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 */
301 * A bio_pair is used when we need to split a bio.
302 * This can only happen for a bio that refers to just one
303 * page of data, and in the unusual situation when the
304 * page crosses a chunk/device boundary
306 * The address of the master bio is stored in bio1.bi_private
307 * The address of the pool the pair was allocated from is stored
311 struct bio bio1
, bio2
;
312 struct bio_vec bv1
, bv2
;
313 #if defined(CONFIG_BLK_DEV_INTEGRITY)
314 struct bio_integrity_payload bip1
, bip2
;
315 struct bio_vec iv1
, iv2
;
320 extern struct bio_pair
*bio_split(struct bio
*bi
, int first_sectors
);
321 extern void bio_pair_release(struct bio_pair
*dbio
);
322 extern void bio_trim(struct bio
*bio
, int offset
, int size
);
324 extern struct bio_set
*bioset_create(unsigned int, unsigned int);
325 extern void bioset_free(struct bio_set
*);
326 extern mempool_t
*biovec_create_pool(struct bio_set
*bs
, int pool_entries
);
328 extern struct bio
*bio_alloc_bioset(gfp_t
, int, struct bio_set
*);
329 extern void bio_put(struct bio
*);
331 extern void __bio_clone_fast(struct bio
*, struct bio
*);
332 extern struct bio
*bio_clone_fast(struct bio
*, gfp_t
, struct bio_set
*);
333 extern struct bio
*bio_clone_bioset(struct bio
*, gfp_t
, struct bio_set
*bs
);
335 extern struct bio_set
*fs_bio_set
;
337 static inline struct bio
*bio_alloc(gfp_t gfp_mask
, unsigned int nr_iovecs
)
339 return bio_alloc_bioset(gfp_mask
, nr_iovecs
, fs_bio_set
);
342 static inline struct bio
*bio_clone(struct bio
*bio
, gfp_t gfp_mask
)
344 return bio_clone_bioset(bio
, gfp_mask
, fs_bio_set
);
347 static inline struct bio
*bio_kmalloc(gfp_t gfp_mask
, unsigned int nr_iovecs
)
349 return bio_alloc_bioset(gfp_mask
, nr_iovecs
, NULL
);
352 static inline struct bio
*bio_clone_kmalloc(struct bio
*bio
, gfp_t gfp_mask
)
354 return bio_clone_bioset(bio
, gfp_mask
, NULL
);
358 extern void bio_endio(struct bio
*, int);
359 extern void bio_endio_nodec(struct bio
*, int);
360 struct request_queue
;
361 extern int bio_phys_segments(struct request_queue
*, struct bio
*);
363 extern int submit_bio_wait(int rw
, struct bio
*bio
);
364 extern void bio_advance(struct bio
*, unsigned);
366 extern void bio_init(struct bio
*);
367 extern void bio_reset(struct bio
*);
368 void bio_chain(struct bio
*, struct bio
*);
370 extern int bio_add_page(struct bio
*, struct page
*, unsigned int,unsigned int);
371 extern int bio_add_pc_page(struct request_queue
*, struct bio
*, struct page
*,
372 unsigned int, unsigned int);
373 extern int bio_get_nr_vecs(struct block_device
*);
374 extern struct bio
*bio_map_user(struct request_queue
*, struct block_device
*,
375 unsigned long, unsigned int, int, gfp_t
);
378 extern struct bio
*bio_map_user_iov(struct request_queue
*,
379 struct block_device
*,
380 struct sg_iovec
*, int, int, gfp_t
);
381 extern void bio_unmap_user(struct bio
*);
382 extern struct bio
*bio_map_kern(struct request_queue
*, void *, unsigned int,
384 extern struct bio
*bio_copy_kern(struct request_queue
*, void *, unsigned int,
386 extern void bio_set_pages_dirty(struct bio
*bio
);
387 extern void bio_check_pages_dirty(struct bio
*bio
);
389 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
390 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
392 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
393 extern void bio_flush_dcache_pages(struct bio
*bi
);
395 static inline void bio_flush_dcache_pages(struct bio
*bi
)
400 extern void bio_copy_data(struct bio
*dst
, struct bio
*src
);
401 extern int bio_alloc_pages(struct bio
*bio
, gfp_t gfp
);
403 extern struct bio
*bio_copy_user(struct request_queue
*, struct rq_map_data
*,
404 unsigned long, unsigned int, int, gfp_t
);
405 extern struct bio
*bio_copy_user_iov(struct request_queue
*,
406 struct rq_map_data
*, struct sg_iovec
*,
408 extern int bio_uncopy_user(struct bio
*);
409 void zero_fill_bio(struct bio
*bio
);
410 extern struct bio_vec
*bvec_alloc(gfp_t
, int, unsigned long *, mempool_t
*);
411 extern void bvec_free(mempool_t
*, struct bio_vec
*, unsigned int);
412 extern unsigned int bvec_nr_vecs(unsigned short idx
);
414 #ifdef CONFIG_BLK_CGROUP
415 int bio_associate_current(struct bio
*bio
);
416 void bio_disassociate_task(struct bio
*bio
);
417 #else /* CONFIG_BLK_CGROUP */
418 static inline int bio_associate_current(struct bio
*bio
) { return -ENOENT
; }
419 static inline void bio_disassociate_task(struct bio
*bio
) { }
420 #endif /* CONFIG_BLK_CGROUP */
422 #ifdef CONFIG_HIGHMEM
424 * remember never ever reenable interrupts between a bvec_kmap_irq and
427 static inline char *bvec_kmap_irq(struct bio_vec
*bvec
, unsigned long *flags
)
432 * might not be a highmem page, but the preempt/irq count
433 * balancing is a lot nicer this way
435 local_irq_save(*flags
);
436 addr
= (unsigned long) kmap_atomic(bvec
->bv_page
);
438 BUG_ON(addr
& ~PAGE_MASK
);
440 return (char *) addr
+ bvec
->bv_offset
;
443 static inline void bvec_kunmap_irq(char *buffer
, unsigned long *flags
)
445 unsigned long ptr
= (unsigned long) buffer
& PAGE_MASK
;
447 kunmap_atomic((void *) ptr
);
448 local_irq_restore(*flags
);
452 static inline char *bvec_kmap_irq(struct bio_vec
*bvec
, unsigned long *flags
)
454 return page_address(bvec
->bv_page
) + bvec
->bv_offset
;
457 static inline void bvec_kunmap_irq(char *buffer
, unsigned long *flags
)
463 static inline char *__bio_kmap_irq(struct bio
*bio
, struct bvec_iter iter
,
464 unsigned long *flags
)
466 return bvec_kmap_irq(&bio_iter_iovec(bio
, iter
), flags
);
468 #define __bio_kunmap_irq(buf, flags) bvec_kunmap_irq(buf, flags)
470 #define bio_kmap_irq(bio, flags) \
471 __bio_kmap_irq((bio), (bio)->bi_iter, (flags))
472 #define bio_kunmap_irq(buf,flags) __bio_kunmap_irq(buf, flags)
475 * BIO list management for use by remapping drivers (e.g. DM or MD) and loop.
477 * A bio_list anchors a singly-linked list of bios chained through the bi_next
478 * member of the bio. The bio_list also caches the last list member to allow
479 * fast access to the tail.
486 static inline int bio_list_empty(const struct bio_list
*bl
)
488 return bl
->head
== NULL
;
491 static inline void bio_list_init(struct bio_list
*bl
)
493 bl
->head
= bl
->tail
= NULL
;
496 #define BIO_EMPTY_LIST { NULL, NULL }
498 #define bio_list_for_each(bio, bl) \
499 for (bio = (bl)->head; bio; bio = bio->bi_next)
501 static inline unsigned bio_list_size(const struct bio_list
*bl
)
506 bio_list_for_each(bio
, bl
)
512 static inline void bio_list_add(struct bio_list
*bl
, struct bio
*bio
)
517 bl
->tail
->bi_next
= bio
;
524 static inline void bio_list_add_head(struct bio_list
*bl
, struct bio
*bio
)
526 bio
->bi_next
= bl
->head
;
534 static inline void bio_list_merge(struct bio_list
*bl
, struct bio_list
*bl2
)
540 bl
->tail
->bi_next
= bl2
->head
;
542 bl
->head
= bl2
->head
;
544 bl
->tail
= bl2
->tail
;
547 static inline void bio_list_merge_head(struct bio_list
*bl
,
548 struct bio_list
*bl2
)
554 bl2
->tail
->bi_next
= bl
->head
;
556 bl
->tail
= bl2
->tail
;
558 bl
->head
= bl2
->head
;
561 static inline struct bio
*bio_list_peek(struct bio_list
*bl
)
566 static inline struct bio
*bio_list_pop(struct bio_list
*bl
)
568 struct bio
*bio
= bl
->head
;
571 bl
->head
= bl
->head
->bi_next
;
581 static inline struct bio
*bio_list_get(struct bio_list
*bl
)
583 struct bio
*bio
= bl
->head
;
585 bl
->head
= bl
->tail
= NULL
;
591 * bio_set is used to allow other portions of the IO system to
592 * allocate their own private memory pools for bio and iovec structures.
593 * These memory pools in turn all allocate from the bio_slab
594 * and the bvec_slabs[].
596 #define BIO_POOL_SIZE 2
597 #define BIOVEC_NR_POOLS 6
598 #define BIOVEC_MAX_IDX (BIOVEC_NR_POOLS - 1)
601 struct kmem_cache
*bio_slab
;
602 unsigned int front_pad
;
605 mempool_t
*bvec_pool
;
606 #if defined(CONFIG_BLK_DEV_INTEGRITY)
607 mempool_t
*bio_integrity_pool
;
608 mempool_t
*bvec_integrity_pool
;
612 * Deadlock avoidance for stacking block drivers: see comments in
613 * bio_alloc_bioset() for details
615 spinlock_t rescue_lock
;
616 struct bio_list rescue_list
;
617 struct work_struct rescue_work
;
618 struct workqueue_struct
*rescue_workqueue
;
624 struct kmem_cache
*slab
;
628 * a small number of entries is fine, not going to be performance critical.
629 * basically we just need to survive
631 #define BIO_SPLIT_ENTRIES 2
633 #if defined(CONFIG_BLK_DEV_INTEGRITY)
637 #define bip_vec_idx(bip, idx) (&(bip->bip_vec[(idx)]))
639 #define bip_for_each_vec(bvl, bip, iter) \
640 for_each_bvec(bvl, (bip)->bip_vec, iter, (bip)->bip_iter)
642 #define bio_for_each_integrity_vec(_bvl, _bio, _iter) \
644 bip_for_each_vec(_bvl, _bio->bi_integrity, _iter)
646 #define bio_integrity(bio) (bio->bi_integrity != NULL)
648 extern struct bio_integrity_payload
*bio_integrity_alloc(struct bio
*, gfp_t
, unsigned int);
649 extern void bio_integrity_free(struct bio
*);
650 extern int bio_integrity_add_page(struct bio
*, struct page
*, unsigned int, unsigned int);
651 extern int bio_integrity_enabled(struct bio
*bio
);
652 extern int bio_integrity_set_tag(struct bio
*, void *, unsigned int);
653 extern int bio_integrity_get_tag(struct bio
*, void *, unsigned int);
654 extern int bio_integrity_prep(struct bio
*);
655 extern void bio_integrity_endio(struct bio
*, int);
656 extern void bio_integrity_advance(struct bio
*, unsigned int);
657 extern void bio_integrity_trim(struct bio
*, unsigned int, unsigned int);
658 extern void bio_integrity_split(struct bio
*, struct bio_pair
*, int);
659 extern int bio_integrity_clone(struct bio
*, struct bio
*, gfp_t
);
660 extern int bioset_integrity_create(struct bio_set
*, int);
661 extern void bioset_integrity_free(struct bio_set
*);
662 extern void bio_integrity_init(void);
664 #else /* CONFIG_BLK_DEV_INTEGRITY */
666 static inline int bio_integrity(struct bio
*bio
)
671 static inline int bio_integrity_enabled(struct bio
*bio
)
676 static inline int bioset_integrity_create(struct bio_set
*bs
, int pool_size
)
681 static inline void bioset_integrity_free (struct bio_set
*bs
)
686 static inline int bio_integrity_prep(struct bio
*bio
)
691 static inline void bio_integrity_free(struct bio
*bio
)
696 static inline int bio_integrity_clone(struct bio
*bio
, struct bio
*bio_src
,
702 static inline void bio_integrity_split(struct bio
*bio
, struct bio_pair
*bp
,
708 static inline void bio_integrity_advance(struct bio
*bio
,
709 unsigned int bytes_done
)
714 static inline void bio_integrity_trim(struct bio
*bio
, unsigned int offset
,
715 unsigned int sectors
)
720 static inline void bio_integrity_init(void)
725 #endif /* CONFIG_BLK_DEV_INTEGRITY */
727 #endif /* CONFIG_BLOCK */
728 #endif /* __LINUX_BIO_H */