4 * Copyright (C) 2001 Jens Axboe <axboe@suse.de>
6 * This program is free software; you can redistribute it and/or modify
7 * it under the terms of the GNU General Public License version 2 as
8 * published by the Free Software Foundation.
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 * GNU General Public License for more details.
16 * You should have received a copy of the GNU General Public Licens
17 * along with this program; if not, write to the Free Software
18 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-
23 #include <linux/highmem.h>
24 #include <linux/mempool.h>
25 #include <linux/ioprio.h>
26 #include <linux/bug.h>
32 /* struct bio, bio_vec and BIO_* flags are defined in blk_types.h */
33 #include <linux/blk_types.h>
38 #define BIO_BUG_ON BUG_ON
43 #define BIO_MAX_PAGES 256
45 #define bio_prio(bio) (bio)->bi_ioprio
46 #define bio_set_prio(bio, prio) ((bio)->bi_ioprio = prio)
48 #define bio_iter_iovec(bio, iter) \
49 bvec_iter_bvec((bio)->bi_io_vec, (iter))
51 #define bio_iter_page(bio, iter) \
52 bvec_iter_page((bio)->bi_io_vec, (iter))
53 #define bio_iter_len(bio, iter) \
54 bvec_iter_len((bio)->bi_io_vec, (iter))
55 #define bio_iter_offset(bio, iter) \
56 bvec_iter_offset((bio)->bi_io_vec, (iter))
58 #define bio_page(bio) bio_iter_page((bio), (bio)->bi_iter)
59 #define bio_offset(bio) bio_iter_offset((bio), (bio)->bi_iter)
60 #define bio_iovec(bio) bio_iter_iovec((bio), (bio)->bi_iter)
62 #define bio_multiple_segments(bio) \
63 ((bio)->bi_iter.bi_size != bio_iovec(bio).bv_len)
64 #define bio_sectors(bio) ((bio)->bi_iter.bi_size >> 9)
65 #define bio_end_sector(bio) ((bio)->bi_iter.bi_sector + bio_sectors((bio)))
68 * Check whether this bio carries any data or not. A NULL bio is allowed.
70 static inline bool bio_has_data(struct bio *bio)
73 bio->bi_iter.bi_size &&
74 bio_op(bio) != REQ_OP_DISCARD)
80 static inline bool bio_no_advance_iter(struct bio *bio)
82 return bio_op(bio) == REQ_OP_DISCARD || bio_op(bio) == REQ_OP_WRITE_SAME;
85 static inline bool bio_is_rw(struct bio *bio)
87 if (!bio_has_data(bio))
90 if (bio_no_advance_iter(bio))
96 static inline bool bio_mergeable(struct bio *bio)
98 if (bio->bi_opf & REQ_NOMERGE_FLAGS)
104 static inline unsigned int bio_cur_bytes(struct bio *bio)
106 if (bio_has_data(bio))
107 return bio_iovec(bio).bv_len;
108 else /* dataless requests such as discard */
109 return bio->bi_iter.bi_size;
112 static inline void *bio_data(struct bio *bio)
114 if (bio_has_data(bio))
115 return page_address(bio_page(bio)) + bio_offset(bio);
123 #define bio_to_phys(bio) (page_to_phys(bio_page((bio))) + (unsigned long) bio_offset((bio)))
124 #define bvec_to_phys(bv) (page_to_phys((bv)->bv_page) + (unsigned long) (bv)->bv_offset)
127 * queues that have highmem support enabled may still need to revert to
128 * PIO transfers occasionally and thus map high pages temporarily. For
129 * permanent PIO fall back, user is probably better off disabling highmem
130 * I/O completely on that queue (see ide-dma for example)
132 #define __bio_kmap_atomic(bio, iter) \
133 (kmap_atomic(bio_iter_iovec((bio), (iter)).bv_page) + \
134 bio_iter_iovec((bio), (iter)).bv_offset)
136 #define __bio_kunmap_atomic(addr) kunmap_atomic(addr)
142 /* Default implementation of BIOVEC_PHYS_MERGEABLE */
143 #define __BIOVEC_PHYS_MERGEABLE(vec1, vec2) \
144 ((bvec_to_phys((vec1)) + (vec1)->bv_len) == bvec_to_phys((vec2)))
147 * allow arch override, for eg virtualized architectures (put in asm/io.h)
149 #ifndef BIOVEC_PHYS_MERGEABLE
150 #define BIOVEC_PHYS_MERGEABLE(vec1, vec2) \
151 __BIOVEC_PHYS_MERGEABLE(vec1, vec2)
154 #define __BIO_SEG_BOUNDARY(addr1, addr2, mask) \
155 (((addr1) | (mask)) == (((addr2) - 1) | (mask)))
156 #define BIOVEC_SEG_BOUNDARY(q, b1, b2) \
157 __BIO_SEG_BOUNDARY(bvec_to_phys((b1)), bvec_to_phys((b2)) + (b2)->bv_len, queue_segment_boundary((q)))
160 * drivers should _never_ use the all version - the bio may have been split
161 * before it got to the driver and the driver won't own all of it
163 #define bio_for_each_segment_all(bvl, bio, i) \
164 for (i = 0, bvl = (bio)->bi_io_vec; i < (bio)->bi_vcnt; i++, bvl++)
166 static inline void bio_advance_iter(struct bio *bio, struct bvec_iter *iter,
169 iter->bi_sector += bytes >> 9;
171 if (bio_no_advance_iter(bio))
172 iter->bi_size -= bytes;
174 bvec_iter_advance(bio->bi_io_vec, iter, bytes);
177 #define __bio_for_each_segment(bvl, bio, iter, start) \
178 for (iter = (start); \
180 ((bvl = bio_iter_iovec((bio), (iter))), 1); \
181 bio_advance_iter((bio), &(iter), (bvl).bv_len))
183 #define bio_for_each_segment(bvl, bio, iter) \
184 __bio_for_each_segment(bvl, bio, iter, (bio)->bi_iter)
186 #define bio_iter_last(bvec, iter) ((iter).bi_size == (bvec).bv_len)
188 static inline unsigned bio_segments(struct bio *bio)
192 struct bvec_iter iter;
195 * We special case discard/write same, because they interpret bi_size
199 if (bio_op(bio) == REQ_OP_DISCARD)
202 if (bio_op(bio) == REQ_OP_WRITE_SAME)
205 bio_for_each_segment(bv, bio, iter)
212 * get a reference to a bio, so it won't disappear. the intended use is
216 * submit_bio(rw, bio);
217 * if (bio->bi_flags ...)
221 * without the bio_get(), it could potentially complete I/O before submit_bio
222 * returns. and then bio would be freed memory when if (bio->bi_flags ...)
225 static inline void bio_get(struct bio *bio)
227 bio->bi_flags |= (1 << BIO_REFFED);
228 smp_mb__before_atomic();
229 atomic_inc(&bio->__bi_cnt);
232 static inline void bio_cnt_set(struct bio *bio, unsigned int count)
235 bio->bi_flags |= (1 << BIO_REFFED);
236 smp_mb__before_atomic();
238 atomic_set(&bio->__bi_cnt, count);
241 static inline bool bio_flagged(struct bio *bio, unsigned int bit)
243 return (bio->bi_flags & (1U << bit)) != 0;
246 static inline void bio_set_flag(struct bio *bio, unsigned int bit)
248 bio->bi_flags |= (1U << bit);
251 static inline void bio_clear_flag(struct bio *bio, unsigned int bit)
253 bio->bi_flags &= ~(1U << bit);
256 static inline void bio_get_first_bvec(struct bio *bio, struct bio_vec *bv)
258 *bv = bio_iovec(bio);
261 static inline void bio_get_last_bvec(struct bio *bio, struct bio_vec *bv)
263 struct bvec_iter iter = bio->bi_iter;
266 if (unlikely(!bio_multiple_segments(bio))) {
267 *bv = bio_iovec(bio);
271 bio_advance_iter(bio, &iter, iter.bi_size);
273 if (!iter.bi_bvec_done)
274 idx = iter.bi_idx - 1;
275 else /* in the middle of bvec */
278 *bv = bio->bi_io_vec[idx];
281 * iter.bi_bvec_done records actual length of the last bvec
282 * if this bio ends in the middle of one io vector
284 if (iter.bi_bvec_done)
285 bv->bv_len = iter.bi_bvec_done;
289 BIP_BLOCK_INTEGRITY = 1 << 0, /* block layer owns integrity data */
290 BIP_MAPPED_INTEGRITY = 1 << 1, /* ref tag has been remapped */
291 BIP_CTRL_NOCHECK = 1 << 2, /* disable HBA integrity checking */
292 BIP_DISK_NOCHECK = 1 << 3, /* disable disk integrity checking */
293 BIP_IP_CHECKSUM = 1 << 4, /* IP checksum */
297 * bio integrity payload
299 struct bio_integrity_payload {
300 struct bio *bip_bio; /* parent bio */
302 struct bvec_iter bip_iter;
304 bio_end_io_t *bip_end_io; /* saved I/O completion fn */
306 unsigned short bip_slab; /* slab the bip came from */
307 unsigned short bip_vcnt; /* # of integrity bio_vecs */
308 unsigned short bip_max_vcnt; /* integrity bio_vec slots */
309 unsigned short bip_flags; /* control flags */
311 struct work_struct bip_work; /* I/O completion */
313 struct bio_vec *bip_vec;
314 struct bio_vec bip_inline_vecs[0];/* embedded bvec array */
317 #if defined(CONFIG_BLK_DEV_INTEGRITY)
319 static inline struct bio_integrity_payload *bio_integrity(struct bio *bio)
321 if (bio->bi_opf & REQ_INTEGRITY)
322 return bio->bi_integrity;
327 static inline bool bio_integrity_flagged(struct bio *bio, enum bip_flags flag)
329 struct bio_integrity_payload *bip = bio_integrity(bio);
332 return bip->bip_flags & flag;
337 static inline sector_t bip_get_seed(struct bio_integrity_payload *bip)
339 return bip->bip_iter.bi_sector;
342 static inline void bip_set_seed(struct bio_integrity_payload *bip,
345 bip->bip_iter.bi_sector = seed;
348 #endif /* CONFIG_BLK_DEV_INTEGRITY */
350 extern void bio_trim(struct bio *bio, int offset, int size);
351 extern struct bio *bio_split(struct bio *bio, int sectors,
352 gfp_t gfp, struct bio_set *bs);
355 * bio_next_split - get next @sectors from a bio, splitting if necessary
357 * @sectors: number of sectors to split from the front of @bio
359 * @bs: bio set to allocate from
361 * Returns a bio representing the next @sectors of @bio - if the bio is smaller
362 * than @sectors, returns the original bio unchanged.
364 static inline struct bio *bio_next_split(struct bio *bio, int sectors,
365 gfp_t gfp, struct bio_set *bs)
367 if (sectors >= bio_sectors(bio))
370 return bio_split(bio, sectors, gfp, bs);
373 extern struct bio_set *bioset_create(unsigned int, unsigned int);
374 extern struct bio_set *bioset_create_nobvec(unsigned int, unsigned int);
375 extern void bioset_free(struct bio_set *);
376 extern mempool_t *biovec_create_pool(int pool_entries);
378 extern struct bio *bio_alloc_bioset(gfp_t, int, struct bio_set *);
379 extern void bio_put(struct bio *);
381 extern void __bio_clone_fast(struct bio *, struct bio *);
382 extern struct bio *bio_clone_fast(struct bio *, gfp_t, struct bio_set *);
383 extern struct bio *bio_clone_bioset(struct bio *, gfp_t, struct bio_set *bs);
385 extern struct bio_set *fs_bio_set;
387 static inline struct bio *bio_alloc(gfp_t gfp_mask, unsigned int nr_iovecs)
389 return bio_alloc_bioset(gfp_mask, nr_iovecs, fs_bio_set);
392 static inline struct bio *bio_clone(struct bio *bio, gfp_t gfp_mask)
394 return bio_clone_bioset(bio, gfp_mask, fs_bio_set);
397 static inline struct bio *bio_kmalloc(gfp_t gfp_mask, unsigned int nr_iovecs)
399 return bio_alloc_bioset(gfp_mask, nr_iovecs, NULL);
402 static inline struct bio *bio_clone_kmalloc(struct bio *bio, gfp_t gfp_mask)
404 return bio_clone_bioset(bio, gfp_mask, NULL);
408 extern void bio_endio(struct bio *);
410 static inline void bio_io_error(struct bio *bio)
412 bio->bi_error = -EIO;
416 struct request_queue;
417 extern int bio_phys_segments(struct request_queue *, struct bio *);
419 extern int submit_bio_wait(struct bio *bio);
420 extern void bio_advance(struct bio *, unsigned);
422 extern void bio_init(struct bio *);
423 extern void bio_reset(struct bio *);
424 void bio_chain(struct bio *, struct bio *);
426 extern int bio_add_page(struct bio *, struct page *, unsigned int,unsigned int);
427 extern int bio_add_pc_page(struct request_queue *, struct bio *, struct page *,
428 unsigned int, unsigned int);
430 extern struct bio *bio_map_user_iov(struct request_queue *,
431 const struct iov_iter *, gfp_t);
432 extern void bio_unmap_user(struct bio *);
433 extern struct bio *bio_map_kern(struct request_queue *, void *, unsigned int,
435 extern struct bio *bio_copy_kern(struct request_queue *, void *, unsigned int,
437 extern void bio_set_pages_dirty(struct bio *bio);
438 extern void bio_check_pages_dirty(struct bio *bio);
440 void generic_start_io_acct(int rw, unsigned long sectors,
441 struct hd_struct *part);
442 void generic_end_io_acct(int rw, struct hd_struct *part,
443 unsigned long start_time);
445 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
446 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
448 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
449 extern void bio_flush_dcache_pages(struct bio *bi);
451 static inline void bio_flush_dcache_pages(struct bio *bi)
456 extern void bio_copy_data(struct bio *dst, struct bio *src);
457 extern int bio_alloc_pages(struct bio *bio, gfp_t gfp);
459 extern struct bio *bio_copy_user_iov(struct request_queue *,
460 struct rq_map_data *,
461 const struct iov_iter *,
463 extern int bio_uncopy_user(struct bio *);
464 void zero_fill_bio(struct bio *bio);
465 extern struct bio_vec *bvec_alloc(gfp_t, int, unsigned long *, mempool_t *);
466 extern void bvec_free(mempool_t *, struct bio_vec *, unsigned int);
467 extern unsigned int bvec_nr_vecs(unsigned short idx);
469 #ifdef CONFIG_BLK_CGROUP
470 int bio_associate_blkcg(struct bio *bio, struct cgroup_subsys_state *blkcg_css);
471 int bio_associate_current(struct bio *bio);
472 void bio_disassociate_task(struct bio *bio);
473 void bio_clone_blkcg_association(struct bio *dst, struct bio *src);
474 #else /* CONFIG_BLK_CGROUP */
475 static inline int bio_associate_blkcg(struct bio *bio,
476 struct cgroup_subsys_state *blkcg_css) { return 0; }
477 static inline int bio_associate_current(struct bio *bio) { return -ENOENT; }
478 static inline void bio_disassociate_task(struct bio *bio) { }
479 static inline void bio_clone_blkcg_association(struct bio *dst,
481 #endif /* CONFIG_BLK_CGROUP */
483 #ifdef CONFIG_HIGHMEM
485 * remember never ever reenable interrupts between a bvec_kmap_irq and
488 static inline char *bvec_kmap_irq(struct bio_vec *bvec, unsigned long *flags)
493 * might not be a highmem page, but the preempt/irq count
494 * balancing is a lot nicer this way
496 local_irq_save(*flags);
497 addr = (unsigned long) kmap_atomic(bvec->bv_page);
499 BUG_ON(addr & ~PAGE_MASK);
501 return (char *) addr + bvec->bv_offset;
504 static inline void bvec_kunmap_irq(char *buffer, unsigned long *flags)
506 unsigned long ptr = (unsigned long) buffer & PAGE_MASK;
508 kunmap_atomic((void *) ptr);
509 local_irq_restore(*flags);
513 static inline char *bvec_kmap_irq(struct bio_vec *bvec, unsigned long *flags)
515 return page_address(bvec->bv_page) + bvec->bv_offset;
518 static inline void bvec_kunmap_irq(char *buffer, unsigned long *flags)
524 static inline char *__bio_kmap_irq(struct bio *bio, struct bvec_iter iter,
525 unsigned long *flags)
527 return bvec_kmap_irq(&bio_iter_iovec(bio, iter), flags);
529 #define __bio_kunmap_irq(buf, flags) bvec_kunmap_irq(buf, flags)
531 #define bio_kmap_irq(bio, flags) \
532 __bio_kmap_irq((bio), (bio)->bi_iter, (flags))
533 #define bio_kunmap_irq(buf,flags) __bio_kunmap_irq(buf, flags)
536 * BIO list management for use by remapping drivers (e.g. DM or MD) and loop.
538 * A bio_list anchors a singly-linked list of bios chained through the bi_next
539 * member of the bio. The bio_list also caches the last list member to allow
540 * fast access to the tail.
547 static inline int bio_list_empty(const struct bio_list *bl)
549 return bl->head == NULL;
552 static inline void bio_list_init(struct bio_list *bl)
554 bl->head = bl->tail = NULL;
557 #define BIO_EMPTY_LIST { NULL, NULL }
559 #define bio_list_for_each(bio, bl) \
560 for (bio = (bl)->head; bio; bio = bio->bi_next)
562 static inline unsigned bio_list_size(const struct bio_list *bl)
567 bio_list_for_each(bio, bl)
573 static inline void bio_list_add(struct bio_list *bl, struct bio *bio)
578 bl->tail->bi_next = bio;
585 static inline void bio_list_add_head(struct bio_list *bl, struct bio *bio)
587 bio->bi_next = bl->head;
595 static inline void bio_list_merge(struct bio_list *bl, struct bio_list *bl2)
601 bl->tail->bi_next = bl2->head;
603 bl->head = bl2->head;
605 bl->tail = bl2->tail;
608 static inline void bio_list_merge_head(struct bio_list *bl,
609 struct bio_list *bl2)
615 bl2->tail->bi_next = bl->head;
617 bl->tail = bl2->tail;
619 bl->head = bl2->head;
622 static inline struct bio *bio_list_peek(struct bio_list *bl)
627 static inline struct bio *bio_list_pop(struct bio_list *bl)
629 struct bio *bio = bl->head;
632 bl->head = bl->head->bi_next;
642 static inline struct bio *bio_list_get(struct bio_list *bl)
644 struct bio *bio = bl->head;
646 bl->head = bl->tail = NULL;
652 * Increment chain count for the bio. Make sure the CHAIN flag update
653 * is visible before the raised count.
655 static inline void bio_inc_remaining(struct bio *bio)
657 bio_set_flag(bio, BIO_CHAIN);
658 smp_mb__before_atomic();
659 atomic_inc(&bio->__bi_remaining);
663 * bio_set is used to allow other portions of the IO system to
664 * allocate their own private memory pools for bio and iovec structures.
665 * These memory pools in turn all allocate from the bio_slab
666 * and the bvec_slabs[].
668 #define BIO_POOL_SIZE 2
671 struct kmem_cache *bio_slab;
672 unsigned int front_pad;
675 mempool_t *bvec_pool;
676 #if defined(CONFIG_BLK_DEV_INTEGRITY)
677 mempool_t *bio_integrity_pool;
678 mempool_t *bvec_integrity_pool;
682 * Deadlock avoidance for stacking block drivers: see comments in
683 * bio_alloc_bioset() for details
685 spinlock_t rescue_lock;
686 struct bio_list rescue_list;
687 struct work_struct rescue_work;
688 struct workqueue_struct *rescue_workqueue;
694 struct kmem_cache *slab;
698 * a small number of entries is fine, not going to be performance critical.
699 * basically we just need to survive
701 #define BIO_SPLIT_ENTRIES 2
703 #if defined(CONFIG_BLK_DEV_INTEGRITY)
705 #define bip_for_each_vec(bvl, bip, iter) \
706 for_each_bvec(bvl, (bip)->bip_vec, iter, (bip)->bip_iter)
708 #define bio_for_each_integrity_vec(_bvl, _bio, _iter) \
710 bip_for_each_vec(_bvl, _bio->bi_integrity, _iter)
712 extern struct bio_integrity_payload *bio_integrity_alloc(struct bio *, gfp_t, unsigned int);
713 extern void bio_integrity_free(struct bio *);
714 extern int bio_integrity_add_page(struct bio *, struct page *, unsigned int, unsigned int);
715 extern bool bio_integrity_enabled(struct bio *bio);
716 extern int bio_integrity_prep(struct bio *);
717 extern void bio_integrity_endio(struct bio *);
718 extern void bio_integrity_advance(struct bio *, unsigned int);
719 extern void bio_integrity_trim(struct bio *, unsigned int, unsigned int);
720 extern int bio_integrity_clone(struct bio *, struct bio *, gfp_t);
721 extern int bioset_integrity_create(struct bio_set *, int);
722 extern void bioset_integrity_free(struct bio_set *);
723 extern void bio_integrity_init(void);
725 #else /* CONFIG_BLK_DEV_INTEGRITY */
727 static inline void *bio_integrity(struct bio *bio)
732 static inline bool bio_integrity_enabled(struct bio *bio)
737 static inline int bioset_integrity_create(struct bio_set *bs, int pool_size)
742 static inline void bioset_integrity_free (struct bio_set *bs)
747 static inline int bio_integrity_prep(struct bio *bio)
752 static inline void bio_integrity_free(struct bio *bio)
757 static inline int bio_integrity_clone(struct bio *bio, struct bio *bio_src,
763 static inline void bio_integrity_advance(struct bio *bio,
764 unsigned int bytes_done)
769 static inline void bio_integrity_trim(struct bio *bio, unsigned int offset,
770 unsigned int sectors)
775 static inline void bio_integrity_init(void)
780 static inline bool bio_integrity_flagged(struct bio *bio, enum bip_flags flag)
785 static inline void *bio_integrity_alloc(struct bio * bio, gfp_t gfp,
788 return ERR_PTR(-EINVAL);
791 static inline int bio_integrity_add_page(struct bio *bio, struct page *page,
792 unsigned int len, unsigned int offset)
797 #endif /* CONFIG_BLK_DEV_INTEGRITY */
799 #endif /* CONFIG_BLOCK */
800 #endif /* __LINUX_BIO_H */