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>
24 #include <linux/bug.h>
30 /* struct bio, bio_vec and BIO_* flags are defined in blk_types.h */
31 #include <linux/blk_types.h>
36 #define BIO_BUG_ON BUG_ON
41 #ifdef CONFIG_THP_SWAP
42 #if HPAGE_PMD_NR > 256
43 #define BIO_MAX_PAGES HPAGE_PMD_NR
45 #define BIO_MAX_PAGES 256
48 #define BIO_MAX_PAGES 256
51 #define bio_prio(bio) (bio)->bi_ioprio
52 #define bio_set_prio(bio, prio) ((bio)->bi_ioprio = prio)
54 #define bio_iter_iovec(bio, iter) \
55 bvec_iter_bvec((bio)->bi_io_vec, (iter))
57 #define bio_iter_page(bio, iter) \
58 bvec_iter_page((bio)->bi_io_vec, (iter))
59 #define bio_iter_len(bio, iter) \
60 bvec_iter_len((bio)->bi_io_vec, (iter))
61 #define bio_iter_offset(bio, iter) \
62 bvec_iter_offset((bio)->bi_io_vec, (iter))
64 #define bio_page(bio) bio_iter_page((bio), (bio)->bi_iter)
65 #define bio_offset(bio) bio_iter_offset((bio), (bio)->bi_iter)
66 #define bio_iovec(bio) bio_iter_iovec((bio), (bio)->bi_iter)
68 #define bio_multiple_segments(bio) \
69 ((bio)->bi_iter.bi_size != bio_iovec(bio).bv_len)
70 #define bio_sectors(bio) ((bio)->bi_iter.bi_size >> 9)
71 #define bio_end_sector(bio) ((bio)->bi_iter.bi_sector + bio_sectors((bio)))
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);
129 #define bvec_to_phys(bv) (page_to_phys((bv)->bv_page) + (unsigned long) (bv)->bv_offset)
132 * queues that have highmem support enabled may still need to revert to
133 * PIO transfers occasionally and thus map high pages temporarily. For
134 * permanent PIO fall back, user is probably better off disabling highmem
135 * I/O completely on that queue (see ide-dma for example)
137 #define __bio_kmap_atomic(bio, iter) \
138 (kmap_atomic(bio_iter_iovec((bio), (iter)).bv_page) + \
139 bio_iter_iovec((bio), (iter)).bv_offset)
141 #define __bio_kunmap_atomic(addr) kunmap_atomic(addr)
147 /* Default implementation of BIOVEC_PHYS_MERGEABLE */
148 #define __BIOVEC_PHYS_MERGEABLE(vec1, vec2) \
149 ((bvec_to_phys((vec1)) + (vec1)->bv_len) == bvec_to_phys((vec2)))
152 * allow arch override, for eg virtualized architectures (put in asm/io.h)
154 #ifndef BIOVEC_PHYS_MERGEABLE
155 #define BIOVEC_PHYS_MERGEABLE(vec1, vec2) \
156 __BIOVEC_PHYS_MERGEABLE(vec1, vec2)
159 #define __BIO_SEG_BOUNDARY(addr1, addr2, mask) \
160 (((addr1) | (mask)) == (((addr2) - 1) | (mask)))
161 #define BIOVEC_SEG_BOUNDARY(q, b1, b2) \
162 __BIO_SEG_BOUNDARY(bvec_to_phys((b1)), bvec_to_phys((b2)) + (b2)->bv_len, queue_segment_boundary((q)))
165 * drivers should _never_ use the all version - the bio may have been split
166 * before it got to the driver and the driver won't own all of it
168 #define bio_for_each_segment_all(bvl, bio, i) \
169 for (i = 0, bvl = (bio)->bi_io_vec; i < (bio)->bi_vcnt; i++, bvl++)
171 static inline void bio_advance_iter(struct bio *bio, struct bvec_iter *iter,
174 iter->bi_sector += bytes >> 9;
176 if (bio_no_advance_iter(bio)) {
177 iter->bi_size -= bytes;
178 iter->bi_done += bytes;
180 bvec_iter_advance(bio->bi_io_vec, iter, bytes);
181 /* TODO: It is reasonable to complete bio with error here. */
185 static inline bool bio_rewind_iter(struct bio *bio, struct bvec_iter *iter,
188 iter->bi_sector -= bytes >> 9;
190 if (bio_no_advance_iter(bio)) {
191 iter->bi_size += bytes;
192 iter->bi_done -= bytes;
196 return bvec_iter_rewind(bio->bi_io_vec, iter, bytes);
199 #define __bio_for_each_segment(bvl, bio, iter, start) \
200 for (iter = (start); \
202 ((bvl = bio_iter_iovec((bio), (iter))), 1); \
203 bio_advance_iter((bio), &(iter), (bvl).bv_len))
205 #define bio_for_each_segment(bvl, bio, iter) \
206 __bio_for_each_segment(bvl, bio, iter, (bio)->bi_iter)
208 #define bio_iter_last(bvec, iter) ((iter).bi_size == (bvec).bv_len)
210 static inline unsigned bio_segments(struct bio *bio)
214 struct bvec_iter iter;
217 * We special case discard/write same/write zeroes, because they
218 * interpret bi_size differently:
221 switch (bio_op(bio)) {
223 case REQ_OP_SECURE_ERASE:
224 case REQ_OP_WRITE_ZEROES:
226 case REQ_OP_WRITE_SAME:
232 bio_for_each_segment(bv, bio, iter)
239 * get a reference to a bio, so it won't disappear. the intended use is
243 * submit_bio(rw, bio);
244 * if (bio->bi_flags ...)
248 * without the bio_get(), it could potentially complete I/O before submit_bio
249 * returns. and then bio would be freed memory when if (bio->bi_flags ...)
252 static inline void bio_get(struct bio *bio)
254 bio->bi_flags |= (1 << BIO_REFFED);
255 smp_mb__before_atomic();
256 atomic_inc(&bio->__bi_cnt);
259 static inline void bio_cnt_set(struct bio *bio, unsigned int count)
262 bio->bi_flags |= (1 << BIO_REFFED);
263 smp_mb__before_atomic();
265 atomic_set(&bio->__bi_cnt, count);
268 static inline bool bio_flagged(struct bio *bio, unsigned int bit)
270 return (bio->bi_flags & (1U << bit)) != 0;
273 static inline void bio_set_flag(struct bio *bio, unsigned int bit)
275 bio->bi_flags |= (1U << bit);
278 static inline void bio_clear_flag(struct bio *bio, unsigned int bit)
280 bio->bi_flags &= ~(1U << bit);
283 static inline void bio_get_first_bvec(struct bio *bio, struct bio_vec *bv)
285 *bv = bio_iovec(bio);
288 static inline void bio_get_last_bvec(struct bio *bio, struct bio_vec *bv)
290 struct bvec_iter iter = bio->bi_iter;
293 if (unlikely(!bio_multiple_segments(bio))) {
294 *bv = bio_iovec(bio);
298 bio_advance_iter(bio, &iter, iter.bi_size);
300 if (!iter.bi_bvec_done)
301 idx = iter.bi_idx - 1;
302 else /* in the middle of bvec */
305 *bv = bio->bi_io_vec[idx];
308 * iter.bi_bvec_done records actual length of the last bvec
309 * if this bio ends in the middle of one io vector
311 if (iter.bi_bvec_done)
312 bv->bv_len = iter.bi_bvec_done;
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 work_struct bip_work; /* I/O completion */
338 struct bio_vec *bip_vec;
339 struct bio_vec bip_inline_vecs[0];/* embedded bvec array */
342 #if defined(CONFIG_BLK_DEV_INTEGRITY)
344 static inline struct bio_integrity_payload *bio_integrity(struct bio *bio)
346 if (bio->bi_opf & REQ_INTEGRITY)
347 return bio->bi_integrity;
352 static inline bool bio_integrity_flagged(struct bio *bio, enum bip_flags flag)
354 struct bio_integrity_payload *bip = bio_integrity(bio);
357 return bip->bip_flags & flag;
362 static inline sector_t bip_get_seed(struct bio_integrity_payload *bip)
364 return bip->bip_iter.bi_sector;
367 static inline void bip_set_seed(struct bio_integrity_payload *bip,
370 bip->bip_iter.bi_sector = seed;
373 #endif /* CONFIG_BLK_DEV_INTEGRITY */
375 extern void bio_trim(struct bio *bio, int offset, int size);
376 extern struct bio *bio_split(struct bio *bio, int sectors,
377 gfp_t gfp, struct bio_set *bs);
380 * bio_next_split - get next @sectors from a bio, splitting if necessary
382 * @sectors: number of sectors to split from the front of @bio
384 * @bs: bio set to allocate from
386 * Returns a bio representing the next @sectors of @bio - if the bio is smaller
387 * than @sectors, returns the original bio unchanged.
389 static inline struct bio *bio_next_split(struct bio *bio, int sectors,
390 gfp_t gfp, struct bio_set *bs)
392 if (sectors >= bio_sectors(bio))
395 return bio_split(bio, sectors, gfp, bs);
398 extern struct bio_set *bioset_create(unsigned int, unsigned int, int flags);
400 BIOSET_NEED_BVECS = BIT(0),
401 BIOSET_NEED_RESCUER = BIT(1),
403 extern void bioset_free(struct bio_set *);
404 extern mempool_t *biovec_create_pool(int pool_entries);
406 extern struct bio *bio_alloc_bioset(gfp_t, unsigned int, struct bio_set *);
407 extern void bio_put(struct bio *);
409 extern void __bio_clone_fast(struct bio *, struct bio *);
410 extern struct bio *bio_clone_fast(struct bio *, gfp_t, struct bio_set *);
411 extern struct bio *bio_clone_bioset(struct bio *, gfp_t, struct bio_set *bs);
413 extern struct bio_set *fs_bio_set;
415 static inline struct bio *bio_alloc(gfp_t gfp_mask, unsigned int nr_iovecs)
417 return bio_alloc_bioset(gfp_mask, nr_iovecs, fs_bio_set);
420 static inline struct bio *bio_kmalloc(gfp_t gfp_mask, unsigned int nr_iovecs)
422 return bio_alloc_bioset(gfp_mask, nr_iovecs, NULL);
425 static inline struct bio *bio_clone_kmalloc(struct bio *bio, gfp_t gfp_mask)
427 return bio_clone_bioset(bio, gfp_mask, NULL);
431 extern blk_qc_t submit_bio(struct bio *);
433 extern void bio_endio(struct bio *);
435 static inline void bio_io_error(struct bio *bio)
437 bio->bi_status = BLK_STS_IOERR;
441 static inline void bio_wouldblock_error(struct bio *bio)
443 bio->bi_status = BLK_STS_AGAIN;
447 struct request_queue;
448 extern int bio_phys_segments(struct request_queue *, struct bio *);
450 extern int submit_bio_wait(struct bio *bio);
451 extern void bio_advance(struct bio *, unsigned);
453 extern void bio_init(struct bio *bio, struct bio_vec *table,
454 unsigned short max_vecs);
455 extern void bio_uninit(struct bio *);
456 extern void bio_reset(struct bio *);
457 void bio_chain(struct bio *, struct bio *);
459 extern int bio_add_page(struct bio *, struct page *, unsigned int,unsigned int);
460 extern int bio_add_pc_page(struct request_queue *, struct bio *, struct page *,
461 unsigned int, unsigned int);
462 int bio_iov_iter_get_pages(struct bio *bio, struct iov_iter *iter);
464 extern struct bio *bio_map_user_iov(struct request_queue *,
465 const struct iov_iter *, gfp_t);
466 extern void bio_unmap_user(struct bio *);
467 extern struct bio *bio_map_kern(struct request_queue *, void *, unsigned int,
469 extern struct bio *bio_copy_kern(struct request_queue *, void *, unsigned int,
471 extern void bio_set_pages_dirty(struct bio *bio);
472 extern void bio_check_pages_dirty(struct bio *bio);
474 void generic_start_io_acct(struct request_queue *q, int rw,
475 unsigned long sectors, struct hd_struct *part);
476 void generic_end_io_acct(struct request_queue *q, int rw,
477 struct hd_struct *part,
478 unsigned long start_time);
480 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
481 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
483 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
484 extern void bio_flush_dcache_pages(struct bio *bi);
486 static inline void bio_flush_dcache_pages(struct bio *bi)
491 extern void bio_copy_data(struct bio *dst, struct bio *src);
492 extern int bio_alloc_pages(struct bio *bio, gfp_t gfp);
493 extern void bio_free_pages(struct bio *bio);
495 extern struct bio *bio_copy_user_iov(struct request_queue *,
496 struct rq_map_data *,
497 const struct iov_iter *,
499 extern int bio_uncopy_user(struct bio *);
500 void zero_fill_bio(struct bio *bio);
501 extern struct bio_vec *bvec_alloc(gfp_t, int, unsigned long *, mempool_t *);
502 extern void bvec_free(mempool_t *, struct bio_vec *, unsigned int);
503 extern unsigned int bvec_nr_vecs(unsigned short idx);
505 #define bio_set_dev(bio, bdev) \
507 (bio)->bi_disk = (bdev)->bd_disk; \
508 (bio)->bi_partno = (bdev)->bd_partno; \
511 #define bio_copy_dev(dst, src) \
513 (dst)->bi_disk = (src)->bi_disk; \
514 (dst)->bi_partno = (src)->bi_partno; \
517 #define bio_dev(bio) \
518 disk_devt((bio)->bi_disk)
520 #define bio_devname(bio, buf) \
521 __bdevname(bio_dev(bio), (buf))
523 #ifdef CONFIG_BLK_CGROUP
524 int bio_associate_blkcg(struct bio *bio, struct cgroup_subsys_state *blkcg_css);
525 int bio_associate_current(struct bio *bio);
526 void bio_disassociate_task(struct bio *bio);
527 void bio_clone_blkcg_association(struct bio *dst, struct bio *src);
528 #else /* CONFIG_BLK_CGROUP */
529 static inline int bio_associate_blkcg(struct bio *bio,
530 struct cgroup_subsys_state *blkcg_css) { return 0; }
531 static inline int bio_associate_current(struct bio *bio) { return -ENOENT; }
532 static inline void bio_disassociate_task(struct bio *bio) { }
533 static inline void bio_clone_blkcg_association(struct bio *dst,
535 #endif /* CONFIG_BLK_CGROUP */
537 #ifdef CONFIG_HIGHMEM
539 * remember never ever reenable interrupts between a bvec_kmap_irq and
542 static inline char *bvec_kmap_irq(struct bio_vec *bvec, unsigned long *flags)
547 * might not be a highmem page, but the preempt/irq count
548 * balancing is a lot nicer this way
550 local_irq_save(*flags);
551 addr = (unsigned long) kmap_atomic(bvec->bv_page);
553 BUG_ON(addr & ~PAGE_MASK);
555 return (char *) addr + bvec->bv_offset;
558 static inline void bvec_kunmap_irq(char *buffer, unsigned long *flags)
560 unsigned long ptr = (unsigned long) buffer & PAGE_MASK;
562 kunmap_atomic((void *) ptr);
563 local_irq_restore(*flags);
567 static inline char *bvec_kmap_irq(struct bio_vec *bvec, unsigned long *flags)
569 return page_address(bvec->bv_page) + bvec->bv_offset;
572 static inline void bvec_kunmap_irq(char *buffer, unsigned long *flags)
578 static inline char *__bio_kmap_irq(struct bio *bio, struct bvec_iter iter,
579 unsigned long *flags)
581 return bvec_kmap_irq(&bio_iter_iovec(bio, iter), flags);
583 #define __bio_kunmap_irq(buf, flags) bvec_kunmap_irq(buf, flags)
585 #define bio_kmap_irq(bio, flags) \
586 __bio_kmap_irq((bio), (bio)->bi_iter, (flags))
587 #define bio_kunmap_irq(buf,flags) __bio_kunmap_irq(buf, flags)
590 * BIO list management for use by remapping drivers (e.g. DM or MD) and loop.
592 * A bio_list anchors a singly-linked list of bios chained through the bi_next
593 * member of the bio. The bio_list also caches the last list member to allow
594 * fast access to the tail.
601 static inline int bio_list_empty(const struct bio_list *bl)
603 return bl->head == NULL;
606 static inline void bio_list_init(struct bio_list *bl)
608 bl->head = bl->tail = NULL;
611 #define BIO_EMPTY_LIST { NULL, NULL }
613 #define bio_list_for_each(bio, bl) \
614 for (bio = (bl)->head; bio; bio = bio->bi_next)
616 static inline unsigned bio_list_size(const struct bio_list *bl)
621 bio_list_for_each(bio, bl)
627 static inline void bio_list_add(struct bio_list *bl, struct bio *bio)
632 bl->tail->bi_next = bio;
639 static inline void bio_list_add_head(struct bio_list *bl, struct bio *bio)
641 bio->bi_next = bl->head;
649 static inline void bio_list_merge(struct bio_list *bl, struct bio_list *bl2)
655 bl->tail->bi_next = bl2->head;
657 bl->head = bl2->head;
659 bl->tail = bl2->tail;
662 static inline void bio_list_merge_head(struct bio_list *bl,
663 struct bio_list *bl2)
669 bl2->tail->bi_next = bl->head;
671 bl->tail = bl2->tail;
673 bl->head = bl2->head;
676 static inline struct bio *bio_list_peek(struct bio_list *bl)
681 static inline struct bio *bio_list_pop(struct bio_list *bl)
683 struct bio *bio = bl->head;
686 bl->head = bl->head->bi_next;
696 static inline struct bio *bio_list_get(struct bio_list *bl)
698 struct bio *bio = bl->head;
700 bl->head = bl->tail = NULL;
706 * Increment chain count for the bio. Make sure the CHAIN flag update
707 * is visible before the raised count.
709 static inline void bio_inc_remaining(struct bio *bio)
711 bio_set_flag(bio, BIO_CHAIN);
712 smp_mb__before_atomic();
713 atomic_inc(&bio->__bi_remaining);
717 * bio_set is used to allow other portions of the IO system to
718 * allocate their own private memory pools for bio and iovec structures.
719 * These memory pools in turn all allocate from the bio_slab
720 * and the bvec_slabs[].
722 #define BIO_POOL_SIZE 2
725 struct kmem_cache *bio_slab;
726 unsigned int front_pad;
729 mempool_t *bvec_pool;
730 #if defined(CONFIG_BLK_DEV_INTEGRITY)
731 mempool_t *bio_integrity_pool;
732 mempool_t *bvec_integrity_pool;
736 * Deadlock avoidance for stacking block drivers: see comments in
737 * bio_alloc_bioset() for details
739 spinlock_t rescue_lock;
740 struct bio_list rescue_list;
741 struct work_struct rescue_work;
742 struct workqueue_struct *rescue_workqueue;
748 struct kmem_cache *slab;
752 * a small number of entries is fine, not going to be performance critical.
753 * basically we just need to survive
755 #define BIO_SPLIT_ENTRIES 2
757 #if defined(CONFIG_BLK_DEV_INTEGRITY)
759 #define bip_for_each_vec(bvl, bip, iter) \
760 for_each_bvec(bvl, (bip)->bip_vec, iter, (bip)->bip_iter)
762 #define bio_for_each_integrity_vec(_bvl, _bio, _iter) \
764 bip_for_each_vec(_bvl, _bio->bi_integrity, _iter)
766 extern struct bio_integrity_payload *bio_integrity_alloc(struct bio *, gfp_t, unsigned int);
767 extern int bio_integrity_add_page(struct bio *, struct page *, unsigned int, unsigned int);
768 extern bool bio_integrity_prep(struct bio *);
769 extern void bio_integrity_advance(struct bio *, unsigned int);
770 extern void bio_integrity_trim(struct bio *);
771 extern int bio_integrity_clone(struct bio *, struct bio *, gfp_t);
772 extern int bioset_integrity_create(struct bio_set *, int);
773 extern void bioset_integrity_free(struct bio_set *);
774 extern void bio_integrity_init(void);
776 #else /* CONFIG_BLK_DEV_INTEGRITY */
778 static inline void *bio_integrity(struct bio *bio)
783 static inline int bioset_integrity_create(struct bio_set *bs, int pool_size)
788 static inline void bioset_integrity_free (struct bio_set *bs)
793 static inline bool bio_integrity_prep(struct bio *bio)
798 static inline int bio_integrity_clone(struct bio *bio, struct bio *bio_src,
804 static inline void bio_integrity_advance(struct bio *bio,
805 unsigned int bytes_done)
810 static inline void bio_integrity_trim(struct bio *bio)
815 static inline void bio_integrity_init(void)
820 static inline bool bio_integrity_flagged(struct bio *bio, enum bip_flags flag)
825 static inline void *bio_integrity_alloc(struct bio * bio, gfp_t gfp,
828 return ERR_PTR(-EINVAL);
831 static inline int bio_integrity_add_page(struct bio *bio, struct page *page,
832 unsigned int len, unsigned int offset)
837 #endif /* CONFIG_BLK_DEV_INTEGRITY */
839 #endif /* CONFIG_BLOCK */
840 #endif /* __LINUX_BIO_H */