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 #ifdef CONFIG_THP_SWAP
42 #if HPAGE_PMD_NR > 256
43 #define BIO_MAX_PAGES HPAGE_PMD_NR
45 #define BIO_MAX_PAGES 256
48 #define BIO_MAX_PAGES 256
51 #define bio_prio(bio) (bio)->bi_ioprio
52 #define bio_set_prio(bio, prio) ((bio)->bi_ioprio = prio)
54 #define bio_iter_iovec(bio, iter) \
55 bvec_iter_bvec((bio)->bi_io_vec, (iter))
57 #define bio_iter_page(bio, iter) \
58 bvec_iter_page((bio)->bi_io_vec, (iter))
59 #define bio_iter_len(bio, iter) \
60 bvec_iter_len((bio)->bi_io_vec, (iter))
61 #define bio_iter_offset(bio, iter) \
62 bvec_iter_offset((bio)->bi_io_vec, (iter))
64 #define bio_page(bio) bio_iter_page((bio), (bio)->bi_iter)
65 #define bio_offset(bio) bio_iter_offset((bio), (bio)->bi_iter)
66 #define bio_iovec(bio) bio_iter_iovec((bio), (bio)->bi_iter)
68 #define bio_multiple_segments(bio) \
69 ((bio)->bi_iter.bi_size != bio_iovec(bio).bv_len)
70 #define bio_sectors(bio) ((bio)->bi_iter.bi_size >> 9)
71 #define bio_end_sector(bio) ((bio)->bi_iter.bi_sector + bio_sectors((bio)))
74 * Return the data direction, READ or WRITE.
76 #define bio_data_dir(bio) \
77 (op_is_write(bio_op(bio)) ? WRITE : READ)
80 * Check whether this bio carries any data or not. A NULL bio is allowed.
82 static inline bool bio_has_data(struct bio *bio)
85 bio->bi_iter.bi_size &&
86 bio_op(bio) != REQ_OP_DISCARD &&
87 bio_op(bio) != REQ_OP_SECURE_ERASE &&
88 bio_op(bio) != REQ_OP_WRITE_ZEROES)
94 static inline bool bio_no_advance_iter(struct bio *bio)
96 return bio_op(bio) == REQ_OP_DISCARD ||
97 bio_op(bio) == REQ_OP_SECURE_ERASE ||
98 bio_op(bio) == REQ_OP_WRITE_SAME ||
99 bio_op(bio) == REQ_OP_WRITE_ZEROES;
102 static inline bool bio_mergeable(struct bio *bio)
104 if (bio->bi_opf & REQ_NOMERGE_FLAGS)
110 static inline unsigned int bio_cur_bytes(struct bio *bio)
112 if (bio_has_data(bio))
113 return bio_iovec(bio).bv_len;
114 else /* dataless requests such as discard */
115 return bio->bi_iter.bi_size;
118 static inline void *bio_data(struct bio *bio)
120 if (bio_has_data(bio))
121 return page_address(bio_page(bio)) + bio_offset(bio);
126 static inline bool bio_full(struct bio *bio)
128 return bio->bi_vcnt >= bio->bi_max_vecs;
134 #define bvec_to_phys(bv) (page_to_phys((bv)->bv_page) + (unsigned long) (bv)->bv_offset)
137 * queues that have highmem support enabled may still need to revert to
138 * PIO transfers occasionally and thus map high pages temporarily. For
139 * permanent PIO fall back, user is probably better off disabling highmem
140 * I/O completely on that queue (see ide-dma for example)
142 #define __bio_kmap_atomic(bio, iter) \
143 (kmap_atomic(bio_iter_iovec((bio), (iter)).bv_page) + \
144 bio_iter_iovec((bio), (iter)).bv_offset)
146 #define __bio_kunmap_atomic(addr) kunmap_atomic(addr)
152 /* Default implementation of BIOVEC_PHYS_MERGEABLE */
153 #define __BIOVEC_PHYS_MERGEABLE(vec1, vec2) \
154 ((bvec_to_phys((vec1)) + (vec1)->bv_len) == bvec_to_phys((vec2)))
157 * allow arch override, for eg virtualized architectures (put in asm/io.h)
159 #ifndef BIOVEC_PHYS_MERGEABLE
160 #define BIOVEC_PHYS_MERGEABLE(vec1, vec2) \
161 __BIOVEC_PHYS_MERGEABLE(vec1, vec2)
164 #define __BIO_SEG_BOUNDARY(addr1, addr2, mask) \
165 (((addr1) | (mask)) == (((addr2) - 1) | (mask)))
166 #define BIOVEC_SEG_BOUNDARY(q, b1, b2) \
167 __BIO_SEG_BOUNDARY(bvec_to_phys((b1)), bvec_to_phys((b2)) + (b2)->bv_len, queue_segment_boundary((q)))
170 * drivers should _never_ use the all version - the bio may have been split
171 * before it got to the driver and the driver won't own all of it
173 #define bio_for_each_segment_all(bvl, bio, i) \
174 for (i = 0, bvl = (bio)->bi_io_vec; i < (bio)->bi_vcnt; i++, bvl++)
176 static inline void bio_advance_iter(struct bio *bio, struct bvec_iter *iter,
179 iter->bi_sector += bytes >> 9;
181 if (bio_no_advance_iter(bio)) {
182 iter->bi_size -= bytes;
183 iter->bi_done += bytes;
185 bvec_iter_advance(bio->bi_io_vec, iter, bytes);
186 /* TODO: It is reasonable to complete bio with error here. */
190 static inline bool bio_rewind_iter(struct bio *bio, struct bvec_iter *iter,
193 iter->bi_sector -= bytes >> 9;
195 if (bio_no_advance_iter(bio)) {
196 iter->bi_size += bytes;
197 iter->bi_done -= bytes;
201 return bvec_iter_rewind(bio->bi_io_vec, iter, bytes);
204 #define __bio_for_each_segment(bvl, bio, iter, start) \
205 for (iter = (start); \
207 ((bvl = bio_iter_iovec((bio), (iter))), 1); \
208 bio_advance_iter((bio), &(iter), (bvl).bv_len))
210 #define bio_for_each_segment(bvl, bio, iter) \
211 __bio_for_each_segment(bvl, bio, iter, (bio)->bi_iter)
213 #define bio_iter_last(bvec, iter) ((iter).bi_size == (bvec).bv_len)
215 static inline unsigned bio_segments(struct bio *bio)
219 struct bvec_iter iter;
222 * We special case discard/write same/write zeroes, because they
223 * interpret bi_size differently:
226 switch (bio_op(bio)) {
228 case REQ_OP_SECURE_ERASE:
229 case REQ_OP_WRITE_ZEROES:
231 case REQ_OP_WRITE_SAME:
237 bio_for_each_segment(bv, bio, iter)
244 * get a reference to a bio, so it won't disappear. the intended use is
248 * submit_bio(rw, bio);
249 * if (bio->bi_flags ...)
253 * without the bio_get(), it could potentially complete I/O before submit_bio
254 * returns. and then bio would be freed memory when if (bio->bi_flags ...)
257 static inline void bio_get(struct bio *bio)
259 bio->bi_flags |= (1 << BIO_REFFED);
260 smp_mb__before_atomic();
261 atomic_inc(&bio->__bi_cnt);
264 static inline void bio_cnt_set(struct bio *bio, unsigned int count)
267 bio->bi_flags |= (1 << BIO_REFFED);
270 atomic_set(&bio->__bi_cnt, count);
273 static inline bool bio_flagged(struct bio *bio, unsigned int bit)
275 return (bio->bi_flags & (1U << bit)) != 0;
278 static inline void bio_set_flag(struct bio *bio, unsigned int bit)
280 bio->bi_flags |= (1U << bit);
283 static inline void bio_clear_flag(struct bio *bio, unsigned int bit)
285 bio->bi_flags &= ~(1U << bit);
288 static inline void bio_get_first_bvec(struct bio *bio, struct bio_vec *bv)
290 *bv = bio_iovec(bio);
293 static inline void bio_get_last_bvec(struct bio *bio, struct bio_vec *bv)
295 struct bvec_iter iter = bio->bi_iter;
298 if (unlikely(!bio_multiple_segments(bio))) {
299 *bv = bio_iovec(bio);
303 bio_advance_iter(bio, &iter, iter.bi_size);
305 if (!iter.bi_bvec_done)
306 idx = iter.bi_idx - 1;
307 else /* in the middle of bvec */
310 *bv = bio->bi_io_vec[idx];
313 * iter.bi_bvec_done records actual length of the last bvec
314 * if this bio ends in the middle of one io vector
316 if (iter.bi_bvec_done)
317 bv->bv_len = iter.bi_bvec_done;
321 BIP_BLOCK_INTEGRITY = 1 << 0, /* block layer owns integrity data */
322 BIP_MAPPED_INTEGRITY = 1 << 1, /* ref tag has been remapped */
323 BIP_CTRL_NOCHECK = 1 << 2, /* disable HBA integrity checking */
324 BIP_DISK_NOCHECK = 1 << 3, /* disable disk integrity checking */
325 BIP_IP_CHECKSUM = 1 << 4, /* IP checksum */
329 * bio integrity payload
331 struct bio_integrity_payload {
332 struct bio *bip_bio; /* parent bio */
334 struct bvec_iter bip_iter;
336 unsigned short bip_slab; /* slab the bip came from */
337 unsigned short bip_vcnt; /* # of integrity bio_vecs */
338 unsigned short bip_max_vcnt; /* integrity bio_vec slots */
339 unsigned short bip_flags; /* control flags */
341 struct work_struct bip_work; /* I/O completion */
343 struct bio_vec *bip_vec;
344 struct bio_vec bip_inline_vecs[0];/* embedded bvec array */
347 #if defined(CONFIG_BLK_DEV_INTEGRITY)
349 static inline struct bio_integrity_payload *bio_integrity(struct bio *bio)
351 if (bio->bi_opf & REQ_INTEGRITY)
352 return bio->bi_integrity;
357 static inline bool bio_integrity_flagged(struct bio *bio, enum bip_flags flag)
359 struct bio_integrity_payload *bip = bio_integrity(bio);
362 return bip->bip_flags & flag;
367 static inline sector_t bip_get_seed(struct bio_integrity_payload *bip)
369 return bip->bip_iter.bi_sector;
372 static inline void bip_set_seed(struct bio_integrity_payload *bip,
375 bip->bip_iter.bi_sector = seed;
378 #endif /* CONFIG_BLK_DEV_INTEGRITY */
380 extern void bio_trim(struct bio *bio, int offset, int size);
381 extern struct bio *bio_split(struct bio *bio, int sectors,
382 gfp_t gfp, struct bio_set *bs);
385 * bio_next_split - get next @sectors from a bio, splitting if necessary
387 * @sectors: number of sectors to split from the front of @bio
389 * @bs: bio set to allocate from
391 * Returns a bio representing the next @sectors of @bio - if the bio is smaller
392 * than @sectors, returns the original bio unchanged.
394 static inline struct bio *bio_next_split(struct bio *bio, int sectors,
395 gfp_t gfp, struct bio_set *bs)
397 if (sectors >= bio_sectors(bio))
400 return bio_split(bio, sectors, gfp, bs);
403 extern struct bio_set *bioset_create(unsigned int, unsigned int, int flags);
405 BIOSET_NEED_BVECS = BIT(0),
406 BIOSET_NEED_RESCUER = BIT(1),
408 extern void bioset_free(struct bio_set *);
409 extern mempool_t *biovec_create_pool(int pool_entries);
411 extern struct bio *bio_alloc_bioset(gfp_t, unsigned int, struct bio_set *);
412 extern void bio_put(struct bio *);
414 extern void __bio_clone_fast(struct bio *, struct bio *);
415 extern struct bio *bio_clone_fast(struct bio *, gfp_t, struct bio_set *);
416 extern struct bio *bio_clone_bioset(struct bio *, gfp_t, struct bio_set *bs);
418 extern struct bio_set *fs_bio_set;
420 static inline struct bio *bio_alloc(gfp_t gfp_mask, unsigned int nr_iovecs)
422 return bio_alloc_bioset(gfp_mask, nr_iovecs, fs_bio_set);
425 static inline struct bio *bio_kmalloc(gfp_t gfp_mask, unsigned int nr_iovecs)
427 return bio_alloc_bioset(gfp_mask, nr_iovecs, NULL);
430 static inline struct bio *bio_clone_kmalloc(struct bio *bio, gfp_t gfp_mask)
432 return bio_clone_bioset(bio, gfp_mask, NULL);
436 extern blk_qc_t submit_bio(struct bio *);
438 extern void bio_endio(struct bio *);
440 static inline void bio_io_error(struct bio *bio)
442 bio->bi_status = BLK_STS_IOERR;
446 static inline void bio_wouldblock_error(struct bio *bio)
448 bio->bi_status = BLK_STS_AGAIN;
452 struct request_queue;
453 extern int bio_phys_segments(struct request_queue *, struct bio *);
455 extern int submit_bio_wait(struct bio *bio);
456 extern void bio_advance(struct bio *, unsigned);
458 extern void bio_init(struct bio *bio, struct bio_vec *table,
459 unsigned short max_vecs);
460 extern void bio_uninit(struct bio *);
461 extern void bio_reset(struct bio *);
462 void bio_chain(struct bio *, struct bio *);
464 extern int bio_add_page(struct bio *, struct page *, unsigned int,unsigned int);
465 extern int bio_add_pc_page(struct request_queue *, struct bio *, struct page *,
466 unsigned int, unsigned int);
467 bool __bio_try_merge_page(struct bio *bio, struct page *page,
468 unsigned int len, unsigned int off);
469 void __bio_add_page(struct bio *bio, struct page *page,
470 unsigned int len, unsigned int off);
471 int bio_iov_iter_get_pages(struct bio *bio, struct iov_iter *iter);
473 extern struct bio *bio_map_user_iov(struct request_queue *,
474 const struct iov_iter *, gfp_t);
475 extern void bio_unmap_user(struct bio *);
476 extern struct bio *bio_map_kern(struct request_queue *, void *, unsigned int,
478 extern struct bio *bio_copy_kern(struct request_queue *, void *, unsigned int,
480 extern void bio_set_pages_dirty(struct bio *bio);
481 extern void bio_check_pages_dirty(struct bio *bio);
483 void generic_start_io_acct(struct request_queue *q, int rw,
484 unsigned long sectors, struct hd_struct *part);
485 void generic_end_io_acct(struct request_queue *q, int rw,
486 struct hd_struct *part,
487 unsigned long start_time);
489 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
490 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
492 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
493 extern void bio_flush_dcache_pages(struct bio *bi);
495 static inline void bio_flush_dcache_pages(struct bio *bi)
500 extern void bio_copy_data(struct bio *dst, struct bio *src);
501 extern int bio_alloc_pages(struct bio *bio, gfp_t gfp);
502 extern void bio_free_pages(struct bio *bio);
504 extern struct bio *bio_copy_user_iov(struct request_queue *,
505 struct rq_map_data *,
506 const struct iov_iter *,
508 extern int bio_uncopy_user(struct bio *);
509 void zero_fill_bio(struct bio *bio);
510 extern struct bio_vec *bvec_alloc(gfp_t, int, unsigned long *, mempool_t *);
511 extern void bvec_free(mempool_t *, struct bio_vec *, unsigned int);
512 extern unsigned int bvec_nr_vecs(unsigned short idx);
513 extern const char *bio_devname(struct bio *bio, char *buffer);
515 #define bio_set_dev(bio, bdev) \
517 if ((bio)->bi_disk != (bdev)->bd_disk) \
518 bio_clear_flag(bio, BIO_THROTTLED);\
519 (bio)->bi_disk = (bdev)->bd_disk; \
520 (bio)->bi_partno = (bdev)->bd_partno; \
523 #define bio_copy_dev(dst, src) \
525 (dst)->bi_disk = (src)->bi_disk; \
526 (dst)->bi_partno = (src)->bi_partno; \
529 #define bio_dev(bio) \
530 disk_devt((bio)->bi_disk)
532 #ifdef CONFIG_BLK_CGROUP
533 int bio_associate_blkcg(struct bio *bio, struct cgroup_subsys_state *blkcg_css);
534 int bio_associate_current(struct bio *bio);
535 void bio_disassociate_task(struct bio *bio);
536 void bio_clone_blkcg_association(struct bio *dst, struct bio *src);
537 #else /* CONFIG_BLK_CGROUP */
538 static inline int bio_associate_blkcg(struct bio *bio,
539 struct cgroup_subsys_state *blkcg_css) { return 0; }
540 static inline int bio_associate_current(struct bio *bio) { return -ENOENT; }
541 static inline void bio_disassociate_task(struct bio *bio) { }
542 static inline void bio_clone_blkcg_association(struct bio *dst,
544 #endif /* CONFIG_BLK_CGROUP */
546 #ifdef CONFIG_HIGHMEM
548 * remember never ever reenable interrupts between a bvec_kmap_irq and
551 static inline char *bvec_kmap_irq(struct bio_vec *bvec, unsigned long *flags)
556 * might not be a highmem page, but the preempt/irq count
557 * balancing is a lot nicer this way
559 local_irq_save(*flags);
560 addr = (unsigned long) kmap_atomic(bvec->bv_page);
562 BUG_ON(addr & ~PAGE_MASK);
564 return (char *) addr + bvec->bv_offset;
567 static inline void bvec_kunmap_irq(char *buffer, unsigned long *flags)
569 unsigned long ptr = (unsigned long) buffer & PAGE_MASK;
571 kunmap_atomic((void *) ptr);
572 local_irq_restore(*flags);
576 static inline char *bvec_kmap_irq(struct bio_vec *bvec, unsigned long *flags)
578 return page_address(bvec->bv_page) + bvec->bv_offset;
581 static inline void bvec_kunmap_irq(char *buffer, unsigned long *flags)
587 static inline char *__bio_kmap_irq(struct bio *bio, struct bvec_iter iter,
588 unsigned long *flags)
590 return bvec_kmap_irq(&bio_iter_iovec(bio, iter), flags);
592 #define __bio_kunmap_irq(buf, flags) bvec_kunmap_irq(buf, flags)
594 #define bio_kmap_irq(bio, flags) \
595 __bio_kmap_irq((bio), (bio)->bi_iter, (flags))
596 #define bio_kunmap_irq(buf,flags) __bio_kunmap_irq(buf, flags)
599 * BIO list management for use by remapping drivers (e.g. DM or MD) and loop.
601 * A bio_list anchors a singly-linked list of bios chained through the bi_next
602 * member of the bio. The bio_list also caches the last list member to allow
603 * fast access to the tail.
610 static inline int bio_list_empty(const struct bio_list *bl)
612 return bl->head == NULL;
615 static inline void bio_list_init(struct bio_list *bl)
617 bl->head = bl->tail = NULL;
620 #define BIO_EMPTY_LIST { NULL, NULL }
622 #define bio_list_for_each(bio, bl) \
623 for (bio = (bl)->head; bio; bio = bio->bi_next)
625 static inline unsigned bio_list_size(const struct bio_list *bl)
630 bio_list_for_each(bio, bl)
636 static inline void bio_list_add(struct bio_list *bl, struct bio *bio)
641 bl->tail->bi_next = bio;
648 static inline void bio_list_add_head(struct bio_list *bl, struct bio *bio)
650 bio->bi_next = bl->head;
658 static inline void bio_list_merge(struct bio_list *bl, struct bio_list *bl2)
664 bl->tail->bi_next = bl2->head;
666 bl->head = bl2->head;
668 bl->tail = bl2->tail;
671 static inline void bio_list_merge_head(struct bio_list *bl,
672 struct bio_list *bl2)
678 bl2->tail->bi_next = bl->head;
680 bl->tail = bl2->tail;
682 bl->head = bl2->head;
685 static inline struct bio *bio_list_peek(struct bio_list *bl)
690 static inline struct bio *bio_list_pop(struct bio_list *bl)
692 struct bio *bio = bl->head;
695 bl->head = bl->head->bi_next;
705 static inline struct bio *bio_list_get(struct bio_list *bl)
707 struct bio *bio = bl->head;
709 bl->head = bl->tail = NULL;
715 * Increment chain count for the bio. Make sure the CHAIN flag update
716 * is visible before the raised count.
718 static inline void bio_inc_remaining(struct bio *bio)
720 bio_set_flag(bio, BIO_CHAIN);
721 smp_mb__before_atomic();
722 atomic_inc(&bio->__bi_remaining);
726 * bio_set is used to allow other portions of the IO system to
727 * allocate their own private memory pools for bio and iovec structures.
728 * These memory pools in turn all allocate from the bio_slab
729 * and the bvec_slabs[].
731 #define BIO_POOL_SIZE 2
734 struct kmem_cache *bio_slab;
735 unsigned int front_pad;
738 mempool_t *bvec_pool;
739 #if defined(CONFIG_BLK_DEV_INTEGRITY)
740 mempool_t *bio_integrity_pool;
741 mempool_t *bvec_integrity_pool;
745 * Deadlock avoidance for stacking block drivers: see comments in
746 * bio_alloc_bioset() for details
748 spinlock_t rescue_lock;
749 struct bio_list rescue_list;
750 struct work_struct rescue_work;
751 struct workqueue_struct *rescue_workqueue;
757 struct kmem_cache *slab;
761 * a small number of entries is fine, not going to be performance critical.
762 * basically we just need to survive
764 #define BIO_SPLIT_ENTRIES 2
766 #if defined(CONFIG_BLK_DEV_INTEGRITY)
768 #define bip_for_each_vec(bvl, bip, iter) \
769 for_each_bvec(bvl, (bip)->bip_vec, iter, (bip)->bip_iter)
771 #define bio_for_each_integrity_vec(_bvl, _bio, _iter) \
773 bip_for_each_vec(_bvl, _bio->bi_integrity, _iter)
775 extern struct bio_integrity_payload *bio_integrity_alloc(struct bio *, gfp_t, unsigned int);
776 extern int bio_integrity_add_page(struct bio *, struct page *, unsigned int, unsigned int);
777 extern bool bio_integrity_prep(struct bio *);
778 extern void bio_integrity_advance(struct bio *, unsigned int);
779 extern void bio_integrity_trim(struct bio *);
780 extern int bio_integrity_clone(struct bio *, struct bio *, gfp_t);
781 extern int bioset_integrity_create(struct bio_set *, int);
782 extern void bioset_integrity_free(struct bio_set *);
783 extern void bio_integrity_init(void);
785 #else /* CONFIG_BLK_DEV_INTEGRITY */
787 static inline void *bio_integrity(struct bio *bio)
792 static inline int bioset_integrity_create(struct bio_set *bs, int pool_size)
797 static inline void bioset_integrity_free (struct bio_set *bs)
802 static inline bool bio_integrity_prep(struct bio *bio)
807 static inline int bio_integrity_clone(struct bio *bio, struct bio *bio_src,
813 static inline void bio_integrity_advance(struct bio *bio,
814 unsigned int bytes_done)
819 static inline void bio_integrity_trim(struct bio *bio)
824 static inline void bio_integrity_init(void)
829 static inline bool bio_integrity_flagged(struct bio *bio, enum bip_flags flag)
834 static inline void *bio_integrity_alloc(struct bio * bio, gfp_t gfp,
837 return ERR_PTR(-EINVAL);
840 static inline int bio_integrity_add_page(struct bio *bio, struct page *page,
841 unsigned int len, unsigned int offset)
846 #endif /* CONFIG_BLK_DEV_INTEGRITY */
848 #endif /* CONFIG_BLOCK */
849 #endif /* __LINUX_BIO_H */