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 * Check whether this bio carries any data or not. A NULL bio is allowed.
68 static inline bool bio_has_data(struct bio *bio)
71 bio->bi_iter.bi_size &&
72 bio_op(bio) != REQ_OP_DISCARD &&
73 bio_op(bio) != REQ_OP_SECURE_ERASE)
79 static inline bool bio_no_advance_iter(struct bio *bio)
81 return bio_op(bio) == REQ_OP_DISCARD ||
82 bio_op(bio) == REQ_OP_SECURE_ERASE ||
83 bio_op(bio) == REQ_OP_WRITE_SAME;
86 static inline bool bio_is_rw(struct bio *bio)
88 if (!bio_has_data(bio))
91 if (bio_no_advance_iter(bio))
97 static inline bool bio_mergeable(struct bio *bio)
99 if (bio->bi_opf & REQ_NOMERGE_FLAGS)
105 static inline unsigned int bio_cur_bytes(struct bio *bio)
107 if (bio_has_data(bio))
108 return bio_iovec(bio).bv_len;
109 else /* dataless requests such as discard */
110 return bio->bi_iter.bi_size;
113 static inline void *bio_data(struct bio *bio)
115 if (bio_has_data(bio))
116 return page_address(bio_page(bio)) + bio_offset(bio);
124 #define bio_to_phys(bio) (page_to_phys(bio_page((bio))) + (unsigned long) bio_offset((bio)))
125 #define bvec_to_phys(bv) (page_to_phys((bv)->bv_page) + (unsigned long) (bv)->bv_offset)
128 * queues that have highmem support enabled may still need to revert to
129 * PIO transfers occasionally and thus map high pages temporarily. For
130 * permanent PIO fall back, user is probably better off disabling highmem
131 * I/O completely on that queue (see ide-dma for example)
133 #define __bio_kmap_atomic(bio, iter) \
134 (kmap_atomic(bio_iter_iovec((bio), (iter)).bv_page) + \
135 bio_iter_iovec((bio), (iter)).bv_offset)
137 #define __bio_kunmap_atomic(addr) kunmap_atomic(addr)
143 /* Default implementation of BIOVEC_PHYS_MERGEABLE */
144 #define __BIOVEC_PHYS_MERGEABLE(vec1, vec2) \
145 ((bvec_to_phys((vec1)) + (vec1)->bv_len) == bvec_to_phys((vec2)))
148 * allow arch override, for eg virtualized architectures (put in asm/io.h)
150 #ifndef BIOVEC_PHYS_MERGEABLE
151 #define BIOVEC_PHYS_MERGEABLE(vec1, vec2) \
152 __BIOVEC_PHYS_MERGEABLE(vec1, vec2)
155 #define __BIO_SEG_BOUNDARY(addr1, addr2, mask) \
156 (((addr1) | (mask)) == (((addr2) - 1) | (mask)))
157 #define BIOVEC_SEG_BOUNDARY(q, b1, b2) \
158 __BIO_SEG_BOUNDARY(bvec_to_phys((b1)), bvec_to_phys((b2)) + (b2)->bv_len, queue_segment_boundary((q)))
161 * drivers should _never_ use the all version - the bio may have been split
162 * before it got to the driver and the driver won't own all of it
164 #define bio_for_each_segment_all(bvl, bio, i) \
165 for (i = 0, bvl = (bio)->bi_io_vec; i < (bio)->bi_vcnt; i++, bvl++)
167 static inline void bio_advance_iter(struct bio *bio, struct bvec_iter *iter,
170 iter->bi_sector += bytes >> 9;
172 if (bio_no_advance_iter(bio))
173 iter->bi_size -= bytes;
175 bvec_iter_advance(bio->bi_io_vec, iter, bytes);
178 #define __bio_for_each_segment(bvl, bio, iter, start) \
179 for (iter = (start); \
181 ((bvl = bio_iter_iovec((bio), (iter))), 1); \
182 bio_advance_iter((bio), &(iter), (bvl).bv_len))
184 #define bio_for_each_segment(bvl, bio, iter) \
185 __bio_for_each_segment(bvl, bio, iter, (bio)->bi_iter)
187 #define bio_iter_last(bvec, iter) ((iter).bi_size == (bvec).bv_len)
189 static inline unsigned bio_segments(struct bio *bio)
193 struct bvec_iter iter;
196 * We special case discard/write same, because they interpret bi_size
200 if (bio_op(bio) == REQ_OP_DISCARD)
203 if (bio_op(bio) == REQ_OP_SECURE_ERASE)
206 if (bio_op(bio) == REQ_OP_WRITE_SAME)
209 bio_for_each_segment(bv, bio, iter)
216 * get a reference to a bio, so it won't disappear. the intended use is
220 * submit_bio(rw, bio);
221 * if (bio->bi_flags ...)
225 * without the bio_get(), it could potentially complete I/O before submit_bio
226 * returns. and then bio would be freed memory when if (bio->bi_flags ...)
229 static inline void bio_get(struct bio *bio)
231 bio->bi_flags |= (1 << BIO_REFFED);
232 smp_mb__before_atomic();
233 atomic_inc(&bio->__bi_cnt);
236 static inline void bio_cnt_set(struct bio *bio, unsigned int count)
239 bio->bi_flags |= (1 << BIO_REFFED);
240 smp_mb__before_atomic();
242 atomic_set(&bio->__bi_cnt, count);
245 static inline bool bio_flagged(struct bio *bio, unsigned int bit)
247 return (bio->bi_flags & (1U << bit)) != 0;
250 static inline void bio_set_flag(struct bio *bio, unsigned int bit)
252 bio->bi_flags |= (1U << bit);
255 static inline void bio_clear_flag(struct bio *bio, unsigned int bit)
257 bio->bi_flags &= ~(1U << bit);
260 static inline void bio_get_first_bvec(struct bio *bio, struct bio_vec *bv)
262 *bv = bio_iovec(bio);
265 static inline void bio_get_last_bvec(struct bio *bio, struct bio_vec *bv)
267 struct bvec_iter iter = bio->bi_iter;
270 if (unlikely(!bio_multiple_segments(bio))) {
271 *bv = bio_iovec(bio);
275 bio_advance_iter(bio, &iter, iter.bi_size);
277 if (!iter.bi_bvec_done)
278 idx = iter.bi_idx - 1;
279 else /* in the middle of bvec */
282 *bv = bio->bi_io_vec[idx];
285 * iter.bi_bvec_done records actual length of the last bvec
286 * if this bio ends in the middle of one io vector
288 if (iter.bi_bvec_done)
289 bv->bv_len = iter.bi_bvec_done;
293 BIP_BLOCK_INTEGRITY = 1 << 0, /* block layer owns integrity data */
294 BIP_MAPPED_INTEGRITY = 1 << 1, /* ref tag has been remapped */
295 BIP_CTRL_NOCHECK = 1 << 2, /* disable HBA integrity checking */
296 BIP_DISK_NOCHECK = 1 << 3, /* disable disk integrity checking */
297 BIP_IP_CHECKSUM = 1 << 4, /* IP checksum */
301 * bio integrity payload
303 struct bio_integrity_payload {
304 struct bio *bip_bio; /* parent bio */
306 struct bvec_iter bip_iter;
308 bio_end_io_t *bip_end_io; /* saved I/O completion fn */
310 unsigned short bip_slab; /* slab the bip came from */
311 unsigned short bip_vcnt; /* # of integrity bio_vecs */
312 unsigned short bip_max_vcnt; /* integrity bio_vec slots */
313 unsigned short bip_flags; /* control flags */
315 struct work_struct bip_work; /* I/O completion */
317 struct bio_vec *bip_vec;
318 struct bio_vec bip_inline_vecs[0];/* embedded bvec array */
321 #if defined(CONFIG_BLK_DEV_INTEGRITY)
323 static inline struct bio_integrity_payload *bio_integrity(struct bio *bio)
325 if (bio->bi_opf & REQ_INTEGRITY)
326 return bio->bi_integrity;
331 static inline bool bio_integrity_flagged(struct bio *bio, enum bip_flags flag)
333 struct bio_integrity_payload *bip = bio_integrity(bio);
336 return bip->bip_flags & flag;
341 static inline sector_t bip_get_seed(struct bio_integrity_payload *bip)
343 return bip->bip_iter.bi_sector;
346 static inline void bip_set_seed(struct bio_integrity_payload *bip,
349 bip->bip_iter.bi_sector = seed;
352 #endif /* CONFIG_BLK_DEV_INTEGRITY */
354 extern void bio_trim(struct bio *bio, int offset, int size);
355 extern struct bio *bio_split(struct bio *bio, int sectors,
356 gfp_t gfp, struct bio_set *bs);
359 * bio_next_split - get next @sectors from a bio, splitting if necessary
361 * @sectors: number of sectors to split from the front of @bio
363 * @bs: bio set to allocate from
365 * Returns a bio representing the next @sectors of @bio - if the bio is smaller
366 * than @sectors, returns the original bio unchanged.
368 static inline struct bio *bio_next_split(struct bio *bio, int sectors,
369 gfp_t gfp, struct bio_set *bs)
371 if (sectors >= bio_sectors(bio))
374 return bio_split(bio, sectors, gfp, bs);
377 extern struct bio_set *bioset_create(unsigned int, unsigned int);
378 extern struct bio_set *bioset_create_nobvec(unsigned int, unsigned int);
379 extern void bioset_free(struct bio_set *);
380 extern mempool_t *biovec_create_pool(int pool_entries);
382 extern struct bio *bio_alloc_bioset(gfp_t, int, struct bio_set *);
383 extern void bio_put(struct bio *);
385 extern void __bio_clone_fast(struct bio *, struct bio *);
386 extern struct bio *bio_clone_fast(struct bio *, gfp_t, struct bio_set *);
387 extern struct bio *bio_clone_bioset(struct bio *, gfp_t, struct bio_set *bs);
389 extern struct bio_set *fs_bio_set;
391 static inline struct bio *bio_alloc(gfp_t gfp_mask, unsigned int nr_iovecs)
393 return bio_alloc_bioset(gfp_mask, nr_iovecs, fs_bio_set);
396 static inline struct bio *bio_clone(struct bio *bio, gfp_t gfp_mask)
398 return bio_clone_bioset(bio, gfp_mask, fs_bio_set);
401 static inline struct bio *bio_kmalloc(gfp_t gfp_mask, unsigned int nr_iovecs)
403 return bio_alloc_bioset(gfp_mask, nr_iovecs, NULL);
406 static inline struct bio *bio_clone_kmalloc(struct bio *bio, gfp_t gfp_mask)
408 return bio_clone_bioset(bio, gfp_mask, NULL);
412 extern void bio_endio(struct bio *);
414 static inline void bio_io_error(struct bio *bio)
416 bio->bi_error = -EIO;
420 struct request_queue;
421 extern int bio_phys_segments(struct request_queue *, struct bio *);
423 extern int submit_bio_wait(struct bio *bio);
424 extern void bio_advance(struct bio *, unsigned);
426 extern void bio_init(struct bio *);
427 extern void bio_reset(struct bio *);
428 void bio_chain(struct bio *, struct bio *);
430 extern int bio_add_page(struct bio *, struct page *, unsigned int,unsigned int);
431 extern int bio_add_pc_page(struct request_queue *, struct bio *, struct page *,
432 unsigned int, unsigned int);
434 extern struct bio *bio_map_user_iov(struct request_queue *,
435 const struct iov_iter *, gfp_t);
436 extern void bio_unmap_user(struct bio *);
437 extern struct bio *bio_map_kern(struct request_queue *, void *, unsigned int,
439 extern struct bio *bio_copy_kern(struct request_queue *, void *, unsigned int,
441 extern void bio_set_pages_dirty(struct bio *bio);
442 extern void bio_check_pages_dirty(struct bio *bio);
444 void generic_start_io_acct(int rw, unsigned long sectors,
445 struct hd_struct *part);
446 void generic_end_io_acct(int rw, struct hd_struct *part,
447 unsigned long start_time);
449 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
450 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
452 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
453 extern void bio_flush_dcache_pages(struct bio *bi);
455 static inline void bio_flush_dcache_pages(struct bio *bi)
460 extern void bio_copy_data(struct bio *dst, struct bio *src);
461 extern int bio_alloc_pages(struct bio *bio, gfp_t gfp);
462 extern void bio_free_pages(struct bio *bio);
464 extern struct bio *bio_copy_user_iov(struct request_queue *,
465 struct rq_map_data *,
466 const struct iov_iter *,
468 extern int bio_uncopy_user(struct bio *);
469 void zero_fill_bio(struct bio *bio);
470 extern struct bio_vec *bvec_alloc(gfp_t, int, unsigned long *, mempool_t *);
471 extern void bvec_free(mempool_t *, struct bio_vec *, unsigned int);
472 extern unsigned int bvec_nr_vecs(unsigned short idx);
474 #ifdef CONFIG_BLK_CGROUP
475 int bio_associate_blkcg(struct bio *bio, struct cgroup_subsys_state *blkcg_css);
476 int bio_associate_current(struct bio *bio);
477 void bio_disassociate_task(struct bio *bio);
478 void bio_clone_blkcg_association(struct bio *dst, struct bio *src);
479 #else /* CONFIG_BLK_CGROUP */
480 static inline int bio_associate_blkcg(struct bio *bio,
481 struct cgroup_subsys_state *blkcg_css) { return 0; }
482 static inline int bio_associate_current(struct bio *bio) { return -ENOENT; }
483 static inline void bio_disassociate_task(struct bio *bio) { }
484 static inline void bio_clone_blkcg_association(struct bio *dst,
486 #endif /* CONFIG_BLK_CGROUP */
488 #ifdef CONFIG_HIGHMEM
490 * remember never ever reenable interrupts between a bvec_kmap_irq and
493 static inline char *bvec_kmap_irq(struct bio_vec *bvec, unsigned long *flags)
498 * might not be a highmem page, but the preempt/irq count
499 * balancing is a lot nicer this way
501 local_irq_save(*flags);
502 addr = (unsigned long) kmap_atomic(bvec->bv_page);
504 BUG_ON(addr & ~PAGE_MASK);
506 return (char *) addr + bvec->bv_offset;
509 static inline void bvec_kunmap_irq(char *buffer, unsigned long *flags)
511 unsigned long ptr = (unsigned long) buffer & PAGE_MASK;
513 kunmap_atomic((void *) ptr);
514 local_irq_restore(*flags);
518 static inline char *bvec_kmap_irq(struct bio_vec *bvec, unsigned long *flags)
520 return page_address(bvec->bv_page) + bvec->bv_offset;
523 static inline void bvec_kunmap_irq(char *buffer, unsigned long *flags)
529 static inline char *__bio_kmap_irq(struct bio *bio, struct bvec_iter iter,
530 unsigned long *flags)
532 return bvec_kmap_irq(&bio_iter_iovec(bio, iter), flags);
534 #define __bio_kunmap_irq(buf, flags) bvec_kunmap_irq(buf, flags)
536 #define bio_kmap_irq(bio, flags) \
537 __bio_kmap_irq((bio), (bio)->bi_iter, (flags))
538 #define bio_kunmap_irq(buf,flags) __bio_kunmap_irq(buf, flags)
541 * BIO list management for use by remapping drivers (e.g. DM or MD) and loop.
543 * A bio_list anchors a singly-linked list of bios chained through the bi_next
544 * member of the bio. The bio_list also caches the last list member to allow
545 * fast access to the tail.
552 static inline int bio_list_empty(const struct bio_list *bl)
554 return bl->head == NULL;
557 static inline void bio_list_init(struct bio_list *bl)
559 bl->head = bl->tail = NULL;
562 #define BIO_EMPTY_LIST { NULL, NULL }
564 #define bio_list_for_each(bio, bl) \
565 for (bio = (bl)->head; bio; bio = bio->bi_next)
567 static inline unsigned bio_list_size(const struct bio_list *bl)
572 bio_list_for_each(bio, bl)
578 static inline void bio_list_add(struct bio_list *bl, struct bio *bio)
583 bl->tail->bi_next = bio;
590 static inline void bio_list_add_head(struct bio_list *bl, struct bio *bio)
592 bio->bi_next = bl->head;
600 static inline void bio_list_merge(struct bio_list *bl, struct bio_list *bl2)
606 bl->tail->bi_next = bl2->head;
608 bl->head = bl2->head;
610 bl->tail = bl2->tail;
613 static inline void bio_list_merge_head(struct bio_list *bl,
614 struct bio_list *bl2)
620 bl2->tail->bi_next = bl->head;
622 bl->tail = bl2->tail;
624 bl->head = bl2->head;
627 static inline struct bio *bio_list_peek(struct bio_list *bl)
632 static inline struct bio *bio_list_pop(struct bio_list *bl)
634 struct bio *bio = bl->head;
637 bl->head = bl->head->bi_next;
647 static inline struct bio *bio_list_get(struct bio_list *bl)
649 struct bio *bio = bl->head;
651 bl->head = bl->tail = NULL;
657 * Increment chain count for the bio. Make sure the CHAIN flag update
658 * is visible before the raised count.
660 static inline void bio_inc_remaining(struct bio *bio)
662 bio_set_flag(bio, BIO_CHAIN);
663 smp_mb__before_atomic();
664 atomic_inc(&bio->__bi_remaining);
668 * bio_set is used to allow other portions of the IO system to
669 * allocate their own private memory pools for bio and iovec structures.
670 * These memory pools in turn all allocate from the bio_slab
671 * and the bvec_slabs[].
673 #define BIO_POOL_SIZE 2
676 struct kmem_cache *bio_slab;
677 unsigned int front_pad;
680 mempool_t *bvec_pool;
681 #if defined(CONFIG_BLK_DEV_INTEGRITY)
682 mempool_t *bio_integrity_pool;
683 mempool_t *bvec_integrity_pool;
687 * Deadlock avoidance for stacking block drivers: see comments in
688 * bio_alloc_bioset() for details
690 spinlock_t rescue_lock;
691 struct bio_list rescue_list;
692 struct work_struct rescue_work;
693 struct workqueue_struct *rescue_workqueue;
699 struct kmem_cache *slab;
703 * a small number of entries is fine, not going to be performance critical.
704 * basically we just need to survive
706 #define BIO_SPLIT_ENTRIES 2
708 #if defined(CONFIG_BLK_DEV_INTEGRITY)
710 #define bip_for_each_vec(bvl, bip, iter) \
711 for_each_bvec(bvl, (bip)->bip_vec, iter, (bip)->bip_iter)
713 #define bio_for_each_integrity_vec(_bvl, _bio, _iter) \
715 bip_for_each_vec(_bvl, _bio->bi_integrity, _iter)
717 extern struct bio_integrity_payload *bio_integrity_alloc(struct bio *, gfp_t, unsigned int);
718 extern void bio_integrity_free(struct bio *);
719 extern int bio_integrity_add_page(struct bio *, struct page *, unsigned int, unsigned int);
720 extern bool bio_integrity_enabled(struct bio *bio);
721 extern int bio_integrity_prep(struct bio *);
722 extern void bio_integrity_endio(struct bio *);
723 extern void bio_integrity_advance(struct bio *, unsigned int);
724 extern void bio_integrity_trim(struct bio *, unsigned int, unsigned int);
725 extern int bio_integrity_clone(struct bio *, struct bio *, gfp_t);
726 extern int bioset_integrity_create(struct bio_set *, int);
727 extern void bioset_integrity_free(struct bio_set *);
728 extern void bio_integrity_init(void);
730 #else /* CONFIG_BLK_DEV_INTEGRITY */
732 static inline void *bio_integrity(struct bio *bio)
737 static inline bool bio_integrity_enabled(struct bio *bio)
742 static inline int bioset_integrity_create(struct bio_set *bs, int pool_size)
747 static inline void bioset_integrity_free (struct bio_set *bs)
752 static inline int bio_integrity_prep(struct bio *bio)
757 static inline void bio_integrity_free(struct bio *bio)
762 static inline int bio_integrity_clone(struct bio *bio, struct bio *bio_src,
768 static inline void bio_integrity_advance(struct bio *bio,
769 unsigned int bytes_done)
774 static inline void bio_integrity_trim(struct bio *bio, unsigned int offset,
775 unsigned int sectors)
780 static inline void bio_integrity_init(void)
785 static inline bool bio_integrity_flagged(struct bio *bio, enum bip_flags flag)
790 static inline void *bio_integrity_alloc(struct bio * bio, gfp_t gfp,
793 return ERR_PTR(-EINVAL);
796 static inline int bio_integrity_add_page(struct bio *bio, struct page *page,
797 unsigned int len, unsigned int offset)
802 #endif /* CONFIG_BLK_DEV_INTEGRITY */
804 #endif /* CONFIG_BLOCK */
805 #endif /* __LINUX_BIO_H */