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 #ifdef CONFIG_THP_SWAP
38 #if HPAGE_PMD_NR > 256
39 #define BIO_MAX_PAGES HPAGE_PMD_NR
41 #define BIO_MAX_PAGES 256
44 #define BIO_MAX_PAGES 256
47 #define bio_prio(bio) (bio)->bi_ioprio
48 #define bio_set_prio(bio, prio) ((bio)->bi_ioprio = prio)
50 #define bio_iter_iovec(bio, iter) \
51 bvec_iter_bvec((bio)->bi_io_vec, (iter))
53 #define bio_iter_page(bio, iter) \
54 bvec_iter_page((bio)->bi_io_vec, (iter))
55 #define bio_iter_len(bio, iter) \
56 bvec_iter_len((bio)->bi_io_vec, (iter))
57 #define bio_iter_offset(bio, iter) \
58 bvec_iter_offset((bio)->bi_io_vec, (iter))
60 #define bio_page(bio) bio_iter_page((bio), (bio)->bi_iter)
61 #define bio_offset(bio) bio_iter_offset((bio), (bio)->bi_iter)
62 #define bio_iovec(bio) bio_iter_iovec((bio), (bio)->bi_iter)
64 #define bio_multiple_segments(bio) \
65 ((bio)->bi_iter.bi_size != bio_iovec(bio).bv_len)
67 #define bvec_iter_sectors(iter) ((iter).bi_size >> 9)
68 #define bvec_iter_end_sector(iter) ((iter).bi_sector + bvec_iter_sectors((iter)))
70 #define bio_sectors(bio) bvec_iter_sectors((bio)->bi_iter)
71 #define bio_end_sector(bio) bvec_iter_end_sector((bio)->bi_iter)
74 * Return the data direction, READ or WRITE.
76 #define bio_data_dir(bio) \
77 (op_is_write(bio_op(bio)) ? WRITE : READ)
80 * Check whether this bio carries any data or not. A NULL bio is allowed.
82 static inline bool bio_has_data(struct bio *bio)
85 bio->bi_iter.bi_size &&
86 bio_op(bio) != REQ_OP_DISCARD &&
87 bio_op(bio) != REQ_OP_SECURE_ERASE &&
88 bio_op(bio) != REQ_OP_WRITE_ZEROES)
94 static inline bool bio_no_advance_iter(struct bio *bio)
96 return bio_op(bio) == REQ_OP_DISCARD ||
97 bio_op(bio) == REQ_OP_SECURE_ERASE ||
98 bio_op(bio) == REQ_OP_WRITE_SAME ||
99 bio_op(bio) == REQ_OP_WRITE_ZEROES;
102 static inline bool bio_mergeable(struct bio *bio)
104 if (bio->bi_opf & REQ_NOMERGE_FLAGS)
110 static inline unsigned int bio_cur_bytes(struct bio *bio)
112 if (bio_has_data(bio))
113 return bio_iovec(bio).bv_len;
114 else /* dataless requests such as discard */
115 return bio->bi_iter.bi_size;
118 static inline void *bio_data(struct bio *bio)
120 if (bio_has_data(bio))
121 return page_address(bio_page(bio)) + bio_offset(bio);
126 static inline bool bio_full(struct bio *bio)
128 return bio->bi_vcnt >= bio->bi_max_vecs;
131 #define mp_bvec_for_each_segment(bv, bvl, i, iter_all) \
132 for (bv = bvec_init_iter_all(&iter_all); \
133 (iter_all.done < (bvl)->bv_len) && \
134 (mp_bvec_next_segment((bvl), &iter_all), 1); \
135 iter_all.done += bv->bv_len, i += 1)
138 * drivers should _never_ use the all version - the bio may have been split
139 * before it got to the driver and the driver won't own all of it
141 #define bio_for_each_segment_all(bvl, bio, i, iter_all) \
142 for (i = 0, iter_all.idx = 0; iter_all.idx < (bio)->bi_vcnt; iter_all.idx++) \
143 mp_bvec_for_each_segment(bvl, &((bio)->bi_io_vec[iter_all.idx]), i, iter_all)
145 static inline void bio_advance_iter(struct bio *bio, struct bvec_iter *iter,
148 iter->bi_sector += bytes >> 9;
150 if (bio_no_advance_iter(bio))
151 iter->bi_size -= bytes;
153 bvec_iter_advance(bio->bi_io_vec, iter, bytes);
154 /* TODO: It is reasonable to complete bio with error here. */
157 #define __bio_for_each_segment(bvl, bio, iter, start) \
158 for (iter = (start); \
160 ((bvl = bio_iter_iovec((bio), (iter))), 1); \
161 bio_advance_iter((bio), &(iter), (bvl).bv_len))
163 #define bio_for_each_segment(bvl, bio, iter) \
164 __bio_for_each_segment(bvl, bio, iter, (bio)->bi_iter)
166 #define __bio_for_each_bvec(bvl, bio, iter, start) \
167 for (iter = (start); \
169 ((bvl = mp_bvec_iter_bvec((bio)->bi_io_vec, (iter))), 1); \
170 bio_advance_iter((bio), &(iter), (bvl).bv_len))
172 /* iterate over multi-page bvec */
173 #define bio_for_each_bvec(bvl, bio, iter) \
174 __bio_for_each_bvec(bvl, bio, iter, (bio)->bi_iter)
176 #define bio_iter_last(bvec, iter) ((iter).bi_size == (bvec).bv_len)
178 static inline unsigned bio_segments(struct bio *bio)
182 struct bvec_iter iter;
185 * We special case discard/write same/write zeroes, because they
186 * interpret bi_size differently:
189 switch (bio_op(bio)) {
191 case REQ_OP_SECURE_ERASE:
192 case REQ_OP_WRITE_ZEROES:
194 case REQ_OP_WRITE_SAME:
200 bio_for_each_segment(bv, bio, iter)
207 * get a reference to a bio, so it won't disappear. the intended use is
211 * submit_bio(rw, bio);
212 * if (bio->bi_flags ...)
216 * without the bio_get(), it could potentially complete I/O before submit_bio
217 * returns. and then bio would be freed memory when if (bio->bi_flags ...)
220 static inline void bio_get(struct bio *bio)
222 bio->bi_flags |= (1 << BIO_REFFED);
223 smp_mb__before_atomic();
224 atomic_inc(&bio->__bi_cnt);
227 static inline void bio_cnt_set(struct bio *bio, unsigned int count)
230 bio->bi_flags |= (1 << BIO_REFFED);
231 smp_mb__before_atomic();
233 atomic_set(&bio->__bi_cnt, count);
236 static inline bool bio_flagged(struct bio *bio, unsigned int bit)
238 return (bio->bi_flags & (1U << bit)) != 0;
241 static inline void bio_set_flag(struct bio *bio, unsigned int bit)
243 bio->bi_flags |= (1U << bit);
246 static inline void bio_clear_flag(struct bio *bio, unsigned int bit)
248 bio->bi_flags &= ~(1U << bit);
251 static inline void bio_get_first_bvec(struct bio *bio, struct bio_vec *bv)
253 *bv = bio_iovec(bio);
256 static inline void bio_get_last_bvec(struct bio *bio, struct bio_vec *bv)
258 struct bvec_iter iter = bio->bi_iter;
261 if (unlikely(!bio_multiple_segments(bio))) {
262 *bv = bio_iovec(bio);
266 bio_advance_iter(bio, &iter, iter.bi_size);
268 if (!iter.bi_bvec_done)
269 idx = iter.bi_idx - 1;
270 else /* in the middle of bvec */
273 *bv = bio->bi_io_vec[idx];
276 * iter.bi_bvec_done records actual length of the last bvec
277 * if this bio ends in the middle of one io vector
279 if (iter.bi_bvec_done)
280 bv->bv_len = iter.bi_bvec_done;
283 static inline struct bio_vec *bio_first_bvec_all(struct bio *bio)
285 WARN_ON_ONCE(bio_flagged(bio, BIO_CLONED));
286 return bio->bi_io_vec;
289 static inline struct page *bio_first_page_all(struct bio *bio)
291 return bio_first_bvec_all(bio)->bv_page;
294 static inline struct bio_vec *bio_last_bvec_all(struct bio *bio)
296 WARN_ON_ONCE(bio_flagged(bio, BIO_CLONED));
297 return &bio->bi_io_vec[bio->bi_vcnt - 1];
301 BIP_BLOCK_INTEGRITY = 1 << 0, /* block layer owns integrity data */
302 BIP_MAPPED_INTEGRITY = 1 << 1, /* ref tag has been remapped */
303 BIP_CTRL_NOCHECK = 1 << 2, /* disable HBA integrity checking */
304 BIP_DISK_NOCHECK = 1 << 3, /* disable disk integrity checking */
305 BIP_IP_CHECKSUM = 1 << 4, /* IP checksum */
309 * bio integrity payload
311 struct bio_integrity_payload {
312 struct bio *bip_bio; /* parent bio */
314 struct bvec_iter bip_iter;
316 unsigned short bip_slab; /* slab the bip came from */
317 unsigned short bip_vcnt; /* # of integrity bio_vecs */
318 unsigned short bip_max_vcnt; /* integrity bio_vec slots */
319 unsigned short bip_flags; /* control flags */
321 struct bvec_iter bio_iter; /* for rewinding parent bio */
323 struct work_struct bip_work; /* I/O completion */
325 struct bio_vec *bip_vec;
326 struct bio_vec bip_inline_vecs[0];/* embedded bvec array */
329 #if defined(CONFIG_BLK_DEV_INTEGRITY)
331 static inline struct bio_integrity_payload *bio_integrity(struct bio *bio)
333 if (bio->bi_opf & REQ_INTEGRITY)
334 return bio->bi_integrity;
339 static inline bool bio_integrity_flagged(struct bio *bio, enum bip_flags flag)
341 struct bio_integrity_payload *bip = bio_integrity(bio);
344 return bip->bip_flags & flag;
349 static inline sector_t bip_get_seed(struct bio_integrity_payload *bip)
351 return bip->bip_iter.bi_sector;
354 static inline void bip_set_seed(struct bio_integrity_payload *bip,
357 bip->bip_iter.bi_sector = seed;
360 #endif /* CONFIG_BLK_DEV_INTEGRITY */
362 extern void bio_trim(struct bio *bio, int offset, int size);
363 extern struct bio *bio_split(struct bio *bio, int sectors,
364 gfp_t gfp, struct bio_set *bs);
367 * bio_next_split - get next @sectors from a bio, splitting if necessary
369 * @sectors: number of sectors to split from the front of @bio
371 * @bs: bio set to allocate from
373 * Returns a bio representing the next @sectors of @bio - if the bio is smaller
374 * than @sectors, returns the original bio unchanged.
376 static inline struct bio *bio_next_split(struct bio *bio, int sectors,
377 gfp_t gfp, struct bio_set *bs)
379 if (sectors >= bio_sectors(bio))
382 return bio_split(bio, sectors, gfp, bs);
386 BIOSET_NEED_BVECS = BIT(0),
387 BIOSET_NEED_RESCUER = BIT(1),
389 extern int bioset_init(struct bio_set *, unsigned int, unsigned int, int flags);
390 extern void bioset_exit(struct bio_set *);
391 extern int biovec_init_pool(mempool_t *pool, int pool_entries);
392 extern int bioset_init_from_src(struct bio_set *bs, struct bio_set *src);
394 extern struct bio *bio_alloc_bioset(gfp_t, unsigned int, struct bio_set *);
395 extern void bio_put(struct bio *);
397 extern void __bio_clone_fast(struct bio *, struct bio *);
398 extern struct bio *bio_clone_fast(struct bio *, gfp_t, struct bio_set *);
400 extern struct bio_set fs_bio_set;
402 static inline struct bio *bio_alloc(gfp_t gfp_mask, unsigned int nr_iovecs)
404 return bio_alloc_bioset(gfp_mask, nr_iovecs, &fs_bio_set);
407 static inline struct bio *bio_kmalloc(gfp_t gfp_mask, unsigned int nr_iovecs)
409 return bio_alloc_bioset(gfp_mask, nr_iovecs, NULL);
412 extern blk_qc_t submit_bio(struct bio *);
414 extern void bio_endio(struct bio *);
416 static inline void bio_io_error(struct bio *bio)
418 bio->bi_status = BLK_STS_IOERR;
422 static inline void bio_wouldblock_error(struct bio *bio)
424 bio->bi_status = BLK_STS_AGAIN;
428 struct request_queue;
429 extern int bio_phys_segments(struct request_queue *, struct bio *);
431 extern int submit_bio_wait(struct bio *bio);
432 extern void bio_advance(struct bio *, unsigned);
434 extern void bio_init(struct bio *bio, struct bio_vec *table,
435 unsigned short max_vecs);
436 extern void bio_uninit(struct bio *);
437 extern void bio_reset(struct bio *);
438 void bio_chain(struct bio *, struct bio *);
440 extern int bio_add_page(struct bio *, struct page *, unsigned int,unsigned int);
441 extern int bio_add_pc_page(struct request_queue *, struct bio *, struct page *,
442 unsigned int, unsigned int);
443 bool __bio_try_merge_page(struct bio *bio, struct page *page,
444 unsigned int len, unsigned int off);
445 void __bio_add_page(struct bio *bio, struct page *page,
446 unsigned int len, unsigned int off);
447 int bio_iov_iter_get_pages(struct bio *bio, struct iov_iter *iter);
449 extern struct bio *bio_map_user_iov(struct request_queue *,
450 struct iov_iter *, gfp_t);
451 extern void bio_unmap_user(struct bio *);
452 extern struct bio *bio_map_kern(struct request_queue *, void *, unsigned int,
454 extern struct bio *bio_copy_kern(struct request_queue *, void *, unsigned int,
456 extern void bio_set_pages_dirty(struct bio *bio);
457 extern void bio_check_pages_dirty(struct bio *bio);
459 void generic_start_io_acct(struct request_queue *q, int op,
460 unsigned long sectors, struct hd_struct *part);
461 void generic_end_io_acct(struct request_queue *q, int op,
462 struct hd_struct *part,
463 unsigned long start_time);
465 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
466 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
468 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
469 extern void bio_flush_dcache_pages(struct bio *bi);
471 static inline void bio_flush_dcache_pages(struct bio *bi)
476 extern void bio_copy_data_iter(struct bio *dst, struct bvec_iter *dst_iter,
477 struct bio *src, struct bvec_iter *src_iter);
478 extern void bio_copy_data(struct bio *dst, struct bio *src);
479 extern void bio_list_copy_data(struct bio *dst, struct bio *src);
480 extern void bio_free_pages(struct bio *bio);
482 extern struct bio *bio_copy_user_iov(struct request_queue *,
483 struct rq_map_data *,
486 extern int bio_uncopy_user(struct bio *);
487 void zero_fill_bio_iter(struct bio *bio, struct bvec_iter iter);
489 static inline void zero_fill_bio(struct bio *bio)
491 zero_fill_bio_iter(bio, bio->bi_iter);
494 extern struct bio_vec *bvec_alloc(gfp_t, int, unsigned long *, mempool_t *);
495 extern void bvec_free(mempool_t *, struct bio_vec *, unsigned int);
496 extern unsigned int bvec_nr_vecs(unsigned short idx);
497 extern const char *bio_devname(struct bio *bio, char *buffer);
499 #define bio_set_dev(bio, bdev) \
501 if ((bio)->bi_disk != (bdev)->bd_disk) \
502 bio_clear_flag(bio, BIO_THROTTLED);\
503 (bio)->bi_disk = (bdev)->bd_disk; \
504 (bio)->bi_partno = (bdev)->bd_partno; \
505 bio_associate_blkg(bio); \
508 #define bio_copy_dev(dst, src) \
510 (dst)->bi_disk = (src)->bi_disk; \
511 (dst)->bi_partno = (src)->bi_partno; \
512 bio_clone_blkg_association(dst, src); \
515 #define bio_dev(bio) \
516 disk_devt((bio)->bi_disk)
518 #if defined(CONFIG_MEMCG) && defined(CONFIG_BLK_CGROUP)
519 void bio_associate_blkg_from_page(struct bio *bio, struct page *page);
521 static inline void bio_associate_blkg_from_page(struct bio *bio,
522 struct page *page) { }
525 #ifdef CONFIG_BLK_CGROUP
526 void bio_disassociate_blkg(struct bio *bio);
527 void bio_associate_blkg(struct bio *bio);
528 void bio_associate_blkg_from_css(struct bio *bio,
529 struct cgroup_subsys_state *css);
530 void bio_clone_blkg_association(struct bio *dst, struct bio *src);
531 #else /* CONFIG_BLK_CGROUP */
532 static inline void bio_disassociate_blkg(struct bio *bio) { }
533 static inline void bio_associate_blkg(struct bio *bio) { }
534 static inline void bio_associate_blkg_from_css(struct bio *bio,
535 struct cgroup_subsys_state *css)
537 static inline void bio_clone_blkg_association(struct bio *dst,
539 #endif /* CONFIG_BLK_CGROUP */
541 #ifdef CONFIG_HIGHMEM
543 * remember never ever reenable interrupts between a bvec_kmap_irq and
546 static inline char *bvec_kmap_irq(struct bio_vec *bvec, unsigned long *flags)
551 * might not be a highmem page, but the preempt/irq count
552 * balancing is a lot nicer this way
554 local_irq_save(*flags);
555 addr = (unsigned long) kmap_atomic(bvec->bv_page);
557 BUG_ON(addr & ~PAGE_MASK);
559 return (char *) addr + bvec->bv_offset;
562 static inline void bvec_kunmap_irq(char *buffer, unsigned long *flags)
564 unsigned long ptr = (unsigned long) buffer & PAGE_MASK;
566 kunmap_atomic((void *) ptr);
567 local_irq_restore(*flags);
571 static inline char *bvec_kmap_irq(struct bio_vec *bvec, unsigned long *flags)
573 return page_address(bvec->bv_page) + bvec->bv_offset;
576 static inline void bvec_kunmap_irq(char *buffer, unsigned long *flags)
583 * BIO list management for use by remapping drivers (e.g. DM or MD) and loop.
585 * A bio_list anchors a singly-linked list of bios chained through the bi_next
586 * member of the bio. The bio_list also caches the last list member to allow
587 * fast access to the tail.
594 static inline int bio_list_empty(const struct bio_list *bl)
596 return bl->head == NULL;
599 static inline void bio_list_init(struct bio_list *bl)
601 bl->head = bl->tail = NULL;
604 #define BIO_EMPTY_LIST { NULL, NULL }
606 #define bio_list_for_each(bio, bl) \
607 for (bio = (bl)->head; bio; bio = bio->bi_next)
609 static inline unsigned bio_list_size(const struct bio_list *bl)
614 bio_list_for_each(bio, bl)
620 static inline void bio_list_add(struct bio_list *bl, struct bio *bio)
625 bl->tail->bi_next = bio;
632 static inline void bio_list_add_head(struct bio_list *bl, struct bio *bio)
634 bio->bi_next = bl->head;
642 static inline void bio_list_merge(struct bio_list *bl, struct bio_list *bl2)
648 bl->tail->bi_next = bl2->head;
650 bl->head = bl2->head;
652 bl->tail = bl2->tail;
655 static inline void bio_list_merge_head(struct bio_list *bl,
656 struct bio_list *bl2)
662 bl2->tail->bi_next = bl->head;
664 bl->tail = bl2->tail;
666 bl->head = bl2->head;
669 static inline struct bio *bio_list_peek(struct bio_list *bl)
674 static inline struct bio *bio_list_pop(struct bio_list *bl)
676 struct bio *bio = bl->head;
679 bl->head = bl->head->bi_next;
689 static inline struct bio *bio_list_get(struct bio_list *bl)
691 struct bio *bio = bl->head;
693 bl->head = bl->tail = NULL;
699 * Increment chain count for the bio. Make sure the CHAIN flag update
700 * is visible before the raised count.
702 static inline void bio_inc_remaining(struct bio *bio)
704 bio_set_flag(bio, BIO_CHAIN);
705 smp_mb__before_atomic();
706 atomic_inc(&bio->__bi_remaining);
710 * bio_set is used to allow other portions of the IO system to
711 * allocate their own private memory pools for bio and iovec structures.
712 * These memory pools in turn all allocate from the bio_slab
713 * and the bvec_slabs[].
715 #define BIO_POOL_SIZE 2
718 struct kmem_cache *bio_slab;
719 unsigned int front_pad;
723 #if defined(CONFIG_BLK_DEV_INTEGRITY)
724 mempool_t bio_integrity_pool;
725 mempool_t bvec_integrity_pool;
729 * Deadlock avoidance for stacking block drivers: see comments in
730 * bio_alloc_bioset() for details
732 spinlock_t rescue_lock;
733 struct bio_list rescue_list;
734 struct work_struct rescue_work;
735 struct workqueue_struct *rescue_workqueue;
741 struct kmem_cache *slab;
744 static inline bool bioset_initialized(struct bio_set *bs)
746 return bs->bio_slab != NULL;
750 * a small number of entries is fine, not going to be performance critical.
751 * basically we just need to survive
753 #define BIO_SPLIT_ENTRIES 2
755 #if defined(CONFIG_BLK_DEV_INTEGRITY)
757 #define bip_for_each_vec(bvl, bip, iter) \
758 for_each_bvec(bvl, (bip)->bip_vec, iter, (bip)->bip_iter)
760 #define bio_for_each_integrity_vec(_bvl, _bio, _iter) \
762 bip_for_each_vec(_bvl, _bio->bi_integrity, _iter)
764 extern struct bio_integrity_payload *bio_integrity_alloc(struct bio *, gfp_t, unsigned int);
765 extern int bio_integrity_add_page(struct bio *, struct page *, unsigned int, unsigned int);
766 extern bool bio_integrity_prep(struct bio *);
767 extern void bio_integrity_advance(struct bio *, unsigned int);
768 extern void bio_integrity_trim(struct bio *);
769 extern int bio_integrity_clone(struct bio *, struct bio *, gfp_t);
770 extern int bioset_integrity_create(struct bio_set *, int);
771 extern void bioset_integrity_free(struct bio_set *);
772 extern void bio_integrity_init(void);
774 #else /* CONFIG_BLK_DEV_INTEGRITY */
776 static inline void *bio_integrity(struct bio *bio)
781 static inline int bioset_integrity_create(struct bio_set *bs, int pool_size)
786 static inline void bioset_integrity_free (struct bio_set *bs)
791 static inline bool bio_integrity_prep(struct bio *bio)
796 static inline int bio_integrity_clone(struct bio *bio, struct bio *bio_src,
802 static inline void bio_integrity_advance(struct bio *bio,
803 unsigned int bytes_done)
808 static inline void bio_integrity_trim(struct bio *bio)
813 static inline void bio_integrity_init(void)
818 static inline bool bio_integrity_flagged(struct bio *bio, enum bip_flags flag)
823 static inline void *bio_integrity_alloc(struct bio * bio, gfp_t gfp,
826 return ERR_PTR(-EINVAL);
829 static inline int bio_integrity_add_page(struct bio *bio, struct page *page,
830 unsigned int len, unsigned int offset)
835 #endif /* CONFIG_BLK_DEV_INTEGRITY */
837 #endif /* CONFIG_BLOCK */
838 #endif /* __LINUX_BIO_H */