2 * Copyright (C) 2001 Jens Axboe <axboe@suse.de>
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 #define BIO_MAX_PAGES 256
43 #define bio_prio(bio) (bio)->bi_ioprio
44 #define bio_set_prio(bio, prio) ((bio)->bi_ioprio = prio)
46 #define bio_iter_iovec(bio, iter) \
47 bvec_iter_bvec((bio)->bi_io_vec, (iter))
49 #define bio_iter_page(bio, iter) \
50 bvec_iter_page((bio)->bi_io_vec, (iter))
51 #define bio_iter_len(bio, iter) \
52 bvec_iter_len((bio)->bi_io_vec, (iter))
53 #define bio_iter_offset(bio, iter) \
54 bvec_iter_offset((bio)->bi_io_vec, (iter))
56 #define bio_page(bio) bio_iter_page((bio), (bio)->bi_iter)
57 #define bio_offset(bio) bio_iter_offset((bio), (bio)->bi_iter)
58 #define bio_iovec(bio) bio_iter_iovec((bio), (bio)->bi_iter)
60 #define bio_multiple_segments(bio) \
61 ((bio)->bi_iter.bi_size != bio_iovec(bio).bv_len)
62 #define bio_sectors(bio) ((bio)->bi_iter.bi_size >> 9)
63 #define bio_end_sector(bio) ((bio)->bi_iter.bi_sector + bio_sectors((bio)))
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);
121 #define bio_to_phys(bio) (page_to_phys(bio_page((bio))) + (unsigned long) bio_offset((bio)))
122 #define bvec_to_phys(bv) (page_to_phys((bv)->bv_page) + (unsigned long) (bv)->bv_offset)
125 * queues that have highmem support enabled may still need to revert to
126 * PIO transfers occasionally and thus map high pages temporarily. For
127 * permanent PIO fall back, user is probably better off disabling highmem
128 * I/O completely on that queue (see ide-dma for example)
130 #define __bio_kmap_atomic(bio, iter) \
131 (kmap_atomic(bio_iter_iovec((bio), (iter)).bv_page) + \
132 bio_iter_iovec((bio), (iter)).bv_offset)
134 #define __bio_kunmap_atomic(addr) kunmap_atomic(addr)
140 /* Default implementation of BIOVEC_PHYS_MERGEABLE */
141 #define __BIOVEC_PHYS_MERGEABLE(vec1, vec2) \
142 ((bvec_to_phys((vec1)) + (vec1)->bv_len) == bvec_to_phys((vec2)))
145 * allow arch override, for eg virtualized architectures (put in asm/io.h)
147 #ifndef BIOVEC_PHYS_MERGEABLE
148 #define BIOVEC_PHYS_MERGEABLE(vec1, vec2) \
149 __BIOVEC_PHYS_MERGEABLE(vec1, vec2)
152 #define __BIO_SEG_BOUNDARY(addr1, addr2, mask) \
153 (((addr1) | (mask)) == (((addr2) - 1) | (mask)))
154 #define BIOVEC_SEG_BOUNDARY(q, b1, b2) \
155 __BIO_SEG_BOUNDARY(bvec_to_phys((b1)), bvec_to_phys((b2)) + (b2)->bv_len, queue_segment_boundary((q)))
158 * drivers should _never_ use the all version - the bio may have been split
159 * before it got to the driver and the driver won't own all of it
161 #define bio_for_each_segment_all(bvl, bio, i) \
162 for (i = 0, bvl = (bio)->bi_io_vec; i < (bio)->bi_vcnt; i++, bvl++)
164 static inline void bio_advance_iter(struct bio *bio, struct bvec_iter *iter,
167 iter->bi_sector += bytes >> 9;
169 if (bio_no_advance_iter(bio))
170 iter->bi_size -= bytes;
172 bvec_iter_advance(bio->bi_io_vec, iter, bytes);
175 #define __bio_for_each_segment(bvl, bio, iter, start) \
176 for (iter = (start); \
178 ((bvl = bio_iter_iovec((bio), (iter))), 1); \
179 bio_advance_iter((bio), &(iter), (bvl).bv_len))
181 #define bio_for_each_segment(bvl, bio, iter) \
182 __bio_for_each_segment(bvl, bio, iter, (bio)->bi_iter)
184 #define bio_iter_last(bvec, iter) ((iter).bi_size == (bvec).bv_len)
186 static inline unsigned __bio_segments(struct bio *bio, struct bvec_iter *bvec)
190 struct bvec_iter iter;
193 * We special case discard/write same/write zeroes, because they
194 * interpret bi_size differently:
197 switch (bio_op(bio)) {
199 case REQ_OP_SECURE_ERASE:
200 case REQ_OP_WRITE_ZEROES:
202 case REQ_OP_WRITE_SAME:
208 __bio_for_each_segment(bv, bio, iter, *bvec)
214 static inline unsigned bio_segments(struct bio *bio)
216 return __bio_segments(bio, &bio->bi_iter);
220 * get a reference to a bio, so it won't disappear. the intended use is
224 * submit_bio(rw, bio);
225 * if (bio->bi_flags ...)
229 * without the bio_get(), it could potentially complete I/O before submit_bio
230 * returns. and then bio would be freed memory when if (bio->bi_flags ...)
233 static inline void bio_get(struct bio *bio)
235 bio->bi_flags |= (1 << BIO_REFFED);
236 smp_mb__before_atomic();
237 atomic_inc(&bio->__bi_cnt);
240 static inline void bio_cnt_set(struct bio *bio, unsigned int count)
243 bio->bi_flags |= (1 << BIO_REFFED);
244 smp_mb__before_atomic();
246 atomic_set(&bio->__bi_cnt, count);
249 static inline bool bio_flagged(struct bio *bio, unsigned int bit)
251 return (bio->bi_flags & (1U << bit)) != 0;
254 static inline void bio_set_flag(struct bio *bio, unsigned int bit)
256 bio->bi_flags |= (1U << bit);
259 static inline void bio_clear_flag(struct bio *bio, unsigned int bit)
261 bio->bi_flags &= ~(1U << bit);
264 static inline void bio_get_first_bvec(struct bio *bio, struct bio_vec *bv)
266 *bv = bio_iovec(bio);
269 static inline void bio_get_last_bvec(struct bio *bio, struct bio_vec *bv)
271 struct bvec_iter iter = bio->bi_iter;
274 if (unlikely(!bio_multiple_segments(bio))) {
275 *bv = bio_iovec(bio);
279 bio_advance_iter(bio, &iter, iter.bi_size);
281 if (!iter.bi_bvec_done)
282 idx = iter.bi_idx - 1;
283 else /* in the middle of bvec */
286 *bv = bio->bi_io_vec[idx];
289 * iter.bi_bvec_done records actual length of the last bvec
290 * if this bio ends in the middle of one io vector
292 if (iter.bi_bvec_done)
293 bv->bv_len = iter.bi_bvec_done;
297 BIP_BLOCK_INTEGRITY = 1 << 0, /* block layer owns integrity data */
298 BIP_MAPPED_INTEGRITY = 1 << 1, /* ref tag has been remapped */
299 BIP_CTRL_NOCHECK = 1 << 2, /* disable HBA integrity checking */
300 BIP_DISK_NOCHECK = 1 << 3, /* disable disk integrity checking */
301 BIP_IP_CHECKSUM = 1 << 4, /* IP checksum */
305 * bio integrity payload
307 struct bio_integrity_payload {
308 struct bio *bip_bio; /* parent bio */
310 struct bvec_iter bip_iter;
312 bio_end_io_t *bip_end_io; /* saved I/O completion fn */
314 unsigned short bip_slab; /* slab the bip came from */
315 unsigned short bip_vcnt; /* # of integrity bio_vecs */
316 unsigned short bip_max_vcnt; /* integrity bio_vec slots */
317 unsigned short bip_flags; /* control flags */
319 struct work_struct bip_work; /* I/O completion */
321 struct bio_vec *bip_vec;
322 struct bio_vec bip_inline_vecs[0];/* embedded bvec array */
325 #if defined(CONFIG_BLK_DEV_INTEGRITY)
327 static inline struct bio_integrity_payload *bio_integrity(struct bio *bio)
329 if (bio->bi_opf & REQ_INTEGRITY)
330 return bio->bi_integrity;
335 static inline bool bio_integrity_flagged(struct bio *bio, enum bip_flags flag)
337 struct bio_integrity_payload *bip = bio_integrity(bio);
340 return bip->bip_flags & flag;
345 static inline sector_t bip_get_seed(struct bio_integrity_payload *bip)
347 return bip->bip_iter.bi_sector;
350 static inline void bip_set_seed(struct bio_integrity_payload *bip,
353 bip->bip_iter.bi_sector = seed;
356 #endif /* CONFIG_BLK_DEV_INTEGRITY */
358 extern void bio_trim(struct bio *bio, int offset, int size);
359 extern struct bio *bio_split(struct bio *bio, int sectors,
360 gfp_t gfp, struct bio_set *bs);
363 * bio_next_split - get next @sectors from a bio, splitting if necessary
365 * @sectors: number of sectors to split from the front of @bio
367 * @bs: bio set to allocate from
369 * Returns a bio representing the next @sectors of @bio - if the bio is smaller
370 * than @sectors, returns the original bio unchanged.
372 static inline struct bio *bio_next_split(struct bio *bio, int sectors,
373 gfp_t gfp, struct bio_set *bs)
375 if (sectors >= bio_sectors(bio))
378 return bio_split(bio, sectors, gfp, bs);
381 extern struct bio_set *bioset_create(unsigned int, unsigned int);
382 extern struct bio_set *bioset_create_nobvec(unsigned int, unsigned int);
383 extern void bioset_free(struct bio_set *);
384 extern mempool_t *biovec_create_pool(int pool_entries);
386 extern struct bio *bio_alloc_bioset(gfp_t, 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 *);
391 extern struct bio *bio_clone_bioset(struct bio *, gfp_t, struct bio_set *bs);
392 extern struct bio *bio_clone_bioset_partial(struct bio *, gfp_t,
393 struct bio_set *, int, int);
395 extern struct bio_set *fs_bio_set;
397 static inline struct bio *bio_alloc(gfp_t gfp_mask, unsigned int nr_iovecs)
399 return bio_alloc_bioset(gfp_mask, nr_iovecs, fs_bio_set);
402 static inline struct bio *bio_clone(struct bio *bio, gfp_t gfp_mask)
404 return bio_clone_bioset(bio, gfp_mask, 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 static inline struct bio *bio_clone_kmalloc(struct bio *bio, gfp_t gfp_mask)
414 return bio_clone_bioset(bio, gfp_mask, NULL);
418 extern blk_qc_t submit_bio(struct bio *);
420 extern void bio_endio(struct bio *);
422 static inline void bio_io_error(struct bio *bio)
424 bio->bi_error = -EIO;
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_reset(struct bio *);
437 void bio_chain(struct bio *, struct bio *);
439 extern int bio_add_page(struct bio *, struct page *, unsigned int,unsigned int);
440 extern int bio_add_pc_page(struct request_queue *, struct bio *, struct page *,
441 unsigned int, unsigned int);
442 int bio_iov_iter_get_pages(struct bio *bio, struct iov_iter *iter);
444 extern struct bio *bio_map_user_iov(struct request_queue *,
445 const struct iov_iter *, gfp_t);
446 extern void bio_unmap_user(struct bio *);
447 extern struct bio *bio_map_kern(struct request_queue *, void *, unsigned int,
449 extern struct bio *bio_copy_kern(struct request_queue *, void *, unsigned int,
451 extern void bio_set_pages_dirty(struct bio *bio);
452 extern void bio_check_pages_dirty(struct bio *bio);
454 void generic_start_io_acct(int rw, unsigned long sectors,
455 struct hd_struct *part);
456 void generic_end_io_acct(int rw, struct hd_struct *part,
457 unsigned long start_time);
459 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
460 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
462 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
463 extern void bio_flush_dcache_pages(struct bio *bi);
465 static inline void bio_flush_dcache_pages(struct bio *bi)
470 extern void bio_copy_data(struct bio *dst, struct bio *src);
471 extern int bio_alloc_pages(struct bio *bio, gfp_t gfp);
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 *,
476 const struct iov_iter *,
478 extern int bio_uncopy_user(struct bio *);
479 void zero_fill_bio(struct bio *bio);
480 extern struct bio_vec *bvec_alloc(gfp_t, int, unsigned long *, mempool_t *);
481 extern void bvec_free(mempool_t *, struct bio_vec *, unsigned int);
482 extern unsigned int bvec_nr_vecs(unsigned short idx);
484 #ifdef CONFIG_BLK_CGROUP
485 int bio_associate_blkcg(struct bio *bio, struct cgroup_subsys_state *blkcg_css);
486 int bio_associate_current(struct bio *bio);
487 void bio_disassociate_task(struct bio *bio);
488 void bio_clone_blkcg_association(struct bio *dst, struct bio *src);
489 #else /* CONFIG_BLK_CGROUP */
490 static inline int bio_associate_blkcg(struct bio *bio,
491 struct cgroup_subsys_state *blkcg_css) { return 0; }
492 static inline int bio_associate_current(struct bio *bio) { return -ENOENT; }
493 static inline void bio_disassociate_task(struct bio *bio) { }
494 static inline void bio_clone_blkcg_association(struct bio *dst,
496 #endif /* CONFIG_BLK_CGROUP */
498 #ifdef CONFIG_HIGHMEM
500 * remember never ever reenable interrupts between a bvec_kmap_irq and
503 static inline char *bvec_kmap_irq(struct bio_vec *bvec, unsigned long *flags)
508 * might not be a highmem page, but the preempt/irq count
509 * balancing is a lot nicer this way
511 local_irq_save(*flags);
512 addr = (unsigned long) kmap_atomic(bvec->bv_page);
514 BUG_ON(addr & ~PAGE_MASK);
516 return (char *) addr + bvec->bv_offset;
519 static inline void bvec_kunmap_irq(char *buffer, unsigned long *flags)
521 unsigned long ptr = (unsigned long) buffer & PAGE_MASK;
523 kunmap_atomic((void *) ptr);
524 local_irq_restore(*flags);
528 static inline char *bvec_kmap_irq(struct bio_vec *bvec, unsigned long *flags)
530 return page_address(bvec->bv_page) + bvec->bv_offset;
533 static inline void bvec_kunmap_irq(char *buffer, unsigned long *flags)
539 static inline char *__bio_kmap_irq(struct bio *bio, struct bvec_iter iter,
540 unsigned long *flags)
542 return bvec_kmap_irq(&bio_iter_iovec(bio, iter), flags);
544 #define __bio_kunmap_irq(buf, flags) bvec_kunmap_irq(buf, flags)
546 #define bio_kmap_irq(bio, flags) \
547 __bio_kmap_irq((bio), (bio)->bi_iter, (flags))
548 #define bio_kunmap_irq(buf,flags) __bio_kunmap_irq(buf, flags)
551 * BIO list management for use by remapping drivers (e.g. DM or MD) and loop.
553 * A bio_list anchors a singly-linked list of bios chained through the bi_next
554 * member of the bio. The bio_list also caches the last list member to allow
555 * fast access to the tail.
562 static inline int bio_list_empty(const struct bio_list *bl)
564 return bl->head == NULL;
567 static inline void bio_list_init(struct bio_list *bl)
569 bl->head = bl->tail = NULL;
572 #define BIO_EMPTY_LIST { NULL, NULL }
574 #define bio_list_for_each(bio, bl) \
575 for (bio = (bl)->head; bio; bio = bio->bi_next)
577 static inline unsigned bio_list_size(const struct bio_list *bl)
582 bio_list_for_each(bio, bl)
588 static inline void bio_list_add(struct bio_list *bl, struct bio *bio)
593 bl->tail->bi_next = bio;
600 static inline void bio_list_add_head(struct bio_list *bl, struct bio *bio)
602 bio->bi_next = bl->head;
610 static inline void bio_list_merge(struct bio_list *bl, struct bio_list *bl2)
616 bl->tail->bi_next = bl2->head;
618 bl->head = bl2->head;
620 bl->tail = bl2->tail;
623 static inline void bio_list_merge_head(struct bio_list *bl,
624 struct bio_list *bl2)
630 bl2->tail->bi_next = bl->head;
632 bl->tail = bl2->tail;
634 bl->head = bl2->head;
637 static inline struct bio *bio_list_peek(struct bio_list *bl)
642 static inline struct bio *bio_list_pop(struct bio_list *bl)
644 struct bio *bio = bl->head;
647 bl->head = bl->head->bi_next;
657 static inline struct bio *bio_list_get(struct bio_list *bl)
659 struct bio *bio = bl->head;
661 bl->head = bl->tail = NULL;
667 * Increment chain count for the bio. Make sure the CHAIN flag update
668 * is visible before the raised count.
670 static inline void bio_inc_remaining(struct bio *bio)
672 bio_set_flag(bio, BIO_CHAIN);
673 smp_mb__before_atomic();
674 atomic_inc(&bio->__bi_remaining);
678 * bio_set is used to allow other portions of the IO system to
679 * allocate their own private memory pools for bio and iovec structures.
680 * These memory pools in turn all allocate from the bio_slab
681 * and the bvec_slabs[].
683 #define BIO_POOL_SIZE 2
686 struct kmem_cache *bio_slab;
687 unsigned int front_pad;
690 mempool_t *bvec_pool;
691 #if defined(CONFIG_BLK_DEV_INTEGRITY)
692 mempool_t *bio_integrity_pool;
693 mempool_t *bvec_integrity_pool;
697 * Deadlock avoidance for stacking block drivers: see comments in
698 * bio_alloc_bioset() for details
700 spinlock_t rescue_lock;
701 struct bio_list rescue_list;
702 struct work_struct rescue_work;
703 struct workqueue_struct *rescue_workqueue;
709 struct kmem_cache *slab;
713 * a small number of entries is fine, not going to be performance critical.
714 * basically we just need to survive
716 #define BIO_SPLIT_ENTRIES 2
718 #if defined(CONFIG_BLK_DEV_INTEGRITY)
720 #define bip_for_each_vec(bvl, bip, iter) \
721 for_each_bvec(bvl, (bip)->bip_vec, iter, (bip)->bip_iter)
723 #define bio_for_each_integrity_vec(_bvl, _bio, _iter) \
725 bip_for_each_vec(_bvl, _bio->bi_integrity, _iter)
727 extern struct bio_integrity_payload *bio_integrity_alloc(struct bio *, gfp_t, unsigned int);
728 extern void bio_integrity_free(struct bio *);
729 extern int bio_integrity_add_page(struct bio *, struct page *, unsigned int, unsigned int);
730 extern bool bio_integrity_enabled(struct bio *bio);
731 extern int bio_integrity_prep(struct bio *);
732 extern void bio_integrity_endio(struct bio *);
733 extern void bio_integrity_advance(struct bio *, unsigned int);
734 extern void bio_integrity_trim(struct bio *, unsigned int, unsigned int);
735 extern int bio_integrity_clone(struct bio *, struct bio *, gfp_t);
736 extern int bioset_integrity_create(struct bio_set *, int);
737 extern void bioset_integrity_free(struct bio_set *);
738 extern void bio_integrity_init(void);
740 #else /* CONFIG_BLK_DEV_INTEGRITY */
742 static inline void *bio_integrity(struct bio *bio)
747 static inline bool bio_integrity_enabled(struct bio *bio)
752 static inline int bioset_integrity_create(struct bio_set *bs, int pool_size)
757 static inline void bioset_integrity_free (struct bio_set *bs)
762 static inline int bio_integrity_prep(struct bio *bio)
767 static inline void bio_integrity_free(struct bio *bio)
772 static inline int bio_integrity_clone(struct bio *bio, struct bio *bio_src,
778 static inline void bio_integrity_advance(struct bio *bio,
779 unsigned int bytes_done)
784 static inline void bio_integrity_trim(struct bio *bio, unsigned int offset,
785 unsigned int sectors)
790 static inline void bio_integrity_init(void)
795 static inline bool bio_integrity_flagged(struct bio *bio, enum bip_flags flag)
800 static inline void *bio_integrity_alloc(struct bio * bio, gfp_t gfp,
803 return ERR_PTR(-EINVAL);
806 static inline int bio_integrity_add_page(struct bio *bio, struct page *page,
807 unsigned int len, unsigned int offset)
812 #endif /* CONFIG_BLK_DEV_INTEGRITY */
814 #endif /* CONFIG_BLOCK */
815 #endif /* __LINUX_BIO_H */