio(bio) (bio)->bi_ioprio
25 #define bio_set_prio(bio, prio) ((bio)->bi_ioprio = prio)
27 #define bio_iter_iovec(bio, iter) \
28 bvec_iter_bvec((bio)->bi_io_vec, (iter))
30 #define bio_iter_page(bio, iter) \
31 bvec_iter_page((bio)->bi_io_vec, (iter))
32 #define bio_iter_len(bio, iter) \
33 bvec_iter_len((bio)->bi_io_vec, (iter))
34 #define bio_iter_offset(bio, iter) \
35 bvec_iter_offset((bio)->bi_io_vec, (iter))
37 #define bio_page(bio) bio_iter_page((bio), (bio)->bi_iter)
38 #define bio_offset(bio) bio_iter_offset((bio), (bio)->bi_iter)
39 #define bio_iovec(bio) bio_iter_iovec((bio), (bio)->bi_iter)
41 #define bio_multiple_segments(bio) \
42 ((bio)->bi_iter.bi_size != bio_iovec(bio).bv_len)
44 #define bvec_iter_sectors(iter) ((iter).bi_size >> 9)
45 #define bvec_iter_end_sector(iter) ((iter).bi_sector + bvec_iter_sectors((iter)))
47 #define bio_sectors(bio) bvec_iter_sectors((bio)->bi_iter)
48 #define bio_end_sector(bio) bvec_iter_end_sector((bio)->bi_iter)
51 * Return the data direction, READ or WRITE.
53 #define bio_data_dir(bio) \
54 (op_is_write(bio_op(bio)) ? WRITE : READ)
57 * Check whether this bio carries any data or not. A NULL bio is allowed.
59 static inline bool bio_has_data(struct bio *bio)
62 bio->bi_iter.bi_size &&
63 bio_op(bio) != REQ_OP_DISCARD &&
64 bio_op(bio) != REQ_OP_SECURE_ERASE &&
65 bio_op(bio) != REQ_OP_WRITE_ZEROES)
71 static inline bool bio_no_advance_iter(const struct bio *bio)
73 return bio_op(bio) == REQ_OP_DISCARD ||
74 bio_op(bio) == REQ_OP_SECURE_ERASE ||
75 bio_op(bio) == REQ_OP_WRITE_SAME ||
76 bio_op(bio) == REQ_OP_WRITE_ZEROES;
79 static inline bool bio_mergeable(struct bio *bio)
81 if (bio->bi_opf & REQ_NOMERGE_FLAGS)
87 static inline unsigned int bio_cur_bytes(struct bio *bio)
89 if (bio_has_data(bio))
90 return bio_iovec(bio).bv_len;
91 else /* dataless requests such as discard */
92 return bio->bi_iter.bi_size;
95 static inline void *bio_data(struct bio *bio)
97 if (bio_has_data(bio))
98 return page_address(bio_page(bio)) + bio_offset(bio);
104 * bio_full - check if the bio is full
106 * @len: length of one segment to be added
108 * Return true if @bio is full and one segment with @len bytes can't be
109 * added to the bio, otherwise return false
111 static inline bool bio_full(struct bio *bio, unsigned len)
113 if (bio->bi_vcnt >= bio->bi_max_vecs)
116 if (bio->bi_iter.bi_size > UINT_MAX - len)
122 static inline bool bio_next_segment(const struct bio *bio,
123 struct bvec_iter_all *iter)
125 if (iter->idx >= bio->bi_vcnt)
128 bvec_advance(&bio->bi_io_vec[iter->idx], iter);
133 * drivers should _never_ use the all version - the bio may have been split
134 * before it got to the driver and the driver won't own all of it
136 #define bio_for_each_segment_all(bvl, bio, iter) \
137 for (bvl = bvec_init_iter_all(&iter); bio_next_segment((bio), &iter); )
139 static inline void bio_advance_iter(const struct bio *bio,
140 struct bvec_iter *iter, unsigned int bytes)
142 iter->bi_sector += bytes >> 9;
144 if (bio_no_advance_iter(bio))
145 iter->bi_size -= bytes;
147 bvec_iter_advance(bio->bi_io_vec, iter, bytes);
148 /* TODO: It is reasonable to complete bio with error here. */
151 /* @bytes should be less or equal to bvec[i->bi_idx].bv_len */
152 static inline void bio_advance_iter_single(const struct bio *bio,
153 struct bvec_iter *iter,
156 iter->bi_sector += bytes >> 9;
158 if (bio_no_advance_iter(bio))
159 iter->bi_size -= bytes;
161 bvec_iter_advance_single(bio->bi_io_vec, iter, bytes);
164 #define __bio_for_each_segment(bvl, bio, iter, start) \
165 for (iter = (start); \
167 ((bvl = bio_iter_iovec((bio), (iter))), 1); \
168 bio_advance_iter_single((bio), &(iter), (bvl).bv_len))
170 #define bio_for_each_segment(bvl, bio, iter) \
171 __bio_for_each_segment(bvl, bio, iter, (bio)->bi_iter)
173 #define __bio_for_each_bvec(bvl, bio, iter, start) \
174 for (iter = (start); \
176 ((bvl = mp_bvec_iter_bvec((bio)->bi_io_vec, (iter))), 1); \
177 bio_advance_iter_single((bio), &(iter), (bvl).bv_len))
179 /* iterate over multi-page bvec */
180 #define bio_for_each_bvec(bvl, bio, iter) \
181 __bio_for_each_bvec(bvl, bio, iter, (bio)->bi_iter)
184 * Iterate over all multi-page bvecs. Drivers shouldn't use this version for the
185 * same reasons as bio_for_each_segment_all().
187 #define bio_for_each_bvec_all(bvl, bio, i) \
188 for (i = 0, bvl = bio_first_bvec_all(bio); \
189 i < (bio)->bi_vcnt; i++, bvl++) \
191 #define bio_iter_last(bvec, iter) ((iter).bi_size == (bvec).bv_len)
193 static inline unsigned bio_segments(struct bio *bio)
197 struct bvec_iter iter;
200 * We special case discard/write same/write zeroes, because they
201 * interpret bi_size differently:
204 switch (bio_op(bio)) {
206 case REQ_OP_SECURE_ERASE:
207 case REQ_OP_WRITE_ZEROES:
209 case REQ_OP_WRITE_SAME:
215 bio_for_each_segment(bv, bio, iter)
222 * get a reference to a bio, so it won't disappear. the intended use is
226 * submit_bio(rw, bio);
227 * if (bio->bi_flags ...)
231 * without the bio_get(), it could potentially complete I/O before submit_bio
232 * returns. and then bio would be freed memory when if (bio->bi_flags ...)
235 static inline void bio_get(struct bio *bio)
237 bio->bi_flags |= (1 << BIO_REFFED);
238 smp_mb__before_atomic();
239 atomic_inc(&bio->__bi_cnt);
242 static inline void bio_cnt_set(struct bio *bio, unsigned int count)
245 bio->bi_flags |= (1 << BIO_REFFED);
248 atomic_set(&bio->__bi_cnt, count);
251 static inline bool bio_flagged(struct bio *bio, unsigned int bit)
253 return (bio->bi_flags & (1U << bit)) != 0;
256 static inline void bio_set_flag(struct bio *bio, unsigned int bit)
258 bio->bi_flags |= (1U << bit);
261 static inline void bio_clear_flag(struct bio *bio, unsigned int bit)
263 bio->bi_flags &= ~(1U << bit);
266 static inline void bio_get_first_bvec(struct bio *bio, struct bio_vec *bv)
268 *bv = bio_iovec(bio);
271 static inline void bio_get_last_bvec(struct bio *bio, struct bio_vec *bv)
273 struct bvec_iter iter = bio->bi_iter;
276 if (unlikely(!bio_multiple_segments(bio))) {
277 *bv = bio_iovec(bio);
281 bio_advance_iter(bio, &iter, iter.bi_size);
283 if (!iter.bi_bvec_done)
284 idx = iter.bi_idx - 1;
285 else /* in the middle of bvec */
288 *bv = bio->bi_io_vec[idx];
291 * iter.bi_bvec_done records actual length of the last bvec
292 * if this bio ends in the middle of one io vector
294 if (iter.bi_bvec_done)
295 bv->bv_len = iter.bi_bvec_done;
298 static inline struct bio_vec *bio_first_bvec_all(struct bio *bio)
300 WARN_ON_ONCE(bio_flagged(bio, BIO_CLONED));
301 return bio->bi_io_vec;
304 static inline struct page *bio_first_page_all(struct bio *bio)
306 return bio_first_bvec_all(bio)->bv_page;
309 static inline struct bio_vec *bio_last_bvec_all(struct bio *bio)
311 WARN_ON_ONCE(bio_flagged(bio, BIO_CLONED));
312 return &bio->bi_io_vec[bio->bi_vcnt - 1];
316 BIP_BLOCK_INTEGRITY = 1 << 0, /* block layer owns integrity data */
317 BIP_MAPPED_INTEGRITY = 1 << 1, /* ref tag has been remapped */
318 BIP_CTRL_NOCHECK = 1 << 2, /* disable HBA integrity checking */
319 BIP_DISK_NOCHECK = 1 << 3, /* disable disk integrity checking */
320 BIP_IP_CHECKSUM = 1 << 4, /* IP checksum */
324 * bio integrity payload
326 struct bio_integrity_payload {
327 struct bio *bip_bio; /* parent bio */
329 struct bvec_iter bip_iter;
331 unsigned short bip_slab; /* slab the bip came from */
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 extern struct bio *bio_alloc_bioset(gfp_t, unsigned int, struct bio_set *);
410 extern void bio_put(struct bio *);
412 extern void __bio_clone_fast(struct bio *, struct bio *);
413 extern struct bio *bio_clone_fast(struct bio *, gfp_t, struct bio_set *);
415 extern struct bio_set fs_bio_set;
417 static inline struct bio *bio_alloc(gfp_t gfp_mask, unsigned int nr_iovecs)
419 return bio_alloc_bioset(gfp_mask, nr_iovecs, &fs_bio_set);
422 static inline struct bio *bio_kmalloc(gfp_t gfp_mask, unsigned int nr_iovecs)
424 return bio_alloc_bioset(gfp_mask, nr_iovecs, NULL);
427 extern blk_qc_t submit_bio(struct bio *);
429 extern void bio_endio(struct bio *);
431 static inline void bio_io_error(struct bio *bio)
433 bio->bi_status = BLK_STS_IOERR;
437 static inline void bio_wouldblock_error(struct bio *bio)
439 bio_set_flag(bio, BIO_QUIET);
440 bio->bi_status = BLK_STS_AGAIN;
444 struct request_queue;
446 extern int submit_bio_wait(struct bio *bio);
447 extern void bio_advance(struct bio *, unsigned);
449 extern void bio_init(struct bio *bio, struct bio_vec *table,
450 unsigned short max_vecs);
451 extern void bio_uninit(struct bio *);
452 extern void bio_reset(struct bio *);
453 void bio_chain(struct bio *, struct bio *);
455 extern int bio_add_page(struct bio *, struct page *, unsigned int,unsigned int);
456 extern int bio_add_pc_page(struct request_queue *, struct bio *, struct page *,
457 unsigned int, unsigned int);
458 bool __bio_try_merge_page(struct bio *bio, struct page *page,
459 unsigned int len, unsigned int off, bool *same_page);
460 void __bio_add_page(struct bio *bio, struct page *page,
461 unsigned int len, unsigned int off);
462 int bio_iov_iter_get_pages(struct bio *bio, struct iov_iter *iter);
463 void bio_release_pages(struct bio *bio, bool mark_dirty);
464 extern void bio_set_pages_dirty(struct bio *bio);
465 extern void bio_check_pages_dirty(struct bio *bio);
467 extern void bio_copy_data_iter(struct bio *dst, struct bvec_iter *dst_iter,
468 struct bio *src, struct bvec_iter *src_iter);
469 extern void bio_copy_data(struct bio *dst, struct bio *src);
470 extern void bio_list_copy_data(struct bio *dst, struct bio *src);
471 extern void bio_free_pages(struct bio *bio);
472 void zero_fill_bio_iter(struct bio *bio, struct bvec_iter iter);
473 void bio_truncate(struct bio *bio, unsigned new_size);
474 void guard_bio_eod(struct bio *bio);
476 static inline void zero_fill_bio(struct bio *bio)
478 zero_fill_bio_iter(bio, bio->bi_iter);
481 extern struct bio_vec *bvec_alloc(gfp_t, int, unsigned long *, mempool_t *);
482 extern void bvec_free(mempool_t *, struct bio_vec *, unsigned int);
483 extern unsigned int bvec_nr_vecs(unsigned short idx);
484 extern const char *bio_devname(struct bio *bio, char *buffer);
486 #define bio_set_dev(bio, bdev) \
488 if ((bio)->bi_disk != (bdev)->bd_disk) \
489 bio_clear_flag(bio, BIO_THROTTLED);\
490 (bio)->bi_disk = (bdev)->bd_disk; \
491 (bio)->bi_partno = (bdev)->bd_partno; \
492 bio_associate_blkg(bio); \
495 #define bio_copy_dev(dst, src) \
497 (dst)->bi_disk = (src)->bi_disk; \
498 (dst)->bi_partno = (src)->bi_partno; \
499 bio_clone_blkg_association(dst, src); \
502 #define bio_dev(bio) \
503 disk_devt((bio)->bi_disk)
505 #ifdef CONFIG_BLK_CGROUP
506 void bio_associate_blkg(struct bio *bio);
507 void bio_associate_blkg_from_css(struct bio *bio,
508 struct cgroup_subsys_state *css);
509 void bio_clone_blkg_association(struct bio *dst, struct bio *src);
510 #else /* CONFIG_BLK_CGROUP */
511 static inline void bio_associate_blkg(struct bio *bio) { }
512 static inline void bio_associate_blkg_from_css(struct bio *bio,
513 struct cgroup_subsys_state *css)
515 static inline void bio_clone_blkg_association(struct bio *dst,
517 #endif /* CONFIG_BLK_CGROUP */
519 #ifdef CONFIG_HIGHMEM
521 * remember never ever reenable interrupts between a bvec_kmap_irq and
524 static inline char *bvec_kmap_irq(struct bio_vec *bvec, unsigned long *flags)
529 * might not be a highmem page, but the preempt/irq count
530 * balancing is a lot nicer this way
532 local_irq_save(*flags);
533 addr = (unsigned long) kmap_atomic(bvec->bv_page);
535 BUG_ON(addr & ~PAGE_MASK);
537 return (char *) addr + bvec->bv_offset;
540 static inline void bvec_kunmap_irq(char *buffer, unsigned long *flags)
542 unsigned long ptr = (unsigned long) buffer & PAGE_MASK;
544 kunmap_atomic((void *) ptr);
545 local_irq_restore(*flags);
549 static inline char *bvec_kmap_irq(struct bio_vec *bvec, unsigned long *flags)
551 return page_address(bvec->bv_page) + bvec->bv_offset;
554 static inline void bvec_kunmap_irq(char *buffer, unsigned long *flags)
561 * BIO list management for use by remapping drivers (e.g. DM or MD) and loop.
563 * A bio_list anchors a singly-linked list of bios chained through the bi_next
564 * member of the bio. The bio_list also caches the last list member to allow
565 * fast access to the tail.
572 static inline int bio_list_empty(const struct bio_list *bl)
574 return bl->head == NULL;
577 static inline void bio_list_init(struct bio_list *bl)
579 bl->head = bl->tail = NULL;
582 #define BIO_EMPTY_LIST { NULL, NULL }
584 #define bio_list_for_each(bio, bl) \
585 for (bio = (bl)->head; bio; bio = bio->bi_next)
587 static inline unsigned bio_list_size(const struct bio_list *bl)
592 bio_list_for_each(bio, bl)
598 static inline void bio_list_add(struct bio_list *bl, struct bio *bio)
603 bl->tail->bi_next = bio;
610 static inline void bio_list_add_head(struct bio_list *bl, struct bio *bio)
612 bio->bi_next = bl->head;
620 static inline void bio_list_merge(struct bio_list *bl, struct bio_list *bl2)
626 bl->tail->bi_next = bl2->head;
628 bl->head = bl2->head;
630 bl->tail = bl2->tail;
633 static inline void bio_list_merge_head(struct bio_list *bl,
634 struct bio_list *bl2)
640 bl2->tail->bi_next = bl->head;
642 bl->tail = bl2->tail;
644 bl->head = bl2->head;
647 static inline struct bio *bio_list_peek(struct bio_list *bl)
652 static inline struct bio *bio_list_pop(struct bio_list *bl)
654 struct bio *bio = bl->head;
657 bl->head = bl->head->bi_next;
667 static inline struct bio *bio_list_get(struct bio_list *bl)
669 struct bio *bio = bl->head;
671 bl->head = bl->tail = NULL;
677 * Increment chain count for the bio. Make sure the CHAIN flag update
678 * is visible before the raised count.
680 static inline void bio_inc_remaining(struct bio *bio)
682 bio_set_flag(bio, BIO_CHAIN);
683 smp_mb__before_atomic();
684 atomic_inc(&bio->__bi_remaining);
688 * bio_set is used to allow other portions of the IO system to
689 * allocate their own private memory pools for bio and iovec structures.
690 * These memory pools in turn all allocate from the bio_slab
691 * and the bvec_slabs[].
693 #define BIO_POOL_SIZE 2
696 struct kmem_cache *bio_slab;
697 unsigned int front_pad;
701 #if defined(CONFIG_BLK_DEV_INTEGRITY)
702 mempool_t bio_integrity_pool;
703 mempool_t bvec_integrity_pool;
707 * Deadlock avoidance for stacking block drivers: see comments in
708 * bio_alloc_bioset() for details
710 spinlock_t rescue_lock;
711 struct bio_list rescue_list;
712 struct work_struct rescue_work;
713 struct workqueue_struct *rescue_workqueue;
719 struct kmem_cache *slab;
722 static inline bool bioset_initialized(struct bio_set *bs)
724 return bs->bio_slab != NULL;
727 #if defined(CONFIG_BLK_DEV_INTEGRITY)
729 #define bip_for_each_vec(bvl, bip, iter) \
730 for_each_bvec(bvl, (bip)->bip_vec, iter, (bip)->bip_iter)
732 #define bio_for_each_integrity_vec(_bvl, _bio, _iter) \
734 bip_for_each_vec(_bvl, _bio->bi_integrity, _iter)
736 extern struct bio_integrity_payload *bio_integrity_alloc(struct bio *, gfp_t, unsigned int);
737 extern int bio_integrity_add_page(struct bio *, struct page *, unsigned int, unsigned int);
738 extern bool bio_integrity_prep(struct bio *);
739 extern void bio_integrity_advance(struct bio *, unsigned int);
740 extern void bio_integrity_trim(struct bio *);
741 extern int bio_integrity_clone(struct bio *, struct bio *, gfp_t);
742 extern int bioset_integrity_create(struct bio_set *, int);
743 extern void bioset_integrity_free(struct bio_set *);
744 extern void bio_integrity_init(void);
746 #else /* CONFIG_BLK_DEV_INTEGRITY */
748 static inline void *bio_integrity(struct bio *bio)
753 static inline int bioset_integrity_create(struct bio_set *bs, int pool_size)
758 static inline void bioset_integrity_free (struct bio_set *bs)
763 static inline bool bio_integrity_prep(struct bio *bio)
768 static inline int bio_integrity_clone(struct bio *bio, struct bio *bio_src,
774 static inline void bio_integrity_advance(struct bio *bio,
775 unsigned int bytes_done)
780 static inline void bio_integrity_trim(struct bio *bio)
785 static inline void bio_integrity_init(void)
790 static inline bool bio_integrity_flagged(struct bio *bio, enum bip_flags flag)
795 static inline void *bio_integrity_alloc(struct bio * bio, gfp_t gfp,
798 return ERR_PTR(-EINVAL);
801 static inline int bio_integrity_add_page(struct bio *bio, struct page *page,
802 unsigned int len, unsigned int offset)
807 #endif /* CONFIG_BLK_DEV_INTEGRITY */
810 * Mark a bio as polled. Note that for async polled IO, the caller must
811 * expect -EWOULDBLOCK if we cannot allocate a request (or other resources).
812 * We cannot block waiting for requests on polled IO, as those completions
813 * must be found by the caller. This is different than IRQ driven IO, where
814 * it's safe to wait for IO to complete.
816 static inline void bio_set_polled(struct bio *bio, struct kiocb *kiocb)
818 bio->bi_opf |= REQ_HIPRI;
819 if (!is_sync_kiocb(kiocb))
820 bio->bi_opf |= REQ_NOWAIT;
823 #endif /* __LINUX_BIO_H */