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 bvec_to_phys(bv) (page_to_phys((bv)->bv_page) + (unsigned long) (bv)->bv_offset)
124 * queues that have highmem support enabled may still need to revert to
125 * PIO transfers occasionally and thus map high pages temporarily. For
126 * permanent PIO fall back, user is probably better off disabling highmem
127 * I/O completely on that queue (see ide-dma for example)
129 #define __bio_kmap_atomic(bio, iter) \
130 (kmap_atomic(bio_iter_iovec((bio), (iter)).bv_page) + \
131 bio_iter_iovec((bio), (iter)).bv_offset)
133 #define __bio_kunmap_atomic(addr) kunmap_atomic(addr)
139 /* Default implementation of BIOVEC_PHYS_MERGEABLE */
140 #define __BIOVEC_PHYS_MERGEABLE(vec1, vec2) \
141 ((bvec_to_phys((vec1)) + (vec1)->bv_len) == bvec_to_phys((vec2)))
144 * allow arch override, for eg virtualized architectures (put in asm/io.h)
146 #ifndef BIOVEC_PHYS_MERGEABLE
147 #define BIOVEC_PHYS_MERGEABLE(vec1, vec2) \
148 __BIOVEC_PHYS_MERGEABLE(vec1, vec2)
151 #define __BIO_SEG_BOUNDARY(addr1, addr2, mask) \
152 (((addr1) | (mask)) == (((addr2) - 1) | (mask)))
153 #define BIOVEC_SEG_BOUNDARY(q, b1, b2) \
154 __BIO_SEG_BOUNDARY(bvec_to_phys((b1)), bvec_to_phys((b2)) + (b2)->bv_len, queue_segment_boundary((q)))
157 * drivers should _never_ use the all version - the bio may have been split
158 * before it got to the driver and the driver won't own all of it
160 #define bio_for_each_segment_all(bvl, bio, i) \
161 for (i = 0, bvl = (bio)->bi_io_vec; i < (bio)->bi_vcnt; i++, bvl++)
163 static inline void bio_advance_iter(struct bio *bio, struct bvec_iter *iter,
166 iter->bi_sector += bytes >> 9;
168 if (bio_no_advance_iter(bio)) {
169 iter->bi_size -= bytes;
170 iter->bi_done += bytes;
172 bvec_iter_advance(bio->bi_io_vec, iter, bytes);
173 /* TODO: It is reasonable to complete bio with error here. */
177 static inline bool bio_rewind_iter(struct bio *bio, struct bvec_iter *iter,
180 iter->bi_sector -= bytes >> 9;
182 if (bio_no_advance_iter(bio)) {
183 iter->bi_size += bytes;
184 iter->bi_done -= bytes;
188 return bvec_iter_rewind(bio->bi_io_vec, iter, bytes);
191 #define __bio_for_each_segment(bvl, bio, iter, start) \
192 for (iter = (start); \
194 ((bvl = bio_iter_iovec((bio), (iter))), 1); \
195 bio_advance_iter((bio), &(iter), (bvl).bv_len))
197 #define bio_for_each_segment(bvl, bio, iter) \
198 __bio_for_each_segment(bvl, bio, iter, (bio)->bi_iter)
200 #define bio_iter_last(bvec, iter) ((iter).bi_size == (bvec).bv_len)
202 static inline unsigned bio_segments(struct bio *bio)
206 struct bvec_iter iter;
209 * We special case discard/write same/write zeroes, because they
210 * interpret bi_size differently:
213 switch (bio_op(bio)) {
215 case REQ_OP_SECURE_ERASE:
216 case REQ_OP_WRITE_ZEROES:
218 case REQ_OP_WRITE_SAME:
224 bio_for_each_segment(bv, bio, iter)
231 * get a reference to a bio, so it won't disappear. the intended use is
235 * submit_bio(rw, bio);
236 * if (bio->bi_flags ...)
240 * without the bio_get(), it could potentially complete I/O before submit_bio
241 * returns. and then bio would be freed memory when if (bio->bi_flags ...)
244 static inline void bio_get(struct bio *bio)
246 bio->bi_flags |= (1 << BIO_REFFED);
247 smp_mb__before_atomic();
248 atomic_inc(&bio->__bi_cnt);
251 static inline void bio_cnt_set(struct bio *bio, unsigned int count)
254 bio->bi_flags |= (1 << BIO_REFFED);
255 smp_mb__before_atomic();
257 atomic_set(&bio->__bi_cnt, count);
260 static inline bool bio_flagged(struct bio *bio, unsigned int bit)
262 return (bio->bi_flags & (1U << bit)) != 0;
265 static inline void bio_set_flag(struct bio *bio, unsigned int bit)
267 bio->bi_flags |= (1U << bit);
270 static inline void bio_clear_flag(struct bio *bio, unsigned int bit)
272 bio->bi_flags &= ~(1U << bit);
275 static inline void bio_get_first_bvec(struct bio *bio, struct bio_vec *bv)
277 *bv = bio_iovec(bio);
280 static inline void bio_get_last_bvec(struct bio *bio, struct bio_vec *bv)
282 struct bvec_iter iter = bio->bi_iter;
285 if (unlikely(!bio_multiple_segments(bio))) {
286 *bv = bio_iovec(bio);
290 bio_advance_iter(bio, &iter, iter.bi_size);
292 if (!iter.bi_bvec_done)
293 idx = iter.bi_idx - 1;
294 else /* in the middle of bvec */
297 *bv = bio->bi_io_vec[idx];
300 * iter.bi_bvec_done records actual length of the last bvec
301 * if this bio ends in the middle of one io vector
303 if (iter.bi_bvec_done)
304 bv->bv_len = iter.bi_bvec_done;
308 BIP_BLOCK_INTEGRITY = 1 << 0, /* block layer owns integrity data */
309 BIP_MAPPED_INTEGRITY = 1 << 1, /* ref tag has been remapped */
310 BIP_CTRL_NOCHECK = 1 << 2, /* disable HBA integrity checking */
311 BIP_DISK_NOCHECK = 1 << 3, /* disable disk integrity checking */
312 BIP_IP_CHECKSUM = 1 << 4, /* IP checksum */
316 * bio integrity payload
318 struct bio_integrity_payload {
319 struct bio *bip_bio; /* parent bio */
321 struct bvec_iter bip_iter;
323 bio_end_io_t *bip_end_io; /* saved I/O completion fn */
325 unsigned short bip_slab; /* slab the bip came from */
326 unsigned short bip_vcnt; /* # of integrity bio_vecs */
327 unsigned short bip_max_vcnt; /* integrity bio_vec slots */
328 unsigned short bip_flags; /* control flags */
330 struct work_struct bip_work; /* I/O completion */
332 struct bio_vec *bip_vec;
333 struct bio_vec bip_inline_vecs[0];/* embedded bvec array */
336 #if defined(CONFIG_BLK_DEV_INTEGRITY)
338 static inline struct bio_integrity_payload *bio_integrity(struct bio *bio)
340 if (bio->bi_opf & REQ_INTEGRITY)
341 return bio->bi_integrity;
346 static inline bool bio_integrity_flagged(struct bio *bio, enum bip_flags flag)
348 struct bio_integrity_payload *bip = bio_integrity(bio);
351 return bip->bip_flags & flag;
356 static inline sector_t bip_get_seed(struct bio_integrity_payload *bip)
358 return bip->bip_iter.bi_sector;
361 static inline void bip_set_seed(struct bio_integrity_payload *bip,
364 bip->bip_iter.bi_sector = seed;
367 #endif /* CONFIG_BLK_DEV_INTEGRITY */
369 extern void bio_trim(struct bio *bio, int offset, int size);
370 extern struct bio *bio_split(struct bio *bio, int sectors,
371 gfp_t gfp, struct bio_set *bs);
374 * bio_next_split - get next @sectors from a bio, splitting if necessary
376 * @sectors: number of sectors to split from the front of @bio
378 * @bs: bio set to allocate from
380 * Returns a bio representing the next @sectors of @bio - if the bio is smaller
381 * than @sectors, returns the original bio unchanged.
383 static inline struct bio *bio_next_split(struct bio *bio, int sectors,
384 gfp_t gfp, struct bio_set *bs)
386 if (sectors >= bio_sectors(bio))
389 return bio_split(bio, sectors, gfp, bs);
392 extern struct bio_set *bioset_create(unsigned int, unsigned int, int flags);
394 BIOSET_NEED_BVECS = BIT(0),
395 BIOSET_NEED_RESCUER = BIT(1),
397 extern void bioset_free(struct bio_set *);
398 extern mempool_t *biovec_create_pool(int pool_entries);
400 extern struct bio *bio_alloc_bioset(gfp_t, unsigned int, struct bio_set *);
401 extern void bio_put(struct bio *);
403 extern void __bio_clone_fast(struct bio *, struct bio *);
404 extern struct bio *bio_clone_fast(struct bio *, gfp_t, struct bio_set *);
405 extern struct bio *bio_clone_bioset(struct bio *, gfp_t, struct bio_set *bs);
407 extern struct bio_set *fs_bio_set;
409 static inline struct bio *bio_alloc(gfp_t gfp_mask, unsigned int nr_iovecs)
411 return bio_alloc_bioset(gfp_mask, nr_iovecs, fs_bio_set);
414 static inline struct bio *bio_kmalloc(gfp_t gfp_mask, unsigned int nr_iovecs)
416 return bio_alloc_bioset(gfp_mask, nr_iovecs, NULL);
419 static inline struct bio *bio_clone_kmalloc(struct bio *bio, gfp_t gfp_mask)
421 return bio_clone_bioset(bio, gfp_mask, NULL);
425 extern blk_qc_t submit_bio(struct bio *);
427 extern void bio_endio(struct bio *);
429 static inline void bio_io_error(struct bio *bio)
431 bio->bi_status = BLK_STS_IOERR;
435 static inline void bio_wouldblock_error(struct bio *bio)
437 bio->bi_status = BLK_STS_AGAIN;
441 struct request_queue;
442 extern int bio_phys_segments(struct request_queue *, struct bio *);
444 extern int submit_bio_wait(struct bio *bio);
445 extern void bio_advance(struct bio *, unsigned);
447 extern void bio_init(struct bio *bio, struct bio_vec *table,
448 unsigned short max_vecs);
449 extern void bio_uninit(struct bio *);
450 extern void bio_reset(struct bio *);
451 void bio_chain(struct bio *, struct bio *);
453 extern int bio_add_page(struct bio *, struct page *, unsigned int,unsigned int);
454 extern int bio_add_pc_page(struct request_queue *, struct bio *, struct page *,
455 unsigned int, unsigned int);
456 int bio_iov_iter_get_pages(struct bio *bio, struct iov_iter *iter);
458 extern struct bio *bio_map_user_iov(struct request_queue *,
459 const struct iov_iter *, gfp_t);
460 extern void bio_unmap_user(struct bio *);
461 extern struct bio *bio_map_kern(struct request_queue *, void *, unsigned int,
463 extern struct bio *bio_copy_kern(struct request_queue *, void *, unsigned int,
465 extern void bio_set_pages_dirty(struct bio *bio);
466 extern void bio_check_pages_dirty(struct bio *bio);
468 void generic_start_io_acct(int rw, unsigned long sectors,
469 struct hd_struct *part);
470 void generic_end_io_acct(int rw, struct hd_struct *part,
471 unsigned long start_time);
473 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
474 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
476 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
477 extern void bio_flush_dcache_pages(struct bio *bi);
479 static inline void bio_flush_dcache_pages(struct bio *bi)
484 extern void bio_copy_data(struct bio *dst, struct bio *src);
485 extern int bio_alloc_pages(struct bio *bio, gfp_t gfp);
486 extern void bio_free_pages(struct bio *bio);
488 extern struct bio *bio_copy_user_iov(struct request_queue *,
489 struct rq_map_data *,
490 const struct iov_iter *,
492 extern int bio_uncopy_user(struct bio *);
493 void zero_fill_bio(struct bio *bio);
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);
498 #ifdef CONFIG_BLK_CGROUP
499 int bio_associate_blkcg(struct bio *bio, struct cgroup_subsys_state *blkcg_css);
500 int bio_associate_current(struct bio *bio);
501 void bio_disassociate_task(struct bio *bio);
502 void bio_clone_blkcg_association(struct bio *dst, struct bio *src);
503 #else /* CONFIG_BLK_CGROUP */
504 static inline int bio_associate_blkcg(struct bio *bio,
505 struct cgroup_subsys_state *blkcg_css) { return 0; }
506 static inline int bio_associate_current(struct bio *bio) { return -ENOENT; }
507 static inline void bio_disassociate_task(struct bio *bio) { }
508 static inline void bio_clone_blkcg_association(struct bio *dst,
510 #endif /* CONFIG_BLK_CGROUP */
512 #ifdef CONFIG_HIGHMEM
514 * remember never ever reenable interrupts between a bvec_kmap_irq and
517 static inline char *bvec_kmap_irq(struct bio_vec *bvec, unsigned long *flags)
522 * might not be a highmem page, but the preempt/irq count
523 * balancing is a lot nicer this way
525 local_irq_save(*flags);
526 addr = (unsigned long) kmap_atomic(bvec->bv_page);
528 BUG_ON(addr & ~PAGE_MASK);
530 return (char *) addr + bvec->bv_offset;
533 static inline void bvec_kunmap_irq(char *buffer, unsigned long *flags)
535 unsigned long ptr = (unsigned long) buffer & PAGE_MASK;
537 kunmap_atomic((void *) ptr);
538 local_irq_restore(*flags);
542 static inline char *bvec_kmap_irq(struct bio_vec *bvec, unsigned long *flags)
544 return page_address(bvec->bv_page) + bvec->bv_offset;
547 static inline void bvec_kunmap_irq(char *buffer, unsigned long *flags)
553 static inline char *__bio_kmap_irq(struct bio *bio, struct bvec_iter iter,
554 unsigned long *flags)
556 return bvec_kmap_irq(&bio_iter_iovec(bio, iter), flags);
558 #define __bio_kunmap_irq(buf, flags) bvec_kunmap_irq(buf, flags)
560 #define bio_kmap_irq(bio, flags) \
561 __bio_kmap_irq((bio), (bio)->bi_iter, (flags))
562 #define bio_kunmap_irq(buf,flags) __bio_kunmap_irq(buf, flags)
565 * BIO list management for use by remapping drivers (e.g. DM or MD) and loop.
567 * A bio_list anchors a singly-linked list of bios chained through the bi_next
568 * member of the bio. The bio_list also caches the last list member to allow
569 * fast access to the tail.
576 static inline int bio_list_empty(const struct bio_list *bl)
578 return bl->head == NULL;
581 static inline void bio_list_init(struct bio_list *bl)
583 bl->head = bl->tail = NULL;
586 #define BIO_EMPTY_LIST { NULL, NULL }
588 #define bio_list_for_each(bio, bl) \
589 for (bio = (bl)->head; bio; bio = bio->bi_next)
591 static inline unsigned bio_list_size(const struct bio_list *bl)
596 bio_list_for_each(bio, bl)
602 static inline void bio_list_add(struct bio_list *bl, struct bio *bio)
607 bl->tail->bi_next = bio;
614 static inline void bio_list_add_head(struct bio_list *bl, struct bio *bio)
616 bio->bi_next = bl->head;
624 static inline void bio_list_merge(struct bio_list *bl, struct bio_list *bl2)
630 bl->tail->bi_next = bl2->head;
632 bl->head = bl2->head;
634 bl->tail = bl2->tail;
637 static inline void bio_list_merge_head(struct bio_list *bl,
638 struct bio_list *bl2)
644 bl2->tail->bi_next = bl->head;
646 bl->tail = bl2->tail;
648 bl->head = bl2->head;
651 static inline struct bio *bio_list_peek(struct bio_list *bl)
656 static inline struct bio *bio_list_pop(struct bio_list *bl)
658 struct bio *bio = bl->head;
661 bl->head = bl->head->bi_next;
671 static inline struct bio *bio_list_get(struct bio_list *bl)
673 struct bio *bio = bl->head;
675 bl->head = bl->tail = NULL;
681 * Increment chain count for the bio. Make sure the CHAIN flag update
682 * is visible before the raised count.
684 static inline void bio_inc_remaining(struct bio *bio)
686 bio_set_flag(bio, BIO_CHAIN);
687 smp_mb__before_atomic();
688 atomic_inc(&bio->__bi_remaining);
692 * bio_set is used to allow other portions of the IO system to
693 * allocate their own private memory pools for bio and iovec structures.
694 * These memory pools in turn all allocate from the bio_slab
695 * and the bvec_slabs[].
697 #define BIO_POOL_SIZE 2
700 struct kmem_cache *bio_slab;
701 unsigned int front_pad;
704 mempool_t *bvec_pool;
705 #if defined(CONFIG_BLK_DEV_INTEGRITY)
706 mempool_t *bio_integrity_pool;
707 mempool_t *bvec_integrity_pool;
711 * Deadlock avoidance for stacking block drivers: see comments in
712 * bio_alloc_bioset() for details
714 spinlock_t rescue_lock;
715 struct bio_list rescue_list;
716 struct work_struct rescue_work;
717 struct workqueue_struct *rescue_workqueue;
723 struct kmem_cache *slab;
727 * a small number of entries is fine, not going to be performance critical.
728 * basically we just need to survive
730 #define BIO_SPLIT_ENTRIES 2
732 #if defined(CONFIG_BLK_DEV_INTEGRITY)
734 #define bip_for_each_vec(bvl, bip, iter) \
735 for_each_bvec(bvl, (bip)->bip_vec, iter, (bip)->bip_iter)
737 #define bio_for_each_integrity_vec(_bvl, _bio, _iter) \
739 bip_for_each_vec(_bvl, _bio->bi_integrity, _iter)
741 extern struct bio_integrity_payload *bio_integrity_alloc(struct bio *, gfp_t, unsigned int);
742 extern void bio_integrity_free(struct bio *);
743 extern int bio_integrity_add_page(struct bio *, struct page *, unsigned int, unsigned int);
744 extern bool bio_integrity_prep(struct bio *);
745 extern void bio_integrity_endio(struct bio *);
746 extern void bio_integrity_advance(struct bio *, unsigned int);
747 extern void bio_integrity_trim(struct bio *);
748 extern int bio_integrity_clone(struct bio *, struct bio *, gfp_t);
749 extern int bioset_integrity_create(struct bio_set *, int);
750 extern void bioset_integrity_free(struct bio_set *);
751 extern void bio_integrity_init(void);
753 #else /* CONFIG_BLK_DEV_INTEGRITY */
755 static inline void *bio_integrity(struct bio *bio)
760 static inline int bioset_integrity_create(struct bio_set *bs, int pool_size)
765 static inline void bioset_integrity_free (struct bio_set *bs)
770 static inline bool bio_integrity_prep(struct bio *bio)
775 static inline void bio_integrity_free(struct bio *bio)
780 static inline int bio_integrity_clone(struct bio *bio, struct bio *bio_src,
786 static inline void bio_integrity_advance(struct bio *bio,
787 unsigned int bytes_done)
792 static inline void bio_integrity_trim(struct bio *bio)
797 static inline void bio_integrity_init(void)
802 static inline bool bio_integrity_flagged(struct bio *bio, enum bip_flags flag)
807 static inline void *bio_integrity_alloc(struct bio * bio, gfp_t gfp,
810 return ERR_PTR(-EINVAL);
813 static inline int bio_integrity_add_page(struct bio *bio, struct page *page,
814 unsigned int len, unsigned int offset)
819 #endif /* CONFIG_BLK_DEV_INTEGRITY */
821 #endif /* CONFIG_BLOCK */
822 #endif /* __LINUX_BIO_H */