4 * This program is free software; you can redistribute it and/or modify
5 * it under the terms of the GNU General Public License version 2 as
6 * published by the Free Software Foundation.
8 * This program is distributed in the hope that it will be useful,
9 * but WITHOUT ANY WARRANTY; without even the implied warranty of
11 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 * GNU General Public License for more details.
14 * You should have received a copy of the GNU General Public Licens
15 * along with this program; if not, write to the Free Software
16 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-
21 #include <linux/highmem.h>
22 #include <linux/mempool.h>
23 #include <linux/ioprio.h>
26 /* struct bio, bio_vec and BIO_* flags are defined in blk_types.h */
27 #include <linux/blk_types.h>
32 #define BIO_BUG_ON BUG_ON
37 #define BIO_MAX_PAGES 256
39 #define bio_prio(bio) (bio)->bi_ioprio
40 #define bio_set_prio(bio, prio) ((bio)->bi_ioprio = prio)
42 #define bio_iter_iovec(bio, iter) \
43 bvec_iter_bvec((bio)->bi_io_vec, (iter))
45 #define bio_iter_page(bio, iter) \
46 bvec_iter_page((bio)->bi_io_vec, (iter))
47 #define bio_iter_len(bio, iter) \
48 bvec_iter_len((bio)->bi_io_vec, (iter))
49 #define bio_iter_offset(bio, iter) \
50 bvec_iter_offset((bio)->bi_io_vec, (iter))
52 #define bio_page(bio) bio_iter_page((bio), (bio)->bi_iter)
53 #define bio_offset(bio) bio_iter_offset((bio), (bio)->bi_iter)
54 #define bio_iovec(bio) bio_iter_iovec((bio), (bio)->bi_iter)
56 #define bio_multiple_segments(bio) \
57 ((bio)->bi_iter.bi_size != bio_iovec(bio).bv_len)
59 #define bvec_iter_sectors(iter) ((iter).bi_size >> 9)
60 #define bvec_iter_end_sector(iter) ((iter).bi_sector + bvec_iter_sectors((iter)))
62 #define bio_sectors(bio) bvec_iter_sectors((bio)->bi_iter)
63 #define bio_end_sector(bio) bvec_iter_end_sector((bio)->bi_iter)
66 * Return the data direction, READ or WRITE.
68 #define bio_data_dir(bio) \
69 (op_is_write(bio_op(bio)) ? WRITE : READ)
72 * Check whether this bio carries any data or not. A NULL bio is allowed.
74 static inline bool bio_has_data(struct bio *bio)
77 bio->bi_iter.bi_size &&
78 bio_op(bio) != REQ_OP_DISCARD &&
79 bio_op(bio) != REQ_OP_SECURE_ERASE &&
80 bio_op(bio) != REQ_OP_WRITE_ZEROES)
86 static inline bool bio_no_advance_iter(struct bio *bio)
88 return bio_op(bio) == REQ_OP_DISCARD ||
89 bio_op(bio) == REQ_OP_SECURE_ERASE ||
90 bio_op(bio) == REQ_OP_WRITE_SAME ||
91 bio_op(bio) == REQ_OP_WRITE_ZEROES;
94 static inline bool bio_mergeable(struct bio *bio)
96 if (bio->bi_opf & REQ_NOMERGE_FLAGS)
102 static inline unsigned int bio_cur_bytes(struct bio *bio)
104 if (bio_has_data(bio))
105 return bio_iovec(bio).bv_len;
106 else /* dataless requests such as discard */
107 return bio->bi_iter.bi_size;
110 static inline void *bio_data(struct bio *bio)
112 if (bio_has_data(bio))
113 return page_address(bio_page(bio)) + bio_offset(bio);
118 static inline bool bio_full(struct bio *bio)
120 return bio->bi_vcnt >= bio->bi_max_vecs;
123 #define mp_bvec_for_each_segment(bv, bvl, i, iter_all) \
124 for (bv = bvec_init_iter_all(&iter_all); \
125 (iter_all.done < (bvl)->bv_len) && \
126 (mp_bvec_next_segment((bvl), &iter_all), 1); \
127 iter_all.done += bv->bv_len, i += 1)
130 * drivers should _never_ use the all version - the bio may have been split
131 * before it got to the driver and the driver won't own all of it
133 #define bio_for_each_segment_all(bvl, bio, i, iter_all) \
134 for (i = 0, iter_all.idx = 0; iter_all.idx < (bio)->bi_vcnt; iter_all.idx++) \
135 mp_bvec_for_each_segment(bvl, &((bio)->bi_io_vec[iter_all.idx]), i, iter_all)
137 static inline void bio_advance_iter(struct bio *bio, struct bvec_iter *iter,
140 iter->bi_sector += bytes >> 9;
142 if (bio_no_advance_iter(bio))
143 iter->bi_size -= bytes;
145 bvec_iter_advance(bio->bi_io_vec, iter, bytes);
146 /* TODO: It is reasonable to complete bio with error here. */
149 #define __bio_for_each_segment(bvl, bio, iter, start) \
150 for (iter = (start); \
152 ((bvl = bio_iter_iovec((bio), (iter))), 1); \
153 bio_advance_iter((bio), &(iter), (bvl).bv_len))
155 #define bio_for_each_segment(bvl, bio, iter) \
156 __bio_for_each_segment(bvl, bio, iter, (bio)->bi_iter)
158 #define __bio_for_each_bvec(bvl, bio, iter, start) \
159 for (iter = (start); \
161 ((bvl = mp_bvec_iter_bvec((bio)->bi_io_vec, (iter))), 1); \
162 bio_advance_iter((bio), &(iter), (bvl).bv_len))
164 /* iterate over multi-page bvec */
165 #define bio_for_each_bvec(bvl, bio, iter) \
166 __bio_for_each_bvec(bvl, bio, iter, (bio)->bi_iter)
168 #define bio_iter_last(bvec, iter) ((iter).bi_size == (bvec).bv_len)
170 static inline unsigned bio_segments(struct bio *bio)
174 struct bvec_iter iter;
177 * We special case discard/write same/write zeroes, because they
178 * interpret bi_size differently:
181 switch (bio_op(bio)) {
183 case REQ_OP_SECURE_ERASE:
184 case REQ_OP_WRITE_ZEROES:
186 case REQ_OP_WRITE_SAME:
192 bio_for_each_segment(bv, bio, iter)
199 * get a reference to a bio, so it won't disappear. the intended use is
203 * submit_bio(rw, bio);
204 * if (bio->bi_flags ...)
208 * without the bio_get(), it could potentially complete I/O before submit_bio
209 * returns. and then bio would be freed memory when if (bio->bi_flags ...)
212 static inline void bio_get(struct bio *bio)
214 bio->bi_flags |= (1 << BIO_REFFED);
215 smp_mb__before_atomic();
216 atomic_inc(&bio->__bi_cnt);
219 static inline void bio_cnt_set(struct bio *bio, unsigned int count)
222 bio->bi_flags |= (1 << BIO_REFFED);
223 smp_mb__before_atomic();
225 atomic_set(&bio->__bi_cnt, count);
228 static inline bool bio_flagged(struct bio *bio, unsigned int bit)
230 return (bio->bi_flags & (1U << bit)) != 0;
233 static inline void bio_set_flag(struct bio *bio, unsigned int bit)
235 bio->bi_flags |= (1U << bit);
238 static inline void bio_clear_flag(struct bio *bio, unsigned int bit)
240 bio->bi_flags &= ~(1U << bit);
243 static inline void bio_get_first_bvec(struct bio *bio, struct bio_vec *bv)
245 *bv = bio_iovec(bio);
248 static inline void bio_get_last_bvec(struct bio *bio, struct bio_vec *bv)
250 struct bvec_iter iter = bio->bi_iter;
253 if (unlikely(!bio_multiple_segments(bio))) {
254 *bv = bio_iovec(bio);
258 bio_advance_iter(bio, &iter, iter.bi_size);
260 if (!iter.bi_bvec_done)
261 idx = iter.bi_idx - 1;
262 else /* in the middle of bvec */
265 *bv = bio->bi_io_vec[idx];
268 * iter.bi_bvec_done records actual length of the last bvec
269 * if this bio ends in the middle of one io vector
271 if (iter.bi_bvec_done)
272 bv->bv_len = iter.bi_bvec_done;
275 static inline struct bio_vec *bio_first_bvec_all(struct bio *bio)
277 WARN_ON_ONCE(bio_flagged(bio, BIO_CLONED));
278 return bio->bi_io_vec;
281 static inline struct page *bio_first_page_all(struct bio *bio)
283 return bio_first_bvec_all(bio)->bv_page;
286 static inline struct bio_vec *bio_last_bvec_all(struct bio *bio)
288 WARN_ON_ONCE(bio_flagged(bio, BIO_CLONED));
289 return &bio->bi_io_vec[bio->bi_vcnt - 1];
293 BIP_BLOCK_INTEGRITY = 1 << 0, /* block layer owns integrity data */
294 BIP_MAPPED_INTEGRITY = 1 << 1, /* ref tag has been remapped */
295 BIP_CTRL_NOCHECK = 1 << 2, /* disable HBA integrity checking */
296 BIP_DISK_NOCHECK = 1 << 3, /* disable disk integrity checking */
297 BIP_IP_CHECKSUM = 1 << 4, /* IP checksum */
301 * bio integrity payload
303 struct bio_integrity_payload {
304 struct bio *bip_bio; /* parent bio */
306 struct bvec_iter bip_iter;
308 unsigned short bip_slab; /* slab the bip came from */
309 unsigned short bip_vcnt; /* # of integrity bio_vecs */
310 unsigned short bip_max_vcnt; /* integrity bio_vec slots */
311 unsigned short bip_flags; /* control flags */
313 struct bvec_iter bio_iter; /* for rewinding parent bio */
315 struct work_struct bip_work; /* I/O completion */
317 struct bio_vec *bip_vec;
318 struct bio_vec bip_inline_vecs[0];/* embedded bvec array */
321 #if defined(CONFIG_BLK_DEV_INTEGRITY)
323 static inline struct bio_integrity_payload *bio_integrity(struct bio *bio)
325 if (bio->bi_opf & REQ_INTEGRITY)
326 return bio->bi_integrity;
331 static inline bool bio_integrity_flagged(struct bio *bio, enum bip_flags flag)
333 struct bio_integrity_payload *bip = bio_integrity(bio);
336 return bip->bip_flags & flag;
341 static inline sector_t bip_get_seed(struct bio_integrity_payload *bip)
343 return bip->bip_iter.bi_sector;
346 static inline void bip_set_seed(struct bio_integrity_payload *bip,
349 bip->bip_iter.bi_sector = seed;
352 #endif /* CONFIG_BLK_DEV_INTEGRITY */
354 extern void bio_trim(struct bio *bio, int offset, int size);
355 extern struct bio *bio_split(struct bio *bio, int sectors,
356 gfp_t gfp, struct bio_set *bs);
359 * bio_next_split - get next @sectors from a bio, splitting if necessary
361 * @sectors: number of sectors to split from the front of @bio
363 * @bs: bio set to allocate from
365 * Returns a bio representing the next @sectors of @bio - if the bio is smaller
366 * than @sectors, returns the original bio unchanged.
368 static inline struct bio *bio_next_split(struct bio *bio, int sectors,
369 gfp_t gfp, struct bio_set *bs)
371 if (sectors >= bio_sectors(bio))
374 return bio_split(bio, sectors, gfp, bs);
378 BIOSET_NEED_BVECS = BIT(0),
379 BIOSET_NEED_RESCUER = BIT(1),
381 extern int bioset_init(struct bio_set *, unsigned int, unsigned int, int flags);
382 extern void bioset_exit(struct bio_set *);
383 extern int biovec_init_pool(mempool_t *pool, int pool_entries);
384 extern int bioset_init_from_src(struct bio_set *bs, struct bio_set *src);
386 extern struct bio *bio_alloc_bioset(gfp_t, unsigned int, struct bio_set *);
387 extern void bio_put(struct bio *);
389 extern void __bio_clone_fast(struct bio *, struct bio *);
390 extern struct bio *bio_clone_fast(struct bio *, gfp_t, struct bio_set *);
392 extern struct bio_set fs_bio_set;
394 static inline struct bio *bio_alloc(gfp_t gfp_mask, unsigned int nr_iovecs)
396 return bio_alloc_bioset(gfp_mask, nr_iovecs, &fs_bio_set);
399 static inline struct bio *bio_kmalloc(gfp_t gfp_mask, unsigned int nr_iovecs)
401 return bio_alloc_bioset(gfp_mask, nr_iovecs, NULL);
404 extern blk_qc_t submit_bio(struct bio *);
406 extern void bio_endio(struct bio *);
408 static inline void bio_io_error(struct bio *bio)
410 bio->bi_status = BLK_STS_IOERR;
414 static inline void bio_wouldblock_error(struct bio *bio)
416 bio->bi_status = BLK_STS_AGAIN;
420 struct request_queue;
421 extern int bio_phys_segments(struct request_queue *, struct bio *);
423 extern int submit_bio_wait(struct bio *bio);
424 extern void bio_advance(struct bio *, unsigned);
426 extern void bio_init(struct bio *bio, struct bio_vec *table,
427 unsigned short max_vecs);
428 extern void bio_uninit(struct bio *);
429 extern void bio_reset(struct bio *);
430 void bio_chain(struct bio *, struct bio *);
432 extern int bio_add_page(struct bio *, struct page *, unsigned int,unsigned int);
433 extern int bio_add_pc_page(struct request_queue *, struct bio *, struct page *,
434 unsigned int, unsigned int);
435 bool __bio_try_merge_page(struct bio *bio, struct page *page,
436 unsigned int len, unsigned int off, bool same_page);
437 void __bio_add_page(struct bio *bio, struct page *page,
438 unsigned int len, unsigned int off);
439 int bio_iov_iter_get_pages(struct bio *bio, struct iov_iter *iter);
441 extern struct bio *bio_map_user_iov(struct request_queue *,
442 struct iov_iter *, gfp_t);
443 extern void bio_unmap_user(struct bio *);
444 extern struct bio *bio_map_kern(struct request_queue *, void *, unsigned int,
446 extern struct bio *bio_copy_kern(struct request_queue *, void *, unsigned int,
448 extern void bio_set_pages_dirty(struct bio *bio);
449 extern void bio_check_pages_dirty(struct bio *bio);
451 void generic_start_io_acct(struct request_queue *q, int op,
452 unsigned long sectors, struct hd_struct *part);
453 void generic_end_io_acct(struct request_queue *q, int op,
454 struct hd_struct *part,
455 unsigned long start_time);
457 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
458 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
460 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
461 extern void bio_flush_dcache_pages(struct bio *bi);
463 static inline void bio_flush_dcache_pages(struct bio *bi)
468 extern void bio_copy_data_iter(struct bio *dst, struct bvec_iter *dst_iter,
469 struct bio *src, struct bvec_iter *src_iter);
470 extern void bio_copy_data(struct bio *dst, struct bio *src);
471 extern void bio_list_copy_data(struct bio *dst, struct bio *src);
472 extern void bio_free_pages(struct bio *bio);
474 extern struct bio *bio_copy_user_iov(struct request_queue *,
475 struct rq_map_data *,
478 extern int bio_uncopy_user(struct bio *);
479 void zero_fill_bio_iter(struct bio *bio, struct bvec_iter iter);
481 static inline void zero_fill_bio(struct bio *bio)
483 zero_fill_bio_iter(bio, bio->bi_iter);
486 extern struct bio_vec *bvec_alloc(gfp_t, int, unsigned long *, mempool_t *);
487 extern void bvec_free(mempool_t *, struct bio_vec *, unsigned int);
488 extern unsigned int bvec_nr_vecs(unsigned short idx);
489 extern const char *bio_devname(struct bio *bio, char *buffer);
491 #define bio_set_dev(bio, bdev) \
493 if ((bio)->bi_disk != (bdev)->bd_disk) \
494 bio_clear_flag(bio, BIO_THROTTLED);\
495 (bio)->bi_disk = (bdev)->bd_disk; \
496 (bio)->bi_partno = (bdev)->bd_partno; \
497 bio_associate_blkg(bio); \
500 #define bio_copy_dev(dst, src) \
502 (dst)->bi_disk = (src)->bi_disk; \
503 (dst)->bi_partno = (src)->bi_partno; \
504 bio_clone_blkg_association(dst, src); \
507 #define bio_dev(bio) \
508 disk_devt((bio)->bi_disk)
510 #if defined(CONFIG_MEMCG) && defined(CONFIG_BLK_CGROUP)
511 void bio_associate_blkg_from_page(struct bio *bio, struct page *page);
513 static inline void bio_associate_blkg_from_page(struct bio *bio,
514 struct page *page) { }
517 #ifdef CONFIG_BLK_CGROUP
518 void bio_disassociate_blkg(struct bio *bio);
519 void bio_associate_blkg(struct bio *bio);
520 void bio_associate_blkg_from_css(struct bio *bio,
521 struct cgroup_subsys_state *css);
522 void bio_clone_blkg_association(struct bio *dst, struct bio *src);
523 #else /* CONFIG_BLK_CGROUP */
524 static inline void bio_disassociate_blkg(struct bio *bio) { }
525 static inline void bio_associate_blkg(struct bio *bio) { }
526 static inline void bio_associate_blkg_from_css(struct bio *bio,
527 struct cgroup_subsys_state *css)
529 static inline void bio_clone_blkg_association(struct bio *dst,
531 #endif /* CONFIG_BLK_CGROUP */
533 #ifdef CONFIG_HIGHMEM
535 * remember never ever reenable interrupts between a bvec_kmap_irq and
538 static inline char *bvec_kmap_irq(struct bio_vec *bvec, unsigned long *flags)
543 * might not be a highmem page, but the preempt/irq count
544 * balancing is a lot nicer this way
546 local_irq_save(*flags);
547 addr = (unsigned long) kmap_atomic(bvec->bv_page);
549 BUG_ON(addr & ~PAGE_MASK);
551 return (char *) addr + bvec->bv_offset;
554 static inline void bvec_kunmap_irq(char *buffer, unsigned long *flags)
556 unsigned long ptr = (unsigned long) buffer & PAGE_MASK;
558 kunmap_atomic((void *) ptr);
559 local_irq_restore(*flags);
563 static inline char *bvec_kmap_irq(struct bio_vec *bvec, unsigned long *flags)
565 return page_address(bvec->bv_page) + bvec->bv_offset;
568 static inline void bvec_kunmap_irq(char *buffer, unsigned long *flags)
575 * BIO list management for use by remapping drivers (e.g. DM or MD) and loop.
577 * A bio_list anchors a singly-linked list of bios chained through the bi_next
578 * member of the bio. The bio_list also caches the last list member to allow
579 * fast access to the tail.
586 static inline int bio_list_empty(const struct bio_list *bl)
588 return bl->head == NULL;
591 static inline void bio_list_init(struct bio_list *bl)
593 bl->head = bl->tail = NULL;
596 #define BIO_EMPTY_LIST { NULL, NULL }
598 #define bio_list_for_each(bio, bl) \
599 for (bio = (bl)->head; bio; bio = bio->bi_next)
601 static inline unsigned bio_list_size(const struct bio_list *bl)
606 bio_list_for_each(bio, bl)
612 static inline void bio_list_add(struct bio_list *bl, struct bio *bio)
617 bl->tail->bi_next = bio;
624 static inline void bio_list_add_head(struct bio_list *bl, struct bio *bio)
626 bio->bi_next = bl->head;
634 static inline void bio_list_merge(struct bio_list *bl, struct bio_list *bl2)
640 bl->tail->bi_next = bl2->head;
642 bl->head = bl2->head;
644 bl->tail = bl2->tail;
647 static inline void bio_list_merge_head(struct bio_list *bl,
648 struct bio_list *bl2)
654 bl2->tail->bi_next = bl->head;
656 bl->tail = bl2->tail;
658 bl->head = bl2->head;
661 static inline struct bio *bio_list_peek(struct bio_list *bl)
666 static inline struct bio *bio_list_pop(struct bio_list *bl)
668 struct bio *bio = bl->head;
671 bl->head = bl->head->bi_next;
681 static inline struct bio *bio_list_get(struct bio_list *bl)
683 struct bio *bio = bl->head;
685 bl->head = bl->tail = NULL;
691 * Increment chain count for the bio. Make sure the CHAIN flag update
692 * is visible before the raised count.
694 static inline void bio_inc_remaining(struct bio *bio)
696 bio_set_flag(bio, BIO_CHAIN);
697 smp_mb__before_atomic();
698 atomic_inc(&bio->__bi_remaining);
702 * bio_set is used to allow other portions of the IO system to
703 * allocate their own private memory pools for bio and iovec structures.
704 * These memory pools in turn all allocate from the bio_slab
705 * and the bvec_slabs[].
707 #define BIO_POOL_SIZE 2
710 struct kmem_cache *bio_slab;
711 unsigned int front_pad;
715 #if defined(CONFIG_BLK_DEV_INTEGRITY)
716 mempool_t bio_integrity_pool;
717 mempool_t bvec_integrity_pool;
721 * Deadlock avoidance for stacking block drivers: see comments in
722 * bio_alloc_bioset() for details
724 spinlock_t rescue_lock;
725 struct bio_list rescue_list;
726 struct work_struct rescue_work;
727 struct workqueue_struct *rescue_workqueue;
733 struct kmem_cache *slab;
736 static inline bool bioset_initialized(struct bio_set *bs)
738 return bs->bio_slab != NULL;
742 * a small number of entries is fine, not going to be performance critical.
743 * basically we just need to survive
745 #define BIO_SPLIT_ENTRIES 2
747 #if defined(CONFIG_BLK_DEV_INTEGRITY)
749 #define bip_for_each_vec(bvl, bip, iter) \
750 for_each_bvec(bvl, (bip)->bip_vec, iter, (bip)->bip_iter)
752 #define bio_for_each_integrity_vec(_bvl, _bio, _iter) \
754 bip_for_each_vec(_bvl, _bio->bi_integrity, _iter)
756 extern struct bio_integrity_payload *bio_integrity_alloc(struct bio *, gfp_t, unsigned int);
757 extern int bio_integrity_add_page(struct bio *, struct page *, unsigned int, unsigned int);
758 extern bool bio_integrity_prep(struct bio *);
759 extern void bio_integrity_advance(struct bio *, unsigned int);
760 extern void bio_integrity_trim(struct bio *);
761 extern int bio_integrity_clone(struct bio *, struct bio *, gfp_t);
762 extern int bioset_integrity_create(struct bio_set *, int);
763 extern void bioset_integrity_free(struct bio_set *);
764 extern void bio_integrity_init(void);
766 #else /* CONFIG_BLK_DEV_INTEGRITY */
768 static inline void *bio_integrity(struct bio *bio)
773 static inline int bioset_integrity_create(struct bio_set *bs, int pool_size)
778 static inline void bioset_integrity_free (struct bio_set *bs)
783 static inline bool bio_integrity_prep(struct bio *bio)
788 static inline int bio_integrity_clone(struct bio *bio, struct bio *bio_src,
794 static inline void bio_integrity_advance(struct bio *bio,
795 unsigned int bytes_done)
800 static inline void bio_integrity_trim(struct bio *bio)
805 static inline void bio_integrity_init(void)
810 static inline bool bio_integrity_flagged(struct bio *bio, enum bip_flags flag)
815 static inline void *bio_integrity_alloc(struct bio * bio, gfp_t gfp,
818 return ERR_PTR(-EINVAL);
821 static inline int bio_integrity_add_page(struct bio *bio, struct page *page,
822 unsigned int len, unsigned int offset)
827 #endif /* CONFIG_BLK_DEV_INTEGRITY */
830 * Mark a bio as polled. Note that for async polled IO, the caller must
831 * expect -EWOULDBLOCK if we cannot allocate a request (or other resources).
832 * We cannot block waiting for requests on polled IO, as those completions
833 * must be found by the caller. This is different than IRQ driven IO, where
834 * it's safe to wait for IO to complete.
836 static inline void bio_set_polled(struct bio *bio, struct kiocb *kiocb)
838 bio->bi_opf |= REQ_HIPRI;
839 if (!is_sync_kiocb(kiocb))
840 bio->bi_opf |= REQ_NOWAIT;
843 #endif /* CONFIG_BLOCK */
844 #endif /* __LINUX_BIO_H */