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)
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)
215 * get a reference to a bio, so it won't disappear. the intended use is
219 * submit_bio(rw, bio);
220 * if (bio->bi_flags ...)
224 * without the bio_get(), it could potentially complete I/O before submit_bio
225 * returns. and then bio would be freed memory when if (bio->bi_flags ...)
228 static inline void bio_get(struct bio *bio)
230 bio->bi_flags |= (1 << BIO_REFFED);
231 smp_mb__before_atomic();
232 atomic_inc(&bio->__bi_cnt);
235 static inline void bio_cnt_set(struct bio *bio, unsigned int count)
238 bio->bi_flags |= (1 << BIO_REFFED);
239 smp_mb__before_atomic();
241 atomic_set(&bio->__bi_cnt, count);
244 static inline bool bio_flagged(struct bio *bio, unsigned int bit)
246 return (bio->bi_flags & (1U << bit)) != 0;
249 static inline void bio_set_flag(struct bio *bio, unsigned int bit)
251 bio->bi_flags |= (1U << bit);
254 static inline void bio_clear_flag(struct bio *bio, unsigned int bit)
256 bio->bi_flags &= ~(1U << bit);
259 static inline void bio_get_first_bvec(struct bio *bio, struct bio_vec *bv)
261 *bv = bio_iovec(bio);
264 static inline void bio_get_last_bvec(struct bio *bio, struct bio_vec *bv)
266 struct bvec_iter iter = bio->bi_iter;
269 if (unlikely(!bio_multiple_segments(bio))) {
270 *bv = bio_iovec(bio);
274 bio_advance_iter(bio, &iter, iter.bi_size);
276 if (!iter.bi_bvec_done)
277 idx = iter.bi_idx - 1;
278 else /* in the middle of bvec */
281 *bv = bio->bi_io_vec[idx];
284 * iter.bi_bvec_done records actual length of the last bvec
285 * if this bio ends in the middle of one io vector
287 if (iter.bi_bvec_done)
288 bv->bv_len = iter.bi_bvec_done;
292 BIP_BLOCK_INTEGRITY = 1 << 0, /* block layer owns integrity data */
293 BIP_MAPPED_INTEGRITY = 1 << 1, /* ref tag has been remapped */
294 BIP_CTRL_NOCHECK = 1 << 2, /* disable HBA integrity checking */
295 BIP_DISK_NOCHECK = 1 << 3, /* disable disk integrity checking */
296 BIP_IP_CHECKSUM = 1 << 4, /* IP checksum */
300 * bio integrity payload
302 struct bio_integrity_payload {
303 struct bio *bip_bio; /* parent bio */
305 struct bvec_iter bip_iter;
307 bio_end_io_t *bip_end_io; /* saved I/O completion fn */
309 unsigned short bip_slab; /* slab the bip came from */
310 unsigned short bip_vcnt; /* # of integrity bio_vecs */
311 unsigned short bip_max_vcnt; /* integrity bio_vec slots */
312 unsigned short bip_flags; /* control flags */
314 struct work_struct bip_work; /* I/O completion */
316 struct bio_vec *bip_vec;
317 struct bio_vec bip_inline_vecs[0];/* embedded bvec array */
320 #if defined(CONFIG_BLK_DEV_INTEGRITY)
322 static inline struct bio_integrity_payload *bio_integrity(struct bio *bio)
324 if (bio->bi_opf & REQ_INTEGRITY)
325 return bio->bi_integrity;
330 static inline bool bio_integrity_flagged(struct bio *bio, enum bip_flags flag)
332 struct bio_integrity_payload *bip = bio_integrity(bio);
335 return bip->bip_flags & flag;
340 static inline sector_t bip_get_seed(struct bio_integrity_payload *bip)
342 return bip->bip_iter.bi_sector;
345 static inline void bip_set_seed(struct bio_integrity_payload *bip,
348 bip->bip_iter.bi_sector = seed;
351 #endif /* CONFIG_BLK_DEV_INTEGRITY */
353 extern void bio_trim(struct bio *bio, int offset, int size);
354 extern struct bio *bio_split(struct bio *bio, int sectors,
355 gfp_t gfp, struct bio_set *bs);
358 * bio_next_split - get next @sectors from a bio, splitting if necessary
360 * @sectors: number of sectors to split from the front of @bio
362 * @bs: bio set to allocate from
364 * Returns a bio representing the next @sectors of @bio - if the bio is smaller
365 * than @sectors, returns the original bio unchanged.
367 static inline struct bio *bio_next_split(struct bio *bio, int sectors,
368 gfp_t gfp, struct bio_set *bs)
370 if (sectors >= bio_sectors(bio))
373 return bio_split(bio, sectors, gfp, bs);
376 extern struct bio_set *bioset_create(unsigned int, unsigned int);
377 extern struct bio_set *bioset_create_nobvec(unsigned int, unsigned int);
378 extern void bioset_free(struct bio_set *);
379 extern mempool_t *biovec_create_pool(int pool_entries);
381 extern struct bio *bio_alloc_bioset(gfp_t, unsigned int, struct bio_set *);
382 extern void bio_put(struct bio *);
384 extern void __bio_clone_fast(struct bio *, struct bio *);
385 extern struct bio *bio_clone_fast(struct bio *, gfp_t, struct bio_set *);
386 extern struct bio *bio_clone_bioset(struct bio *, gfp_t, struct bio_set *bs);
388 extern struct bio_set *fs_bio_set;
390 static inline struct bio *bio_alloc(gfp_t gfp_mask, unsigned int nr_iovecs)
392 return bio_alloc_bioset(gfp_mask, nr_iovecs, fs_bio_set);
395 static inline struct bio *bio_clone(struct bio *bio, gfp_t gfp_mask)
397 return bio_clone_bioset(bio, gfp_mask, fs_bio_set);
400 static inline struct bio *bio_kmalloc(gfp_t gfp_mask, unsigned int nr_iovecs)
402 return bio_alloc_bioset(gfp_mask, nr_iovecs, NULL);
405 static inline struct bio *bio_clone_kmalloc(struct bio *bio, gfp_t gfp_mask)
407 return bio_clone_bioset(bio, gfp_mask, NULL);
411 extern blk_qc_t submit_bio(struct bio *);
413 extern void bio_endio(struct bio *);
415 static inline void bio_io_error(struct bio *bio)
417 bio->bi_error = -EIO;
421 struct request_queue;
422 extern int bio_phys_segments(struct request_queue *, struct bio *);
424 extern int submit_bio_wait(struct bio *bio);
425 extern void bio_advance(struct bio *, unsigned);
427 extern void bio_init(struct bio *bio, struct bio_vec *table,
428 unsigned short max_vecs);
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 int bio_iov_iter_get_pages(struct bio *bio, struct iov_iter *iter);
437 extern struct bio *bio_map_user_iov(struct request_queue *,
438 const struct iov_iter *, gfp_t);
439 extern void bio_unmap_user(struct bio *);
440 extern struct bio *bio_map_kern(struct request_queue *, void *, unsigned int,
442 extern struct bio *bio_copy_kern(struct request_queue *, void *, unsigned int,
444 extern void bio_set_pages_dirty(struct bio *bio);
445 extern void bio_check_pages_dirty(struct bio *bio);
447 void generic_start_io_acct(int rw, unsigned long sectors,
448 struct hd_struct *part);
449 void generic_end_io_acct(int rw, struct hd_struct *part,
450 unsigned long start_time);
452 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
453 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
455 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
456 extern void bio_flush_dcache_pages(struct bio *bi);
458 static inline void bio_flush_dcache_pages(struct bio *bi)
463 extern void bio_copy_data(struct bio *dst, struct bio *src);
464 extern int bio_alloc_pages(struct bio *bio, gfp_t gfp);
465 extern void bio_free_pages(struct bio *bio);
467 extern struct bio *bio_copy_user_iov(struct request_queue *,
468 struct rq_map_data *,
469 const struct iov_iter *,
471 extern int bio_uncopy_user(struct bio *);
472 void zero_fill_bio(struct bio *bio);
473 extern struct bio_vec *bvec_alloc(gfp_t, int, unsigned long *, mempool_t *);
474 extern void bvec_free(mempool_t *, struct bio_vec *, unsigned int);
475 extern unsigned int bvec_nr_vecs(unsigned short idx);
477 #ifdef CONFIG_BLK_CGROUP
478 int bio_associate_blkcg(struct bio *bio, struct cgroup_subsys_state *blkcg_css);
479 int bio_associate_current(struct bio *bio);
480 void bio_disassociate_task(struct bio *bio);
481 void bio_clone_blkcg_association(struct bio *dst, struct bio *src);
482 #else /* CONFIG_BLK_CGROUP */
483 static inline int bio_associate_blkcg(struct bio *bio,
484 struct cgroup_subsys_state *blkcg_css) { return 0; }
485 static inline int bio_associate_current(struct bio *bio) { return -ENOENT; }
486 static inline void bio_disassociate_task(struct bio *bio) { }
487 static inline void bio_clone_blkcg_association(struct bio *dst,
489 #endif /* CONFIG_BLK_CGROUP */
491 #ifdef CONFIG_HIGHMEM
493 * remember never ever reenable interrupts between a bvec_kmap_irq and
496 static inline char *bvec_kmap_irq(struct bio_vec *bvec, unsigned long *flags)
501 * might not be a highmem page, but the preempt/irq count
502 * balancing is a lot nicer this way
504 local_irq_save(*flags);
505 addr = (unsigned long) kmap_atomic(bvec->bv_page);
507 BUG_ON(addr & ~PAGE_MASK);
509 return (char *) addr + bvec->bv_offset;
512 static inline void bvec_kunmap_irq(char *buffer, unsigned long *flags)
514 unsigned long ptr = (unsigned long) buffer & PAGE_MASK;
516 kunmap_atomic((void *) ptr);
517 local_irq_restore(*flags);
521 static inline char *bvec_kmap_irq(struct bio_vec *bvec, unsigned long *flags)
523 return page_address(bvec->bv_page) + bvec->bv_offset;
526 static inline void bvec_kunmap_irq(char *buffer, unsigned long *flags)
532 static inline char *__bio_kmap_irq(struct bio *bio, struct bvec_iter iter,
533 unsigned long *flags)
535 return bvec_kmap_irq(&bio_iter_iovec(bio, iter), flags);
537 #define __bio_kunmap_irq(buf, flags) bvec_kunmap_irq(buf, flags)
539 #define bio_kmap_irq(bio, flags) \
540 __bio_kmap_irq((bio), (bio)->bi_iter, (flags))
541 #define bio_kunmap_irq(buf,flags) __bio_kunmap_irq(buf, flags)
544 * BIO list management for use by remapping drivers (e.g. DM or MD) and loop.
546 * A bio_list anchors a singly-linked list of bios chained through the bi_next
547 * member of the bio. The bio_list also caches the last list member to allow
548 * fast access to the tail.
555 static inline int bio_list_empty(const struct bio_list *bl)
557 return bl->head == NULL;
560 static inline void bio_list_init(struct bio_list *bl)
562 bl->head = bl->tail = NULL;
565 #define BIO_EMPTY_LIST { NULL, NULL }
567 #define bio_list_for_each(bio, bl) \
568 for (bio = (bl)->head; bio; bio = bio->bi_next)
570 static inline unsigned bio_list_size(const struct bio_list *bl)
575 bio_list_for_each(bio, bl)
581 static inline void bio_list_add(struct bio_list *bl, struct bio *bio)
586 bl->tail->bi_next = bio;
593 static inline void bio_list_add_head(struct bio_list *bl, struct bio *bio)
595 bio->bi_next = bl->head;
603 static inline void bio_list_merge(struct bio_list *bl, struct bio_list *bl2)
609 bl->tail->bi_next = bl2->head;
611 bl->head = bl2->head;
613 bl->tail = bl2->tail;
616 static inline void bio_list_merge_head(struct bio_list *bl,
617 struct bio_list *bl2)
623 bl2->tail->bi_next = bl->head;
625 bl->tail = bl2->tail;
627 bl->head = bl2->head;
630 static inline struct bio *bio_list_peek(struct bio_list *bl)
635 static inline struct bio *bio_list_pop(struct bio_list *bl)
637 struct bio *bio = bl->head;
640 bl->head = bl->head->bi_next;
650 static inline struct bio *bio_list_get(struct bio_list *bl)
652 struct bio *bio = bl->head;
654 bl->head = bl->tail = NULL;
660 * Increment chain count for the bio. Make sure the CHAIN flag update
661 * is visible before the raised count.
663 static inline void bio_inc_remaining(struct bio *bio)
665 bio_set_flag(bio, BIO_CHAIN);
666 smp_mb__before_atomic();
667 atomic_inc(&bio->__bi_remaining);
671 * bio_set is used to allow other portions of the IO system to
672 * allocate their own private memory pools for bio and iovec structures.
673 * These memory pools in turn all allocate from the bio_slab
674 * and the bvec_slabs[].
676 #define BIO_POOL_SIZE 2
679 struct kmem_cache *bio_slab;
680 unsigned int front_pad;
683 mempool_t *bvec_pool;
684 #if defined(CONFIG_BLK_DEV_INTEGRITY)
685 mempool_t *bio_integrity_pool;
686 mempool_t *bvec_integrity_pool;
690 * Deadlock avoidance for stacking block drivers: see comments in
691 * bio_alloc_bioset() for details
693 spinlock_t rescue_lock;
694 struct bio_list rescue_list;
695 struct work_struct rescue_work;
696 struct workqueue_struct *rescue_workqueue;
702 struct kmem_cache *slab;
706 * a small number of entries is fine, not going to be performance critical.
707 * basically we just need to survive
709 #define BIO_SPLIT_ENTRIES 2
711 #if defined(CONFIG_BLK_DEV_INTEGRITY)
713 #define bip_for_each_vec(bvl, bip, iter) \
714 for_each_bvec(bvl, (bip)->bip_vec, iter, (bip)->bip_iter)
716 #define bio_for_each_integrity_vec(_bvl, _bio, _iter) \
718 bip_for_each_vec(_bvl, _bio->bi_integrity, _iter)
720 extern struct bio_integrity_payload *bio_integrity_alloc(struct bio *, gfp_t, unsigned int);
721 extern void bio_integrity_free(struct bio *);
722 extern int bio_integrity_add_page(struct bio *, struct page *, unsigned int, unsigned int);
723 extern bool bio_integrity_enabled(struct bio *bio);
724 extern int bio_integrity_prep(struct bio *);
725 extern void bio_integrity_endio(struct bio *);
726 extern void bio_integrity_advance(struct bio *, unsigned int);
727 extern void bio_integrity_trim(struct bio *, unsigned int, unsigned int);
728 extern int bio_integrity_clone(struct bio *, struct bio *, gfp_t);
729 extern int bioset_integrity_create(struct bio_set *, int);
730 extern void bioset_integrity_free(struct bio_set *);
731 extern void bio_integrity_init(void);
733 #else /* CONFIG_BLK_DEV_INTEGRITY */
735 static inline void *bio_integrity(struct bio *bio)
740 static inline bool bio_integrity_enabled(struct bio *bio)
745 static inline int bioset_integrity_create(struct bio_set *bs, int pool_size)
750 static inline void bioset_integrity_free (struct bio_set *bs)
755 static inline int bio_integrity_prep(struct bio *bio)
760 static inline void bio_integrity_free(struct bio *bio)
765 static inline int bio_integrity_clone(struct bio *bio, struct bio *bio_src,
771 static inline void bio_integrity_advance(struct bio *bio,
772 unsigned int bytes_done)
777 static inline void bio_integrity_trim(struct bio *bio, unsigned int offset,
778 unsigned int sectors)
783 static inline void bio_integrity_init(void)
788 static inline bool bio_integrity_flagged(struct bio *bio, enum bip_flags flag)
793 static inline void *bio_integrity_alloc(struct bio * bio, gfp_t gfp,
796 return ERR_PTR(-EINVAL);
799 static inline int bio_integrity_add_page(struct bio *bio, struct page *page,
800 unsigned int len, unsigned int offset)
805 #endif /* CONFIG_BLK_DEV_INTEGRITY */
807 #endif /* CONFIG_BLOCK */
808 #endif /* __LINUX_BIO_H */