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)))
189 #define bio_io_error(bio) bio_endio((bio), -EIO)
192 * drivers should _never_ use the all version - the bio may have been split
193 * before it got to the driver and the driver won't own all of it
195 #define bio_for_each_segment_all(bvl, bio, i) \
196 for (i = 0, bvl = (bio)->bi_io_vec; i < (bio)->bi_vcnt; i++, bvl++)
198 static inline void bvec_iter_advance(struct bio_vec *bv, struct bvec_iter *iter,
201 WARN_ONCE(bytes > iter->bi_size,
202 "Attempted to advance past end of bvec iter\n");
205 unsigned len = min(bytes, bvec_iter_len(bv, *iter));
208 iter->bi_size -= len;
209 iter->bi_bvec_done += len;
211 if (iter->bi_bvec_done == __bvec_iter_bvec(bv, *iter)->bv_len) {
212 iter->bi_bvec_done = 0;
218 #define for_each_bvec(bvl, bio_vec, iter, start) \
219 for ((iter) = start; \
220 (bvl) = bvec_iter_bvec((bio_vec), (iter)), \
222 bvec_iter_advance((bio_vec), &(iter), (bvl).bv_len))
225 static inline void bio_advance_iter(struct bio *bio, struct bvec_iter *iter,
228 iter->bi_sector += bytes >> 9;
230 if (bio->bi_rw & BIO_NO_ADVANCE_ITER_MASK)
231 iter->bi_size -= bytes;
233 bvec_iter_advance(bio->bi_io_vec, iter, bytes);
236 #define __bio_for_each_segment(bvl, bio, iter, start) \
237 for (iter = (start); \
239 ((bvl = bio_iter_iovec((bio), (iter))), 1); \
240 bio_advance_iter((bio), &(iter), (bvl).bv_len))
242 #define bio_for_each_segment(bvl, bio, iter) \
243 __bio_for_each_segment(bvl, bio, iter, (bio)->bi_iter)
245 #define bio_iter_last(bvec, iter) ((iter).bi_size == (bvec).bv_len)
247 static inline unsigned bio_segments(struct bio *bio)
251 struct bvec_iter iter;
253 bio_for_each_segment(bv, bio, iter)
260 * get a reference to a bio, so it won't disappear. the intended use is
264 * submit_bio(rw, bio);
265 * if (bio->bi_flags ...)
269 * without the bio_get(), it could potentially complete I/O before submit_bio
270 * returns. and then bio would be freed memory when if (bio->bi_flags ...)
273 #define bio_get(bio) atomic_inc(&(bio)->bi_cnt)
275 #if defined(CONFIG_BLK_DEV_INTEGRITY)
277 * bio integrity payload
279 struct bio_integrity_payload {
280 struct bio *bip_bio; /* parent bio */
282 struct bvec_iter bip_iter;
284 /* kill - should just use bip_vec */
285 void *bip_buf; /* generated integrity data */
287 bio_end_io_t *bip_end_io; /* saved I/O completion fn */
289 unsigned short bip_slab; /* slab the bip came from */
290 unsigned short bip_vcnt; /* # of integrity bio_vecs */
291 unsigned bip_owns_buf:1; /* should free bip_buf */
293 struct work_struct bip_work; /* I/O completion */
295 struct bio_vec *bip_vec;
296 struct bio_vec bip_inline_vecs[0];/* embedded bvec array */
298 #endif /* CONFIG_BLK_DEV_INTEGRITY */
300 extern void bio_trim(struct bio *bio, int offset, int size);
301 extern struct bio *bio_split(struct bio *bio, int sectors,
302 gfp_t gfp, struct bio_set *bs);
305 * bio_next_split - get next @sectors from a bio, splitting if necessary
307 * @sectors: number of sectors to split from the front of @bio
309 * @bs: bio set to allocate from
311 * Returns a bio representing the next @sectors of @bio - if the bio is smaller
312 * than @sectors, returns the original bio unchanged.
314 static inline struct bio *bio_next_split(struct bio *bio, int sectors,
315 gfp_t gfp, struct bio_set *bs)
317 if (sectors >= bio_sectors(bio))
320 return bio_split(bio, sectors, gfp, bs);
323 extern struct bio_set *bioset_create(unsigned int, unsigned int);
324 extern void bioset_free(struct bio_set *);
325 extern mempool_t *biovec_create_pool(struct bio_set *bs, int pool_entries);
327 extern struct bio *bio_alloc_bioset(gfp_t, int, struct bio_set *);
328 extern void bio_put(struct bio *);
330 extern void __bio_clone_fast(struct bio *, struct bio *);
331 extern struct bio *bio_clone_fast(struct bio *, gfp_t, struct bio_set *);
332 extern struct bio *bio_clone_bioset(struct bio *, gfp_t, struct bio_set *bs);
334 extern struct bio_set *fs_bio_set;
336 static inline struct bio *bio_alloc(gfp_t gfp_mask, unsigned int nr_iovecs)
338 return bio_alloc_bioset(gfp_mask, nr_iovecs, fs_bio_set);
341 static inline struct bio *bio_clone(struct bio *bio, gfp_t gfp_mask)
343 return bio_clone_bioset(bio, gfp_mask, fs_bio_set);
346 static inline struct bio *bio_kmalloc(gfp_t gfp_mask, unsigned int nr_iovecs)
348 return bio_alloc_bioset(gfp_mask, nr_iovecs, NULL);
351 static inline struct bio *bio_clone_kmalloc(struct bio *bio, gfp_t gfp_mask)
353 return bio_clone_bioset(bio, gfp_mask, NULL);
357 extern void bio_endio(struct bio *, int);
358 extern void bio_endio_nodec(struct bio *, int);
359 struct request_queue;
360 extern int bio_phys_segments(struct request_queue *, struct bio *);
362 extern int submit_bio_wait(int rw, struct bio *bio);
363 extern void bio_advance(struct bio *, unsigned);
365 extern void bio_init(struct bio *);
366 extern void bio_reset(struct bio *);
367 void bio_chain(struct bio *, struct bio *);
369 extern int bio_add_page(struct bio *, struct page *, unsigned int,unsigned int);
370 extern int bio_add_pc_page(struct request_queue *, struct bio *, struct page *,
371 unsigned int, unsigned int);
372 extern int bio_get_nr_vecs(struct block_device *);
373 extern struct bio *bio_map_user(struct request_queue *, struct block_device *,
374 unsigned long, unsigned int, int, gfp_t);
377 extern struct bio *bio_map_user_iov(struct request_queue *,
378 struct block_device *,
379 struct sg_iovec *, int, int, gfp_t);
380 extern void bio_unmap_user(struct bio *);
381 extern struct bio *bio_map_kern(struct request_queue *, void *, unsigned int,
383 extern struct bio *bio_copy_kern(struct request_queue *, void *, unsigned int,
385 extern void bio_set_pages_dirty(struct bio *bio);
386 extern void bio_check_pages_dirty(struct bio *bio);
388 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
389 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
391 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
392 extern void bio_flush_dcache_pages(struct bio *bi);
394 static inline void bio_flush_dcache_pages(struct bio *bi)
399 extern void bio_copy_data(struct bio *dst, struct bio *src);
400 extern int bio_alloc_pages(struct bio *bio, gfp_t gfp);
402 extern struct bio *bio_copy_user(struct request_queue *, struct rq_map_data *,
403 unsigned long, unsigned int, int, gfp_t);
404 extern struct bio *bio_copy_user_iov(struct request_queue *,
405 struct rq_map_data *, struct sg_iovec *,
407 extern int bio_uncopy_user(struct bio *);
408 void zero_fill_bio(struct bio *bio);
409 extern struct bio_vec *bvec_alloc(gfp_t, int, unsigned long *, mempool_t *);
410 extern void bvec_free(mempool_t *, struct bio_vec *, unsigned int);
411 extern unsigned int bvec_nr_vecs(unsigned short idx);
413 #ifdef CONFIG_BLK_CGROUP
414 int bio_associate_current(struct bio *bio);
415 void bio_disassociate_task(struct bio *bio);
416 #else /* CONFIG_BLK_CGROUP */
417 static inline int bio_associate_current(struct bio *bio) { return -ENOENT; }
418 static inline void bio_disassociate_task(struct bio *bio) { }
419 #endif /* CONFIG_BLK_CGROUP */
421 #ifdef CONFIG_HIGHMEM
423 * remember never ever reenable interrupts between a bvec_kmap_irq and
426 static inline char *bvec_kmap_irq(struct bio_vec *bvec, unsigned long *flags)
431 * might not be a highmem page, but the preempt/irq count
432 * balancing is a lot nicer this way
434 local_irq_save(*flags);
435 addr = (unsigned long) kmap_atomic(bvec->bv_page);
437 BUG_ON(addr & ~PAGE_MASK);
439 return (char *) addr + bvec->bv_offset;
442 static inline void bvec_kunmap_irq(char *buffer, unsigned long *flags)
444 unsigned long ptr = (unsigned long) buffer & PAGE_MASK;
446 kunmap_atomic((void *) ptr);
447 local_irq_restore(*flags);
451 static inline char *bvec_kmap_irq(struct bio_vec *bvec, unsigned long *flags)
453 return page_address(bvec->bv_page) + bvec->bv_offset;
456 static inline void bvec_kunmap_irq(char *buffer, unsigned long *flags)
462 static inline char *__bio_kmap_irq(struct bio *bio, struct bvec_iter iter,
463 unsigned long *flags)
465 return bvec_kmap_irq(&bio_iter_iovec(bio, iter), flags);
467 #define __bio_kunmap_irq(buf, flags) bvec_kunmap_irq(buf, flags)
469 #define bio_kmap_irq(bio, flags) \
470 __bio_kmap_irq((bio), (bio)->bi_iter, (flags))
471 #define bio_kunmap_irq(buf,flags) __bio_kunmap_irq(buf, flags)
474 * BIO list management for use by remapping drivers (e.g. DM or MD) and loop.
476 * A bio_list anchors a singly-linked list of bios chained through the bi_next
477 * member of the bio. The bio_list also caches the last list member to allow
478 * fast access to the tail.
485 static inline int bio_list_empty(const struct bio_list *bl)
487 return bl->head == NULL;
490 static inline void bio_list_init(struct bio_list *bl)
492 bl->head = bl->tail = NULL;
495 #define BIO_EMPTY_LIST { NULL, NULL }
497 #define bio_list_for_each(bio, bl) \
498 for (bio = (bl)->head; bio; bio = bio->bi_next)
500 static inline unsigned bio_list_size(const struct bio_list *bl)
505 bio_list_for_each(bio, bl)
511 static inline void bio_list_add(struct bio_list *bl, struct bio *bio)
516 bl->tail->bi_next = bio;
523 static inline void bio_list_add_head(struct bio_list *bl, struct bio *bio)
525 bio->bi_next = bl->head;
533 static inline void bio_list_merge(struct bio_list *bl, struct bio_list *bl2)
539 bl->tail->bi_next = bl2->head;
541 bl->head = bl2->head;
543 bl->tail = bl2->tail;
546 static inline void bio_list_merge_head(struct bio_list *bl,
547 struct bio_list *bl2)
553 bl2->tail->bi_next = bl->head;
555 bl->tail = bl2->tail;
557 bl->head = bl2->head;
560 static inline struct bio *bio_list_peek(struct bio_list *bl)
565 static inline struct bio *bio_list_pop(struct bio_list *bl)
567 struct bio *bio = bl->head;
570 bl->head = bl->head->bi_next;
580 static inline struct bio *bio_list_get(struct bio_list *bl)
582 struct bio *bio = bl->head;
584 bl->head = bl->tail = NULL;
590 * bio_set is used to allow other portions of the IO system to
591 * allocate their own private memory pools for bio and iovec structures.
592 * These memory pools in turn all allocate from the bio_slab
593 * and the bvec_slabs[].
595 #define BIO_POOL_SIZE 2
596 #define BIOVEC_NR_POOLS 6
597 #define BIOVEC_MAX_IDX (BIOVEC_NR_POOLS - 1)
600 struct kmem_cache *bio_slab;
601 unsigned int front_pad;
604 mempool_t *bvec_pool;
605 #if defined(CONFIG_BLK_DEV_INTEGRITY)
606 mempool_t *bio_integrity_pool;
607 mempool_t *bvec_integrity_pool;
611 * Deadlock avoidance for stacking block drivers: see comments in
612 * bio_alloc_bioset() for details
614 spinlock_t rescue_lock;
615 struct bio_list rescue_list;
616 struct work_struct rescue_work;
617 struct workqueue_struct *rescue_workqueue;
623 struct kmem_cache *slab;
627 * a small number of entries is fine, not going to be performance critical.
628 * basically we just need to survive
630 #define BIO_SPLIT_ENTRIES 2
632 #if defined(CONFIG_BLK_DEV_INTEGRITY)
636 #define bip_vec_idx(bip, idx) (&(bip->bip_vec[(idx)]))
638 #define bip_for_each_vec(bvl, bip, iter) \
639 for_each_bvec(bvl, (bip)->bip_vec, iter, (bip)->bip_iter)
641 #define bio_for_each_integrity_vec(_bvl, _bio, _iter) \
643 bip_for_each_vec(_bvl, _bio->bi_integrity, _iter)
645 #define bio_integrity(bio) (bio->bi_integrity != NULL)
647 extern struct bio_integrity_payload *bio_integrity_alloc(struct bio *, gfp_t, unsigned int);
648 extern void bio_integrity_free(struct bio *);
649 extern int bio_integrity_add_page(struct bio *, struct page *, unsigned int, unsigned int);
650 extern int bio_integrity_enabled(struct bio *bio);
651 extern int bio_integrity_set_tag(struct bio *, void *, unsigned int);
652 extern int bio_integrity_get_tag(struct bio *, void *, unsigned int);
653 extern int bio_integrity_prep(struct bio *);
654 extern void bio_integrity_endio(struct bio *, int);
655 extern void bio_integrity_advance(struct bio *, unsigned int);
656 extern void bio_integrity_trim(struct bio *, unsigned int, unsigned 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_advance(struct bio *bio,
701 unsigned int bytes_done)
706 static inline void bio_integrity_trim(struct bio *bio, unsigned int offset,
707 unsigned int sectors)
712 static inline void bio_integrity_init(void)
717 #endif /* CONFIG_BLK_DEV_INTEGRITY */
719 #endif /* CONFIG_BLOCK */
720 #endif /* __LINUX_BIO_H */