1 /* SPDX-License-Identifier: GPL-2.0 */
8 #include <linux/highmem.h>
9 #include <linux/mempool.h>
10 #include <linux/ioprio.h>
13 /* struct bio, bio_vec and BIO_* flags are defined in blk_types.h */
14 #include <linux/blk_types.h>
19 #define BIO_BUG_ON BUG_ON
24 #define BIO_MAX_PAGES 256
26 #define bio_prio(bio) (bio)->bi_ioprio
27 #define bio_set_prio(bio, prio) ((bio)->bi_ioprio = prio)
29 #define bio_iter_iovec(bio, iter) \
30 bvec_iter_bvec((bio)->bi_io_vec, (iter))
32 #define bio_iter_page(bio, iter) \
33 bvec_iter_page((bio)->bi_io_vec, (iter))
34 #define bio_iter_len(bio, iter) \
35 bvec_iter_len((bio)->bi_io_vec, (iter))
36 #define bio_iter_offset(bio, iter) \
37 bvec_iter_offset((bio)->bi_io_vec, (iter))
39 #define bio_page(bio) bio_iter_page((bio), (bio)->bi_iter)
40 #define bio_offset(bio) bio_iter_offset((bio), (bio)->bi_iter)
41 #define bio_iovec(bio) bio_iter_iovec((bio), (bio)->bi_iter)
43 #define bio_multiple_segments(bio) \
44 ((bio)->bi_iter.bi_size != bio_iovec(bio).bv_len)
46 #define bvec_iter_sectors(iter) ((iter).bi_size >> 9)
47 #define bvec_iter_end_sector(iter) ((iter).bi_sector + bvec_iter_sectors((iter)))
49 #define bio_sectors(bio) bvec_iter_sectors((bio)->bi_iter)
50 #define bio_end_sector(bio) bvec_iter_end_sector((bio)->bi_iter)
53 * Return the data direction, READ or WRITE.
55 #define bio_data_dir(bio) \
56 (op_is_write(bio_op(bio)) ? WRITE : READ)
59 * Check whether this bio carries any data or not. A NULL bio is allowed.
61 static inline bool bio_has_data(struct bio *bio)
64 bio->bi_iter.bi_size &&
65 bio_op(bio) != REQ_OP_DISCARD &&
66 bio_op(bio) != REQ_OP_SECURE_ERASE &&
67 bio_op(bio) != REQ_OP_WRITE_ZEROES)
73 static inline bool bio_no_advance_iter(struct bio *bio)
75 return bio_op(bio) == REQ_OP_DISCARD ||
76 bio_op(bio) == REQ_OP_SECURE_ERASE ||
77 bio_op(bio) == REQ_OP_WRITE_SAME ||
78 bio_op(bio) == REQ_OP_WRITE_ZEROES;
81 static inline bool bio_mergeable(struct bio *bio)
83 if (bio->bi_opf & REQ_NOMERGE_FLAGS)
89 static inline unsigned int bio_cur_bytes(struct bio *bio)
91 if (bio_has_data(bio))
92 return bio_iovec(bio).bv_len;
93 else /* dataless requests such as discard */
94 return bio->bi_iter.bi_size;
97 static inline void *bio_data(struct bio *bio)
99 if (bio_has_data(bio))
100 return page_address(bio_page(bio)) + bio_offset(bio);
105 static inline bool bio_full(struct bio *bio)
107 return bio->bi_vcnt >= bio->bi_max_vecs;
110 static inline bool bio_next_segment(const struct bio *bio,
111 struct bvec_iter_all *iter)
113 if (iter->idx >= bio->bi_vcnt)
116 bvec_advance(&bio->bi_io_vec[iter->idx], iter);
121 * drivers should _never_ use the all version - the bio may have been split
122 * before it got to the driver and the driver won't own all of it
124 #define bio_for_each_segment_all(bvl, bio, iter) \
125 for (bvl = bvec_init_iter_all(&iter); bio_next_segment((bio), &iter); )
127 static inline void bio_advance_iter(struct bio *bio, struct bvec_iter *iter,
130 iter->bi_sector += bytes >> 9;
132 if (bio_no_advance_iter(bio))
133 iter->bi_size -= bytes;
135 bvec_iter_advance(bio->bi_io_vec, iter, bytes);
136 /* TODO: It is reasonable to complete bio with error here. */
139 #define __bio_for_each_segment(bvl, bio, iter, start) \
140 for (iter = (start); \
142 ((bvl = bio_iter_iovec((bio), (iter))), 1); \
143 bio_advance_iter((bio), &(iter), (bvl).bv_len))
145 #define bio_for_each_segment(bvl, bio, iter) \
146 __bio_for_each_segment(bvl, bio, iter, (bio)->bi_iter)
148 #define __bio_for_each_bvec(bvl, bio, iter, start) \
149 for (iter = (start); \
151 ((bvl = mp_bvec_iter_bvec((bio)->bi_io_vec, (iter))), 1); \
152 bio_advance_iter((bio), &(iter), (bvl).bv_len))
154 /* iterate over multi-page bvec */
155 #define bio_for_each_bvec(bvl, bio, iter) \
156 __bio_for_each_bvec(bvl, bio, iter, (bio)->bi_iter)
158 #define bio_iter_last(bvec, iter) ((iter).bi_size == (bvec).bv_len)
160 static inline unsigned bio_segments(struct bio *bio)
164 struct bvec_iter iter;
167 * We special case discard/write same/write zeroes, because they
168 * interpret bi_size differently:
171 switch (bio_op(bio)) {
173 case REQ_OP_SECURE_ERASE:
174 case REQ_OP_WRITE_ZEROES:
176 case REQ_OP_WRITE_SAME:
182 bio_for_each_segment(bv, bio, iter)
189 * get a reference to a bio, so it won't disappear. the intended use is
193 * submit_bio(rw, bio);
194 * if (bio->bi_flags ...)
198 * without the bio_get(), it could potentially complete I/O before submit_bio
199 * returns. and then bio would be freed memory when if (bio->bi_flags ...)
202 static inline void bio_get(struct bio *bio)
204 bio->bi_flags |= (1 << BIO_REFFED);
205 smp_mb__before_atomic();
206 atomic_inc(&bio->__bi_cnt);
209 static inline void bio_cnt_set(struct bio *bio, unsigned int count)
212 bio->bi_flags |= (1 << BIO_REFFED);
215 atomic_set(&bio->__bi_cnt, count);
218 static inline bool bio_flagged(struct bio *bio, unsigned int bit)
220 return (bio->bi_flags & (1U << bit)) != 0;
223 static inline void bio_set_flag(struct bio *bio, unsigned int bit)
225 bio->bi_flags |= (1U << bit);
228 static inline void bio_clear_flag(struct bio *bio, unsigned int bit)
230 bio->bi_flags &= ~(1U << bit);
233 static inline void bio_get_first_bvec(struct bio *bio, struct bio_vec *bv)
235 *bv = bio_iovec(bio);
238 static inline void bio_get_last_bvec(struct bio *bio, struct bio_vec *bv)
240 struct bvec_iter iter = bio->bi_iter;
243 if (unlikely(!bio_multiple_segments(bio))) {
244 *bv = bio_iovec(bio);
248 bio_advance_iter(bio, &iter, iter.bi_size);
250 if (!iter.bi_bvec_done)
251 idx = iter.bi_idx - 1;
252 else /* in the middle of bvec */
255 *bv = bio->bi_io_vec[idx];
258 * iter.bi_bvec_done records actual length of the last bvec
259 * if this bio ends in the middle of one io vector
261 if (iter.bi_bvec_done)
262 bv->bv_len = iter.bi_bvec_done;
265 static inline struct bio_vec *bio_first_bvec_all(struct bio *bio)
267 WARN_ON_ONCE(bio_flagged(bio, BIO_CLONED));
268 return bio->bi_io_vec;
271 static inline struct page *bio_first_page_all(struct bio *bio)
273 return bio_first_bvec_all(bio)->bv_page;
276 static inline struct bio_vec *bio_last_bvec_all(struct bio *bio)
278 WARN_ON_ONCE(bio_flagged(bio, BIO_CLONED));
279 return &bio->bi_io_vec[bio->bi_vcnt - 1];
283 BIP_BLOCK_INTEGRITY = 1 << 0, /* block layer owns integrity data */
284 BIP_MAPPED_INTEGRITY = 1 << 1, /* ref tag has been remapped */
285 BIP_CTRL_NOCHECK = 1 << 2, /* disable HBA integrity checking */
286 BIP_DISK_NOCHECK = 1 << 3, /* disable disk integrity checking */
287 BIP_IP_CHECKSUM = 1 << 4, /* IP checksum */
291 * bio integrity payload
293 struct bio_integrity_payload {
294 struct bio *bip_bio; /* parent bio */
296 struct bvec_iter bip_iter;
298 unsigned short bip_slab; /* slab the bip came from */
299 unsigned short bip_vcnt; /* # of integrity bio_vecs */
300 unsigned short bip_max_vcnt; /* integrity bio_vec slots */
301 unsigned short bip_flags; /* control flags */
303 struct bvec_iter bio_iter; /* for rewinding parent bio */
305 struct work_struct bip_work; /* I/O completion */
307 struct bio_vec *bip_vec;
308 struct bio_vec bip_inline_vecs[0];/* embedded bvec array */
311 #if defined(CONFIG_BLK_DEV_INTEGRITY)
313 static inline struct bio_integrity_payload *bio_integrity(struct bio *bio)
315 if (bio->bi_opf & REQ_INTEGRITY)
316 return bio->bi_integrity;
321 static inline bool bio_integrity_flagged(struct bio *bio, enum bip_flags flag)
323 struct bio_integrity_payload *bip = bio_integrity(bio);
326 return bip->bip_flags & flag;
331 static inline sector_t bip_get_seed(struct bio_integrity_payload *bip)
333 return bip->bip_iter.bi_sector;
336 static inline void bip_set_seed(struct bio_integrity_payload *bip,
339 bip->bip_iter.bi_sector = seed;
342 #endif /* CONFIG_BLK_DEV_INTEGRITY */
344 extern void bio_trim(struct bio *bio, int offset, int size);
345 extern struct bio *bio_split(struct bio *bio, int sectors,
346 gfp_t gfp, struct bio_set *bs);
349 * bio_next_split - get next @sectors from a bio, splitting if necessary
351 * @sectors: number of sectors to split from the front of @bio
353 * @bs: bio set to allocate from
355 * Returns a bio representing the next @sectors of @bio - if the bio is smaller
356 * than @sectors, returns the original bio unchanged.
358 static inline struct bio *bio_next_split(struct bio *bio, int sectors,
359 gfp_t gfp, struct bio_set *bs)
361 if (sectors >= bio_sectors(bio))
364 return bio_split(bio, sectors, gfp, bs);
368 BIOSET_NEED_BVECS = BIT(0),
369 BIOSET_NEED_RESCUER = BIT(1),
371 extern int bioset_init(struct bio_set *, unsigned int, unsigned int, int flags);
372 extern void bioset_exit(struct bio_set *);
373 extern int biovec_init_pool(mempool_t *pool, int pool_entries);
374 extern int bioset_init_from_src(struct bio_set *bs, struct bio_set *src);
376 extern struct bio *bio_alloc_bioset(gfp_t, unsigned int, struct bio_set *);
377 extern void bio_put(struct bio *);
379 extern void __bio_clone_fast(struct bio *, struct bio *);
380 extern struct bio *bio_clone_fast(struct bio *, gfp_t, struct bio_set *);
382 extern struct bio_set fs_bio_set;
384 static inline struct bio *bio_alloc(gfp_t gfp_mask, unsigned int nr_iovecs)
386 return bio_alloc_bioset(gfp_mask, nr_iovecs, &fs_bio_set);
389 static inline struct bio *bio_kmalloc(gfp_t gfp_mask, unsigned int nr_iovecs)
391 return bio_alloc_bioset(gfp_mask, nr_iovecs, NULL);
394 extern blk_qc_t submit_bio(struct bio *);
396 extern void bio_endio(struct bio *);
398 static inline void bio_io_error(struct bio *bio)
400 bio->bi_status = BLK_STS_IOERR;
404 static inline void bio_wouldblock_error(struct bio *bio)
406 bio->bi_status = BLK_STS_AGAIN;
410 struct request_queue;
411 extern int bio_phys_segments(struct request_queue *, struct bio *);
413 extern int submit_bio_wait(struct bio *bio);
414 extern void bio_advance(struct bio *, unsigned);
416 extern void bio_init(struct bio *bio, struct bio_vec *table,
417 unsigned short max_vecs);
418 extern void bio_uninit(struct bio *);
419 extern void bio_reset(struct bio *);
420 void bio_chain(struct bio *, struct bio *);
422 extern int bio_add_page(struct bio *, struct page *, unsigned int,unsigned int);
423 extern int bio_add_pc_page(struct request_queue *, struct bio *, struct page *,
424 unsigned int, unsigned int);
425 bool __bio_try_merge_page(struct bio *bio, struct page *page,
426 unsigned int len, unsigned int off, bool *same_page);
427 void __bio_add_page(struct bio *bio, struct page *page,
428 unsigned int len, unsigned int off);
429 int bio_iov_iter_get_pages(struct bio *bio, struct iov_iter *iter);
431 extern struct bio *bio_map_user_iov(struct request_queue *,
432 struct iov_iter *, gfp_t);
433 extern void bio_unmap_user(struct bio *);
434 extern struct bio *bio_map_kern(struct request_queue *, void *, unsigned int,
436 extern struct bio *bio_copy_kern(struct request_queue *, void *, unsigned int,
438 extern void bio_set_pages_dirty(struct bio *bio);
439 extern void bio_check_pages_dirty(struct bio *bio);
441 void generic_start_io_acct(struct request_queue *q, int op,
442 unsigned long sectors, struct hd_struct *part);
443 void generic_end_io_acct(struct request_queue *q, int op,
444 struct hd_struct *part,
445 unsigned long start_time);
447 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
448 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
450 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
451 extern void bio_flush_dcache_pages(struct bio *bi);
453 static inline void bio_flush_dcache_pages(struct bio *bi)
458 extern void bio_copy_data_iter(struct bio *dst, struct bvec_iter *dst_iter,
459 struct bio *src, struct bvec_iter *src_iter);
460 extern void bio_copy_data(struct bio *dst, struct bio *src);
461 extern void bio_list_copy_data(struct bio *dst, struct bio *src);
462 extern void bio_free_pages(struct bio *bio);
464 extern struct bio *bio_copy_user_iov(struct request_queue *,
465 struct rq_map_data *,
468 extern int bio_uncopy_user(struct bio *);
469 void zero_fill_bio_iter(struct bio *bio, struct bvec_iter iter);
471 static inline void zero_fill_bio(struct bio *bio)
473 zero_fill_bio_iter(bio, bio->bi_iter);
476 extern struct bio_vec *bvec_alloc(gfp_t, int, unsigned long *, mempool_t *);
477 extern void bvec_free(mempool_t *, struct bio_vec *, unsigned int);
478 extern unsigned int bvec_nr_vecs(unsigned short idx);
479 extern const char *bio_devname(struct bio *bio, char *buffer);
481 #define bio_set_dev(bio, bdev) \
483 if ((bio)->bi_disk != (bdev)->bd_disk) \
484 bio_clear_flag(bio, BIO_THROTTLED);\
485 (bio)->bi_disk = (bdev)->bd_disk; \
486 (bio)->bi_partno = (bdev)->bd_partno; \
487 bio_associate_blkg(bio); \
490 #define bio_copy_dev(dst, src) \
492 (dst)->bi_disk = (src)->bi_disk; \
493 (dst)->bi_partno = (src)->bi_partno; \
494 bio_clone_blkg_association(dst, src); \
497 #define bio_dev(bio) \
498 disk_devt((bio)->bi_disk)
500 #if defined(CONFIG_MEMCG) && defined(CONFIG_BLK_CGROUP)
501 void bio_associate_blkg_from_page(struct bio *bio, struct page *page);
503 static inline void bio_associate_blkg_from_page(struct bio *bio,
504 struct page *page) { }
507 #ifdef CONFIG_BLK_CGROUP
508 void bio_disassociate_blkg(struct bio *bio);
509 void bio_associate_blkg(struct bio *bio);
510 void bio_associate_blkg_from_css(struct bio *bio,
511 struct cgroup_subsys_state *css);
512 void bio_clone_blkg_association(struct bio *dst, struct bio *src);
513 #else /* CONFIG_BLK_CGROUP */
514 static inline void bio_disassociate_blkg(struct bio *bio) { }
515 static inline void bio_associate_blkg(struct bio *bio) { }
516 static inline void bio_associate_blkg_from_css(struct bio *bio,
517 struct cgroup_subsys_state *css)
519 static inline void bio_clone_blkg_association(struct bio *dst,
521 #endif /* CONFIG_BLK_CGROUP */
523 #ifdef CONFIG_HIGHMEM
525 * remember never ever reenable interrupts between a bvec_kmap_irq and
528 static inline char *bvec_kmap_irq(struct bio_vec *bvec, unsigned long *flags)
533 * might not be a highmem page, but the preempt/irq count
534 * balancing is a lot nicer this way
536 local_irq_save(*flags);
537 addr = (unsigned long) kmap_atomic(bvec->bv_page);
539 BUG_ON(addr & ~PAGE_MASK);
541 return (char *) addr + bvec->bv_offset;
544 static inline void bvec_kunmap_irq(char *buffer, unsigned long *flags)
546 unsigned long ptr = (unsigned long) buffer & PAGE_MASK;
548 kunmap_atomic((void *) ptr);
549 local_irq_restore(*flags);
553 static inline char *bvec_kmap_irq(struct bio_vec *bvec, unsigned long *flags)
555 return page_address(bvec->bv_page) + bvec->bv_offset;
558 static inline void bvec_kunmap_irq(char *buffer, unsigned long *flags)
565 * BIO list management for use by remapping drivers (e.g. DM or MD) and loop.
567 * A bio_list anchors a singly-linked list of bios chained through the bi_next
568 * member of the bio. The bio_list also caches the last list member to allow
569 * fast access to the tail.
576 static inline int bio_list_empty(const struct bio_list *bl)
578 return bl->head == NULL;
581 static inline void bio_list_init(struct bio_list *bl)
583 bl->head = bl->tail = NULL;
586 #define BIO_EMPTY_LIST { NULL, NULL }
588 #define bio_list_for_each(bio, bl) \
589 for (bio = (bl)->head; bio; bio = bio->bi_next)
591 static inline unsigned bio_list_size(const struct bio_list *bl)
596 bio_list_for_each(bio, bl)
602 static inline void bio_list_add(struct bio_list *bl, struct bio *bio)
607 bl->tail->bi_next = bio;
614 static inline void bio_list_add_head(struct bio_list *bl, struct bio *bio)
616 bio->bi_next = bl->head;
624 static inline void bio_list_merge(struct bio_list *bl, struct bio_list *bl2)
630 bl->tail->bi_next = bl2->head;
632 bl->head = bl2->head;
634 bl->tail = bl2->tail;
637 static inline void bio_list_merge_head(struct bio_list *bl,
638 struct bio_list *bl2)
644 bl2->tail->bi_next = bl->head;
646 bl->tail = bl2->tail;
648 bl->head = bl2->head;
651 static inline struct bio *bio_list_peek(struct bio_list *bl)
656 static inline struct bio *bio_list_pop(struct bio_list *bl)
658 struct bio *bio = bl->head;
661 bl->head = bl->head->bi_next;
671 static inline struct bio *bio_list_get(struct bio_list *bl)
673 struct bio *bio = bl->head;
675 bl->head = bl->tail = NULL;
681 * Increment chain count for the bio. Make sure the CHAIN flag update
682 * is visible before the raised count.
684 static inline void bio_inc_remaining(struct bio *bio)
686 bio_set_flag(bio, BIO_CHAIN);
687 smp_mb__before_atomic();
688 atomic_inc(&bio->__bi_remaining);
692 * bio_set is used to allow other portions of the IO system to
693 * allocate their own private memory pools for bio and iovec structures.
694 * These memory pools in turn all allocate from the bio_slab
695 * and the bvec_slabs[].
697 #define BIO_POOL_SIZE 2
700 struct kmem_cache *bio_slab;
701 unsigned int front_pad;
705 #if defined(CONFIG_BLK_DEV_INTEGRITY)
706 mempool_t bio_integrity_pool;
707 mempool_t bvec_integrity_pool;
711 * Deadlock avoidance for stacking block drivers: see comments in
712 * bio_alloc_bioset() for details
714 spinlock_t rescue_lock;
715 struct bio_list rescue_list;
716 struct work_struct rescue_work;
717 struct workqueue_struct *rescue_workqueue;
723 struct kmem_cache *slab;
726 static inline bool bioset_initialized(struct bio_set *bs)
728 return bs->bio_slab != NULL;
732 * a small number of entries is fine, not going to be performance critical.
733 * basically we just need to survive
735 #define BIO_SPLIT_ENTRIES 2
737 #if defined(CONFIG_BLK_DEV_INTEGRITY)
739 #define bip_for_each_vec(bvl, bip, iter) \
740 for_each_bvec(bvl, (bip)->bip_vec, iter, (bip)->bip_iter)
742 #define bio_for_each_integrity_vec(_bvl, _bio, _iter) \
744 bip_for_each_vec(_bvl, _bio->bi_integrity, _iter)
746 extern struct bio_integrity_payload *bio_integrity_alloc(struct bio *, gfp_t, unsigned int);
747 extern int bio_integrity_add_page(struct bio *, struct page *, unsigned int, unsigned int);
748 extern bool bio_integrity_prep(struct bio *);
749 extern void bio_integrity_advance(struct bio *, unsigned int);
750 extern void bio_integrity_trim(struct bio *);
751 extern int bio_integrity_clone(struct bio *, struct bio *, gfp_t);
752 extern int bioset_integrity_create(struct bio_set *, int);
753 extern void bioset_integrity_free(struct bio_set *);
754 extern void bio_integrity_init(void);
756 #else /* CONFIG_BLK_DEV_INTEGRITY */
758 static inline void *bio_integrity(struct bio *bio)
763 static inline int bioset_integrity_create(struct bio_set *bs, int pool_size)
768 static inline void bioset_integrity_free (struct bio_set *bs)
773 static inline bool bio_integrity_prep(struct bio *bio)
778 static inline int bio_integrity_clone(struct bio *bio, struct bio *bio_src,
784 static inline void bio_integrity_advance(struct bio *bio,
785 unsigned int bytes_done)
790 static inline void bio_integrity_trim(struct bio *bio)
795 static inline void bio_integrity_init(void)
800 static inline bool bio_integrity_flagged(struct bio *bio, enum bip_flags flag)
805 static inline void *bio_integrity_alloc(struct bio * bio, gfp_t gfp,
808 return ERR_PTR(-EINVAL);
811 static inline int bio_integrity_add_page(struct bio *bio, struct page *page,
812 unsigned int len, unsigned int offset)
817 #endif /* CONFIG_BLK_DEV_INTEGRITY */
820 * Mark a bio as polled. Note that for async polled IO, the caller must
821 * expect -EWOULDBLOCK if we cannot allocate a request (or other resources).
822 * We cannot block waiting for requests on polled IO, as those completions
823 * must be found by the caller. This is different than IRQ driven IO, where
824 * it's safe to wait for IO to complete.
826 static inline void bio_set_polled(struct bio *bio, struct kiocb *kiocb)
828 bio->bi_opf |= REQ_HIPRI;
829 if (!is_sync_kiocb(kiocb))
830 bio->bi_opf |= REQ_NOWAIT;
833 #endif /* CONFIG_BLOCK */
834 #endif /* __LINUX_BIO_H */