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
44 #define BIO_MAX_SIZE (BIO_MAX_PAGES << PAGE_CACHE_SHIFT)
45 #define BIO_MAX_SECTORS (BIO_MAX_SIZE >> 9)
48 * upper 16 bits of bi_rw define the io priority of this bio
50 #define BIO_PRIO_SHIFT (8 * sizeof(unsigned long) - IOPRIO_BITS)
51 #define bio_prio(bio) ((bio)->bi_rw >> BIO_PRIO_SHIFT)
52 #define bio_prio_valid(bio) ioprio_valid(bio_prio(bio))
54 #define bio_set_prio(bio, prio) do { \
55 WARN_ON(prio >= (1 << IOPRIO_BITS)); \
56 (bio)->bi_rw &= ((1UL << BIO_PRIO_SHIFT) - 1); \
57 (bio)->bi_rw |= ((unsigned long) (prio) << BIO_PRIO_SHIFT); \
61 * various member access, note that bio_data should of course not be used
62 * on highmem page vectors
64 #define __bvec_iter_bvec(bvec, iter) (&(bvec)[(iter).bi_idx])
66 #define bvec_iter_page(bvec, iter) \
67 (__bvec_iter_bvec((bvec), (iter))->bv_page)
69 #define bvec_iter_len(bvec, iter) \
71 __bvec_iter_bvec((bvec), (iter))->bv_len - (iter).bi_bvec_done)
73 #define bvec_iter_offset(bvec, iter) \
74 (__bvec_iter_bvec((bvec), (iter))->bv_offset + (iter).bi_bvec_done)
76 #define bvec_iter_bvec(bvec, iter) \
78 .bv_page = bvec_iter_page((bvec), (iter)), \
79 .bv_len = bvec_iter_len((bvec), (iter)), \
80 .bv_offset = bvec_iter_offset((bvec), (iter)), \
83 #define bio_iter_iovec(bio, iter) \
84 bvec_iter_bvec((bio)->bi_io_vec, (iter))
86 #define bio_iter_page(bio, iter) \
87 bvec_iter_page((bio)->bi_io_vec, (iter))
88 #define bio_iter_len(bio, iter) \
89 bvec_iter_len((bio)->bi_io_vec, (iter))
90 #define bio_iter_offset(bio, iter) \
91 bvec_iter_offset((bio)->bi_io_vec, (iter))
93 #define bio_page(bio) bio_iter_page((bio), (bio)->bi_iter)
94 #define bio_offset(bio) bio_iter_offset((bio), (bio)->bi_iter)
95 #define bio_iovec(bio) bio_iter_iovec((bio), (bio)->bi_iter)
97 #define bio_multiple_segments(bio) \
98 ((bio)->bi_iter.bi_size != bio_iovec(bio).bv_len)
99 #define bio_sectors(bio) ((bio)->bi_iter.bi_size >> 9)
100 #define bio_end_sector(bio) ((bio)->bi_iter.bi_sector + bio_sectors((bio)))
103 * Check whether this bio carries any data or not. A NULL bio is allowed.
105 static inline bool bio_has_data(struct bio *bio)
108 bio->bi_iter.bi_size &&
109 !(bio->bi_rw & REQ_DISCARD))
115 static inline bool bio_is_rw(struct bio *bio)
117 if (!bio_has_data(bio))
120 if (bio->bi_rw & BIO_NO_ADVANCE_ITER_MASK)
126 static inline bool bio_mergeable(struct bio *bio)
128 if (bio->bi_rw & REQ_NOMERGE_FLAGS)
134 static inline unsigned int bio_cur_bytes(struct bio *bio)
136 if (bio_has_data(bio))
137 return bio_iovec(bio).bv_len;
138 else /* dataless requests such as discard */
139 return bio->bi_iter.bi_size;
142 static inline void *bio_data(struct bio *bio)
144 if (bio_has_data(bio))
145 return page_address(bio_page(bio)) + bio_offset(bio);
153 #define bio_to_phys(bio) (page_to_phys(bio_page((bio))) + (unsigned long) bio_offset((bio)))
154 #define bvec_to_phys(bv) (page_to_phys((bv)->bv_page) + (unsigned long) (bv)->bv_offset)
157 * queues that have highmem support enabled may still need to revert to
158 * PIO transfers occasionally and thus map high pages temporarily. For
159 * permanent PIO fall back, user is probably better off disabling highmem
160 * I/O completely on that queue (see ide-dma for example)
162 #define __bio_kmap_atomic(bio, iter) \
163 (kmap_atomic(bio_iter_iovec((bio), (iter)).bv_page) + \
164 bio_iter_iovec((bio), (iter)).bv_offset)
166 #define __bio_kunmap_atomic(addr) kunmap_atomic(addr)
172 /* Default implementation of BIOVEC_PHYS_MERGEABLE */
173 #define __BIOVEC_PHYS_MERGEABLE(vec1, vec2) \
174 ((bvec_to_phys((vec1)) + (vec1)->bv_len) == bvec_to_phys((vec2)))
177 * allow arch override, for eg virtualized architectures (put in asm/io.h)
179 #ifndef BIOVEC_PHYS_MERGEABLE
180 #define BIOVEC_PHYS_MERGEABLE(vec1, vec2) \
181 __BIOVEC_PHYS_MERGEABLE(vec1, vec2)
184 #define __BIO_SEG_BOUNDARY(addr1, addr2, mask) \
185 (((addr1) | (mask)) == (((addr2) - 1) | (mask)))
186 #define BIOVEC_SEG_BOUNDARY(q, b1, b2) \
187 __BIO_SEG_BOUNDARY(bvec_to_phys((b1)), bvec_to_phys((b2)) + (b2)->bv_len, queue_segment_boundary((q)))
190 * Check if adding a bio_vec after bprv with offset would create a gap in
191 * the SG list. Most drivers don't care about this, but some do.
193 static inline bool bvec_gap_to_prev(struct bio_vec *bprv, unsigned int offset)
195 return offset || ((bprv->bv_offset + bprv->bv_len) & (PAGE_SIZE - 1));
199 * drivers should _never_ use the all version - the bio may have been split
200 * before it got to the driver and the driver won't own all of it
202 #define bio_for_each_segment_all(bvl, bio, i) \
203 for (i = 0, bvl = (bio)->bi_io_vec; i < (bio)->bi_vcnt; i++, bvl++)
205 static inline void bvec_iter_advance(struct bio_vec *bv, struct bvec_iter *iter,
208 WARN_ONCE(bytes > iter->bi_size,
209 "Attempted to advance past end of bvec iter\n");
212 unsigned len = min(bytes, bvec_iter_len(bv, *iter));
215 iter->bi_size -= len;
216 iter->bi_bvec_done += len;
218 if (iter->bi_bvec_done == __bvec_iter_bvec(bv, *iter)->bv_len) {
219 iter->bi_bvec_done = 0;
225 #define for_each_bvec(bvl, bio_vec, iter, start) \
226 for (iter = (start); \
228 ((bvl = bvec_iter_bvec((bio_vec), (iter))), 1); \
229 bvec_iter_advance((bio_vec), &(iter), (bvl).bv_len))
232 static inline void bio_advance_iter(struct bio *bio, struct bvec_iter *iter,
235 iter->bi_sector += bytes >> 9;
237 if (bio->bi_rw & BIO_NO_ADVANCE_ITER_MASK)
238 iter->bi_size -= bytes;
240 bvec_iter_advance(bio->bi_io_vec, iter, bytes);
243 #define __bio_for_each_segment(bvl, bio, iter, start) \
244 for (iter = (start); \
246 ((bvl = bio_iter_iovec((bio), (iter))), 1); \
247 bio_advance_iter((bio), &(iter), (bvl).bv_len))
249 #define bio_for_each_segment(bvl, bio, iter) \
250 __bio_for_each_segment(bvl, bio, iter, (bio)->bi_iter)
252 #define bio_iter_last(bvec, iter) ((iter).bi_size == (bvec).bv_len)
254 static inline unsigned bio_segments(struct bio *bio)
258 struct bvec_iter iter;
261 * We special case discard/write same, because they interpret bi_size
265 if (bio->bi_rw & REQ_DISCARD)
268 if (bio->bi_rw & REQ_WRITE_SAME)
271 bio_for_each_segment(bv, bio, iter)
278 * get a reference to a bio, so it won't disappear. the intended use is
282 * submit_bio(rw, bio);
283 * if (bio->bi_flags ...)
287 * without the bio_get(), it could potentially complete I/O before submit_bio
288 * returns. and then bio would be freed memory when if (bio->bi_flags ...)
291 static inline void bio_get(struct bio *bio)
293 bio->bi_flags |= (1 << BIO_REFFED);
294 smp_mb__before_atomic();
295 atomic_inc(&bio->__bi_cnt);
298 static inline void bio_cnt_set(struct bio *bio, unsigned int count)
301 bio->bi_flags |= (1 << BIO_REFFED);
302 smp_mb__before_atomic();
304 atomic_set(&bio->__bi_cnt, count);
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 */
315 #if defined(CONFIG_BLK_DEV_INTEGRITY)
317 static inline struct bio_integrity_payload *bio_integrity(struct bio *bio)
319 if (bio->bi_rw & REQ_INTEGRITY)
320 return bio->bi_integrity;
326 * bio integrity payload
328 struct bio_integrity_payload {
329 struct bio *bip_bio; /* parent bio */
331 struct bvec_iter bip_iter;
333 bio_end_io_t *bip_end_io; /* saved I/O completion fn */
335 unsigned short bip_slab; /* slab the bip came from */
336 unsigned short bip_vcnt; /* # of integrity bio_vecs */
337 unsigned short bip_max_vcnt; /* integrity bio_vec slots */
338 unsigned short bip_flags; /* control flags */
340 struct work_struct bip_work; /* I/O completion */
342 struct bio_vec *bip_vec;
343 struct bio_vec bip_inline_vecs[0];/* embedded bvec array */
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);
393 extern struct bio_set *bioset_create_nobvec(unsigned int, unsigned int);
394 extern void bioset_free(struct bio_set *);
395 extern mempool_t *biovec_create_pool(int pool_entries);
397 extern struct bio *bio_alloc_bioset(gfp_t, int, struct bio_set *);
398 extern void bio_put(struct bio *);
400 extern void __bio_clone_fast(struct bio *, struct bio *);
401 extern struct bio *bio_clone_fast(struct bio *, gfp_t, struct bio_set *);
402 extern struct bio *bio_clone_bioset(struct bio *, gfp_t, struct bio_set *bs);
404 extern struct bio_set *fs_bio_set;
406 static inline struct bio *bio_alloc(gfp_t gfp_mask, unsigned int nr_iovecs)
408 return bio_alloc_bioset(gfp_mask, nr_iovecs, fs_bio_set);
411 static inline struct bio *bio_clone(struct bio *bio, gfp_t gfp_mask)
413 return bio_clone_bioset(bio, gfp_mask, fs_bio_set);
416 static inline struct bio *bio_kmalloc(gfp_t gfp_mask, unsigned int nr_iovecs)
418 return bio_alloc_bioset(gfp_mask, nr_iovecs, NULL);
421 static inline struct bio *bio_clone_kmalloc(struct bio *bio, gfp_t gfp_mask)
423 return bio_clone_bioset(bio, gfp_mask, NULL);
427 extern void bio_endio(struct bio *);
429 static inline void bio_io_error(struct bio *bio)
431 bio->bi_error = -EIO;
435 struct request_queue;
436 extern int bio_phys_segments(struct request_queue *, struct bio *);
438 extern int submit_bio_wait(int rw, struct bio *bio);
439 extern void bio_advance(struct bio *, unsigned);
441 extern void bio_init(struct bio *);
442 extern void bio_reset(struct bio *);
443 void bio_chain(struct bio *, struct bio *);
445 extern int bio_add_page(struct bio *, struct page *, unsigned int,unsigned int);
446 extern int bio_add_pc_page(struct request_queue *, struct bio *, struct page *,
447 unsigned int, unsigned int);
448 extern int bio_get_nr_vecs(struct block_device *);
450 extern struct bio *bio_map_user_iov(struct request_queue *,
451 const struct iov_iter *, gfp_t);
452 extern void bio_unmap_user(struct bio *);
453 extern struct bio *bio_map_kern(struct request_queue *, void *, unsigned int,
455 extern struct bio *bio_copy_kern(struct request_queue *, void *, unsigned int,
457 extern void bio_set_pages_dirty(struct bio *bio);
458 extern void bio_check_pages_dirty(struct bio *bio);
460 void generic_start_io_acct(int rw, unsigned long sectors,
461 struct hd_struct *part);
462 void generic_end_io_acct(int rw, struct hd_struct *part,
463 unsigned long start_time);
465 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
466 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
468 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
469 extern void bio_flush_dcache_pages(struct bio *bi);
471 static inline void bio_flush_dcache_pages(struct bio *bi)
476 extern void bio_copy_data(struct bio *dst, struct bio *src);
477 extern int bio_alloc_pages(struct bio *bio, gfp_t gfp);
479 extern struct bio *bio_copy_user_iov(struct request_queue *,
480 struct rq_map_data *,
481 const struct iov_iter *,
483 extern int bio_uncopy_user(struct bio *);
484 void zero_fill_bio(struct bio *bio);
485 extern struct bio_vec *bvec_alloc(gfp_t, int, unsigned long *, mempool_t *);
486 extern void bvec_free(mempool_t *, struct bio_vec *, unsigned int);
487 extern unsigned int bvec_nr_vecs(unsigned short idx);
489 #ifdef CONFIG_BLK_CGROUP
490 int bio_associate_blkcg(struct bio *bio, struct cgroup_subsys_state *blkcg_css);
491 int bio_associate_current(struct bio *bio);
492 void bio_disassociate_task(struct bio *bio);
493 #else /* CONFIG_BLK_CGROUP */
494 static inline int bio_associate_blkcg(struct bio *bio,
495 struct cgroup_subsys_state *blkcg_css) { return 0; }
496 static inline int bio_associate_current(struct bio *bio) { return -ENOENT; }
497 static inline void bio_disassociate_task(struct bio *bio) { }
498 #endif /* CONFIG_BLK_CGROUP */
500 #ifdef CONFIG_HIGHMEM
502 * remember never ever reenable interrupts between a bvec_kmap_irq and
505 static inline char *bvec_kmap_irq(struct bio_vec *bvec, unsigned long *flags)
510 * might not be a highmem page, but the preempt/irq count
511 * balancing is a lot nicer this way
513 local_irq_save(*flags);
514 addr = (unsigned long) kmap_atomic(bvec->bv_page);
516 BUG_ON(addr & ~PAGE_MASK);
518 return (char *) addr + bvec->bv_offset;
521 static inline void bvec_kunmap_irq(char *buffer, unsigned long *flags)
523 unsigned long ptr = (unsigned long) buffer & PAGE_MASK;
525 kunmap_atomic((void *) ptr);
526 local_irq_restore(*flags);
530 static inline char *bvec_kmap_irq(struct bio_vec *bvec, unsigned long *flags)
532 return page_address(bvec->bv_page) + bvec->bv_offset;
535 static inline void bvec_kunmap_irq(char *buffer, unsigned long *flags)
541 static inline char *__bio_kmap_irq(struct bio *bio, struct bvec_iter iter,
542 unsigned long *flags)
544 return bvec_kmap_irq(&bio_iter_iovec(bio, iter), flags);
546 #define __bio_kunmap_irq(buf, flags) bvec_kunmap_irq(buf, flags)
548 #define bio_kmap_irq(bio, flags) \
549 __bio_kmap_irq((bio), (bio)->bi_iter, (flags))
550 #define bio_kunmap_irq(buf,flags) __bio_kunmap_irq(buf, flags)
553 * BIO list management for use by remapping drivers (e.g. DM or MD) and loop.
555 * A bio_list anchors a singly-linked list of bios chained through the bi_next
556 * member of the bio. The bio_list also caches the last list member to allow
557 * fast access to the tail.
564 static inline int bio_list_empty(const struct bio_list *bl)
566 return bl->head == NULL;
569 static inline void bio_list_init(struct bio_list *bl)
571 bl->head = bl->tail = NULL;
574 #define BIO_EMPTY_LIST { NULL, NULL }
576 #define bio_list_for_each(bio, bl) \
577 for (bio = (bl)->head; bio; bio = bio->bi_next)
579 static inline unsigned bio_list_size(const struct bio_list *bl)
584 bio_list_for_each(bio, bl)
590 static inline void bio_list_add(struct bio_list *bl, struct bio *bio)
595 bl->tail->bi_next = bio;
602 static inline void bio_list_add_head(struct bio_list *bl, struct bio *bio)
604 bio->bi_next = bl->head;
612 static inline void bio_list_merge(struct bio_list *bl, struct bio_list *bl2)
618 bl->tail->bi_next = bl2->head;
620 bl->head = bl2->head;
622 bl->tail = bl2->tail;
625 static inline void bio_list_merge_head(struct bio_list *bl,
626 struct bio_list *bl2)
632 bl2->tail->bi_next = bl->head;
634 bl->tail = bl2->tail;
636 bl->head = bl2->head;
639 static inline struct bio *bio_list_peek(struct bio_list *bl)
644 static inline struct bio *bio_list_pop(struct bio_list *bl)
646 struct bio *bio = bl->head;
649 bl->head = bl->head->bi_next;
659 static inline struct bio *bio_list_get(struct bio_list *bl)
661 struct bio *bio = bl->head;
663 bl->head = bl->tail = NULL;
669 * bio_set is used to allow other portions of the IO system to
670 * allocate their own private memory pools for bio and iovec structures.
671 * These memory pools in turn all allocate from the bio_slab
672 * and the bvec_slabs[].
674 #define BIO_POOL_SIZE 2
675 #define BIOVEC_NR_POOLS 6
676 #define BIOVEC_MAX_IDX (BIOVEC_NR_POOLS - 1)
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 #endif /* CONFIG_BLK_DEV_INTEGRITY */
795 #endif /* CONFIG_BLOCK */
796 #endif /* __LINUX_BIO_H */