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 bio_iovec_idx(bio, idx) (&((bio)->bi_io_vec[(idx)]))
65 #define __bio_iovec(bio) bio_iovec_idx((bio), (bio)->bi_iter.bi_idx)
67 #define __bvec_iter_bvec(bvec, iter) (&(bvec)[(iter).bi_idx])
69 #define bvec_iter_page(bvec, iter) \
70 (__bvec_iter_bvec((bvec), (iter))->bv_page)
72 #define bvec_iter_len(bvec, iter) \
74 __bvec_iter_bvec((bvec), (iter))->bv_len - (iter).bi_bvec_done)
76 #define bvec_iter_offset(bvec, iter) \
77 (__bvec_iter_bvec((bvec), (iter))->bv_offset + (iter).bi_bvec_done)
79 #define bvec_iter_bvec(bvec, iter) \
81 .bv_page = bvec_iter_page((bvec), (iter)), \
82 .bv_len = bvec_iter_len((bvec), (iter)), \
83 .bv_offset = bvec_iter_offset((bvec), (iter)), \
86 #define bio_iter_iovec(bio, iter) \
87 bvec_iter_bvec((bio)->bi_io_vec, (iter))
89 #define bio_iter_page(bio, iter) \
90 bvec_iter_page((bio)->bi_io_vec, (iter))
91 #define bio_iter_len(bio, iter) \
92 bvec_iter_len((bio)->bi_io_vec, (iter))
93 #define bio_iter_offset(bio, iter) \
94 bvec_iter_offset((bio)->bi_io_vec, (iter))
96 #define bio_page(bio) bio_iter_page((bio), (bio)->bi_iter)
97 #define bio_offset(bio) bio_iter_offset((bio), (bio)->bi_iter)
98 #define bio_iovec(bio) bio_iter_iovec((bio), (bio)->bi_iter)
100 #define bio_segments(bio) ((bio)->bi_vcnt - (bio)->bi_iter.bi_idx)
101 #define bio_sectors(bio) ((bio)->bi_iter.bi_size >> 9)
102 #define bio_end_sector(bio) ((bio)->bi_iter.bi_sector + bio_sectors((bio)))
104 static inline unsigned int bio_cur_bytes(struct bio *bio)
107 return bio_iovec(bio).bv_len;
108 else /* dataless requests such as discard */
109 return bio->bi_iter.bi_size;
112 static inline void *bio_data(struct bio *bio)
115 return page_address(bio_page(bio)) + bio_offset(bio);
123 #define bio_to_phys(bio) (page_to_phys(bio_page((bio))) + (unsigned long) bio_offset((bio)))
124 #define bvec_to_phys(bv) (page_to_phys((bv)->bv_page) + (unsigned long) (bv)->bv_offset)
127 * queues that have highmem support enabled may still need to revert to
128 * PIO transfers occasionally and thus map high pages temporarily. For
129 * permanent PIO fall back, user is probably better off disabling highmem
130 * I/O completely on that queue (see ide-dma for example)
132 #define __bio_kmap_atomic(bio, idx) \
133 (kmap_atomic(bio_iovec_idx((bio), (idx))->bv_page) + \
134 bio_iovec_idx((bio), (idx))->bv_offset)
136 #define __bio_kunmap_atomic(addr) kunmap_atomic(addr)
142 #define __BVEC_END(bio) bio_iovec_idx((bio), (bio)->bi_vcnt - 1)
143 #define __BVEC_START(bio) bio_iovec_idx((bio), (bio)->bi_iter.bi_idx)
145 /* Default implementation of BIOVEC_PHYS_MERGEABLE */
146 #define __BIOVEC_PHYS_MERGEABLE(vec1, vec2) \
147 ((bvec_to_phys((vec1)) + (vec1)->bv_len) == bvec_to_phys((vec2)))
150 * allow arch override, for eg virtualized architectures (put in asm/io.h)
152 #ifndef BIOVEC_PHYS_MERGEABLE
153 #define BIOVEC_PHYS_MERGEABLE(vec1, vec2) \
154 __BIOVEC_PHYS_MERGEABLE(vec1, vec2)
157 #define __BIO_SEG_BOUNDARY(addr1, addr2, mask) \
158 (((addr1) | (mask)) == (((addr2) - 1) | (mask)))
159 #define BIOVEC_SEG_BOUNDARY(q, b1, b2) \
160 __BIO_SEG_BOUNDARY(bvec_to_phys((b1)), bvec_to_phys((b2)) + (b2)->bv_len, queue_segment_boundary((q)))
161 #define BIO_SEG_BOUNDARY(q, b1, b2) \
162 BIOVEC_SEG_BOUNDARY((q), __BVEC_END((b1)), __BVEC_START((b2)))
164 #define bio_io_error(bio) bio_endio((bio), -EIO)
167 * drivers should _never_ use the all version - the bio may have been split
168 * before it got to the driver and the driver won't own all of it
170 #define bio_for_each_segment_all(bvl, bio, i) \
172 bvl = bio_iovec_idx((bio), (i)), i < (bio)->bi_vcnt; \
175 static inline void bvec_iter_advance(struct bio_vec *bv, struct bvec_iter *iter,
178 WARN_ONCE(bytes > iter->bi_size,
179 "Attempted to advance past end of bvec iter\n");
182 unsigned len = min(bytes, bvec_iter_len(bv, *iter));
185 iter->bi_size -= len;
186 iter->bi_bvec_done += len;
188 if (iter->bi_bvec_done == __bvec_iter_bvec(bv, *iter)->bv_len) {
189 iter->bi_bvec_done = 0;
195 #define for_each_bvec(bvl, bio_vec, iter, start) \
196 for ((iter) = start; \
197 (bvl) = bvec_iter_bvec((bio_vec), (iter)), \
199 bvec_iter_advance((bio_vec), &(iter), (bvl).bv_len))
202 static inline void bio_advance_iter(struct bio *bio, struct bvec_iter *iter,
205 iter->bi_sector += bytes >> 9;
207 if (bio->bi_rw & BIO_NO_ADVANCE_ITER_MASK)
208 iter->bi_size -= bytes;
210 bvec_iter_advance(bio->bi_io_vec, iter, bytes);
213 #define __bio_for_each_segment(bvl, bio, iter, start) \
214 for (iter = (start); \
216 ((bvl = bio_iter_iovec((bio), (iter))), 1); \
217 bio_advance_iter((bio), &(iter), (bvl).bv_len))
219 #define bio_for_each_segment(bvl, bio, iter) \
220 __bio_for_each_segment(bvl, bio, iter, (bio)->bi_iter)
222 #define bio_iter_last(bvec, iter) ((iter).bi_size == (bvec).bv_len)
225 * get a reference to a bio, so it won't disappear. the intended use is
229 * submit_bio(rw, bio);
230 * if (bio->bi_flags ...)
234 * without the bio_get(), it could potentially complete I/O before submit_bio
235 * returns. and then bio would be freed memory when if (bio->bi_flags ...)
238 #define bio_get(bio) atomic_inc(&(bio)->bi_cnt)
240 #if defined(CONFIG_BLK_DEV_INTEGRITY)
242 * bio integrity payload
244 struct bio_integrity_payload {
245 struct bio *bip_bio; /* parent bio */
247 sector_t bip_sector; /* virtual start sector */
249 void *bip_buf; /* generated integrity data */
250 bio_end_io_t *bip_end_io; /* saved I/O completion fn */
252 unsigned int bip_size;
254 unsigned short bip_slab; /* slab the bip came from */
255 unsigned short bip_vcnt; /* # of integrity bio_vecs */
256 unsigned short bip_idx; /* current bip_vec index */
257 unsigned bip_owns_buf:1; /* should free bip_buf */
259 struct work_struct bip_work; /* I/O completion */
261 struct bio_vec *bip_vec;
262 struct bio_vec bip_inline_vecs[0];/* embedded bvec array */
264 #endif /* CONFIG_BLK_DEV_INTEGRITY */
267 * A bio_pair is used when we need to split a bio.
268 * This can only happen for a bio that refers to just one
269 * page of data, and in the unusual situation when the
270 * page crosses a chunk/device boundary
272 * The address of the master bio is stored in bio1.bi_private
273 * The address of the pool the pair was allocated from is stored
277 struct bio bio1, bio2;
278 struct bio_vec bv1, bv2;
279 #if defined(CONFIG_BLK_DEV_INTEGRITY)
280 struct bio_integrity_payload bip1, bip2;
281 struct bio_vec iv1, iv2;
286 extern struct bio_pair *bio_split(struct bio *bi, int first_sectors);
287 extern void bio_pair_release(struct bio_pair *dbio);
288 extern void bio_trim(struct bio *bio, int offset, int size);
290 extern struct bio_set *bioset_create(unsigned int, unsigned int);
291 extern void bioset_free(struct bio_set *);
292 extern mempool_t *biovec_create_pool(struct bio_set *bs, int pool_entries);
294 extern struct bio *bio_alloc_bioset(gfp_t, int, struct bio_set *);
295 extern void bio_put(struct bio *);
297 extern void __bio_clone(struct bio *, struct bio *);
298 extern struct bio *bio_clone_bioset(struct bio *, gfp_t, struct bio_set *bs);
300 extern struct bio_set *fs_bio_set;
302 static inline struct bio *bio_alloc(gfp_t gfp_mask, unsigned int nr_iovecs)
304 return bio_alloc_bioset(gfp_mask, nr_iovecs, fs_bio_set);
307 static inline struct bio *bio_clone(struct bio *bio, gfp_t gfp_mask)
309 return bio_clone_bioset(bio, gfp_mask, fs_bio_set);
312 static inline struct bio *bio_kmalloc(gfp_t gfp_mask, unsigned int nr_iovecs)
314 return bio_alloc_bioset(gfp_mask, nr_iovecs, NULL);
317 static inline struct bio *bio_clone_kmalloc(struct bio *bio, gfp_t gfp_mask)
319 return bio_clone_bioset(bio, gfp_mask, NULL);
323 extern void bio_endio(struct bio *, int);
324 struct request_queue;
325 extern int bio_phys_segments(struct request_queue *, struct bio *);
327 extern int submit_bio_wait(int rw, struct bio *bio);
328 extern void bio_advance(struct bio *, unsigned);
330 extern void bio_init(struct bio *);
331 extern void bio_reset(struct bio *);
333 extern int bio_add_page(struct bio *, struct page *, unsigned int,unsigned int);
334 extern int bio_add_pc_page(struct request_queue *, struct bio *, struct page *,
335 unsigned int, unsigned int);
336 extern int bio_get_nr_vecs(struct block_device *);
337 extern sector_t bio_sector_offset(struct bio *, unsigned short, unsigned int);
338 extern struct bio *bio_map_user(struct request_queue *, struct block_device *,
339 unsigned long, unsigned int, int, gfp_t);
342 extern struct bio *bio_map_user_iov(struct request_queue *,
343 struct block_device *,
344 struct sg_iovec *, int, int, gfp_t);
345 extern void bio_unmap_user(struct bio *);
346 extern struct bio *bio_map_kern(struct request_queue *, void *, unsigned int,
348 extern struct bio *bio_copy_kern(struct request_queue *, void *, unsigned int,
350 extern void bio_set_pages_dirty(struct bio *bio);
351 extern void bio_check_pages_dirty(struct bio *bio);
353 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
354 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
356 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
357 extern void bio_flush_dcache_pages(struct bio *bi);
359 static inline void bio_flush_dcache_pages(struct bio *bi)
364 extern void bio_copy_data(struct bio *dst, struct bio *src);
365 extern int bio_alloc_pages(struct bio *bio, gfp_t gfp);
367 extern struct bio *bio_copy_user(struct request_queue *, struct rq_map_data *,
368 unsigned long, unsigned int, int, gfp_t);
369 extern struct bio *bio_copy_user_iov(struct request_queue *,
370 struct rq_map_data *, struct sg_iovec *,
372 extern int bio_uncopy_user(struct bio *);
373 void zero_fill_bio(struct bio *bio);
374 extern struct bio_vec *bvec_alloc(gfp_t, int, unsigned long *, mempool_t *);
375 extern void bvec_free(mempool_t *, struct bio_vec *, unsigned int);
376 extern unsigned int bvec_nr_vecs(unsigned short idx);
378 #ifdef CONFIG_BLK_CGROUP
379 int bio_associate_current(struct bio *bio);
380 void bio_disassociate_task(struct bio *bio);
381 #else /* CONFIG_BLK_CGROUP */
382 static inline int bio_associate_current(struct bio *bio) { return -ENOENT; }
383 static inline void bio_disassociate_task(struct bio *bio) { }
384 #endif /* CONFIG_BLK_CGROUP */
386 #ifdef CONFIG_HIGHMEM
388 * remember never ever reenable interrupts between a bvec_kmap_irq and
391 static inline char *bvec_kmap_irq(struct bio_vec *bvec, unsigned long *flags)
396 * might not be a highmem page, but the preempt/irq count
397 * balancing is a lot nicer this way
399 local_irq_save(*flags);
400 addr = (unsigned long) kmap_atomic(bvec->bv_page);
402 BUG_ON(addr & ~PAGE_MASK);
404 return (char *) addr + bvec->bv_offset;
407 static inline void bvec_kunmap_irq(char *buffer, unsigned long *flags)
409 unsigned long ptr = (unsigned long) buffer & PAGE_MASK;
411 kunmap_atomic((void *) ptr);
412 local_irq_restore(*flags);
416 static inline char *bvec_kmap_irq(struct bio_vec *bvec, unsigned long *flags)
418 return page_address(bvec->bv_page) + bvec->bv_offset;
421 static inline void bvec_kunmap_irq(char *buffer, unsigned long *flags)
427 static inline char *__bio_kmap_irq(struct bio *bio, unsigned short idx,
428 unsigned long *flags)
430 return bvec_kmap_irq(bio_iovec_idx(bio, idx), flags);
432 #define __bio_kunmap_irq(buf, flags) bvec_kunmap_irq(buf, flags)
434 #define bio_kmap_irq(bio, flags) \
435 __bio_kmap_irq((bio), (bio)->bi_iter.bi_idx, (flags))
436 #define bio_kunmap_irq(buf,flags) __bio_kunmap_irq(buf, flags)
439 * Check whether this bio carries any data or not. A NULL bio is allowed.
441 static inline bool bio_has_data(struct bio *bio)
443 if (bio && bio->bi_vcnt)
449 static inline bool bio_is_rw(struct bio *bio)
451 if (!bio_has_data(bio))
454 if (bio->bi_rw & REQ_WRITE_SAME)
460 static inline bool bio_mergeable(struct bio *bio)
462 if (bio->bi_rw & REQ_NOMERGE_FLAGS)
469 * BIO list management for use by remapping drivers (e.g. DM or MD) and loop.
471 * A bio_list anchors a singly-linked list of bios chained through the bi_next
472 * member of the bio. The bio_list also caches the last list member to allow
473 * fast access to the tail.
480 static inline int bio_list_empty(const struct bio_list *bl)
482 return bl->head == NULL;
485 static inline void bio_list_init(struct bio_list *bl)
487 bl->head = bl->tail = NULL;
490 #define BIO_EMPTY_LIST { NULL, NULL }
492 #define bio_list_for_each(bio, bl) \
493 for (bio = (bl)->head; bio; bio = bio->bi_next)
495 static inline unsigned bio_list_size(const struct bio_list *bl)
500 bio_list_for_each(bio, bl)
506 static inline void bio_list_add(struct bio_list *bl, struct bio *bio)
511 bl->tail->bi_next = bio;
518 static inline void bio_list_add_head(struct bio_list *bl, struct bio *bio)
520 bio->bi_next = bl->head;
528 static inline void bio_list_merge(struct bio_list *bl, struct bio_list *bl2)
534 bl->tail->bi_next = bl2->head;
536 bl->head = bl2->head;
538 bl->tail = bl2->tail;
541 static inline void bio_list_merge_head(struct bio_list *bl,
542 struct bio_list *bl2)
548 bl2->tail->bi_next = bl->head;
550 bl->tail = bl2->tail;
552 bl->head = bl2->head;
555 static inline struct bio *bio_list_peek(struct bio_list *bl)
560 static inline struct bio *bio_list_pop(struct bio_list *bl)
562 struct bio *bio = bl->head;
565 bl->head = bl->head->bi_next;
575 static inline struct bio *bio_list_get(struct bio_list *bl)
577 struct bio *bio = bl->head;
579 bl->head = bl->tail = NULL;
585 * bio_set is used to allow other portions of the IO system to
586 * allocate their own private memory pools for bio and iovec structures.
587 * These memory pools in turn all allocate from the bio_slab
588 * and the bvec_slabs[].
590 #define BIO_POOL_SIZE 2
591 #define BIOVEC_NR_POOLS 6
592 #define BIOVEC_MAX_IDX (BIOVEC_NR_POOLS - 1)
595 struct kmem_cache *bio_slab;
596 unsigned int front_pad;
599 mempool_t *bvec_pool;
600 #if defined(CONFIG_BLK_DEV_INTEGRITY)
601 mempool_t *bio_integrity_pool;
602 mempool_t *bvec_integrity_pool;
606 * Deadlock avoidance for stacking block drivers: see comments in
607 * bio_alloc_bioset() for details
609 spinlock_t rescue_lock;
610 struct bio_list rescue_list;
611 struct work_struct rescue_work;
612 struct workqueue_struct *rescue_workqueue;
618 struct kmem_cache *slab;
622 * a small number of entries is fine, not going to be performance critical.
623 * basically we just need to survive
625 #define BIO_SPLIT_ENTRIES 2
627 #if defined(CONFIG_BLK_DEV_INTEGRITY)
629 #define bip_vec_idx(bip, idx) (&(bip->bip_vec[(idx)]))
630 #define bip_vec(bip) bip_vec_idx(bip, 0)
632 #define __bip_for_each_vec(bvl, bip, i, start_idx) \
633 for (bvl = bip_vec_idx((bip), (start_idx)), i = (start_idx); \
634 i < (bip)->bip_vcnt; \
637 #define bip_for_each_vec(bvl, bip, i) \
638 __bip_for_each_vec(bvl, bip, i, (bip)->bip_idx)
640 #define bio_for_each_integrity_vec(_bvl, _bio, _iter) \
642 bip_for_each_vec(_bvl, _bio->bi_integrity, _iter)
644 #define bio_integrity(bio) (bio->bi_integrity != NULL)
646 extern struct bio_integrity_payload *bio_integrity_alloc(struct bio *, gfp_t, unsigned int);
647 extern void bio_integrity_free(struct bio *);
648 extern int bio_integrity_add_page(struct bio *, struct page *, unsigned int, unsigned int);
649 extern int bio_integrity_enabled(struct bio *bio);
650 extern int bio_integrity_set_tag(struct bio *, void *, unsigned int);
651 extern int bio_integrity_get_tag(struct bio *, void *, unsigned int);
652 extern int bio_integrity_prep(struct bio *);
653 extern void bio_integrity_endio(struct bio *, int);
654 extern void bio_integrity_advance(struct bio *, unsigned int);
655 extern void bio_integrity_trim(struct bio *, unsigned int, unsigned int);
656 extern void bio_integrity_split(struct bio *, struct bio_pair *, int);
657 extern int bio_integrity_clone(struct bio *, struct bio *, gfp_t);
658 extern int bioset_integrity_create(struct bio_set *, int);
659 extern void bioset_integrity_free(struct bio_set *);
660 extern void bio_integrity_init(void);
662 #else /* CONFIG_BLK_DEV_INTEGRITY */
664 static inline int bio_integrity(struct bio *bio)
669 static inline int bio_integrity_enabled(struct bio *bio)
674 static inline int bioset_integrity_create(struct bio_set *bs, int pool_size)
679 static inline void bioset_integrity_free (struct bio_set *bs)
684 static inline int bio_integrity_prep(struct bio *bio)
689 static inline void bio_integrity_free(struct bio *bio)
694 static inline int bio_integrity_clone(struct bio *bio, struct bio *bio_src,
700 static inline void bio_integrity_split(struct bio *bio, struct bio_pair *bp,
706 static inline void bio_integrity_advance(struct bio *bio,
707 unsigned int bytes_done)
712 static inline void bio_integrity_trim(struct bio *bio, unsigned int offset,
713 unsigned int sectors)
718 static inline void bio_integrity_init(void)
723 #endif /* CONFIG_BLK_DEV_INTEGRITY */
725 #endif /* CONFIG_BLOCK */
726 #endif /* __LINUX_BIO_H */