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_idx)
66 #define bio_page(bio) bio_iovec((bio))->bv_page
67 #define bio_offset(bio) bio_iovec((bio))->bv_offset
68 #define bio_segments(bio) ((bio)->bi_vcnt - (bio)->bi_idx)
69 #define bio_sectors(bio) ((bio)->bi_size >> 9)
70 #define bio_end_sector(bio) ((bio)->bi_sector + bio_sectors((bio)))
72 static inline unsigned int bio_cur_bytes(struct bio
*bio
)
75 return bio_iovec(bio
)->bv_len
;
76 else /* dataless requests such as discard */
80 static inline void *bio_data(struct bio
*bio
)
83 return page_address(bio_page(bio
)) + bio_offset(bio
);
88 static inline int bio_has_allocated_vec(struct bio
*bio
)
90 return bio
->bi_io_vec
&& bio
->bi_io_vec
!= bio
->bi_inline_vecs
;
96 #define bio_to_phys(bio) (page_to_phys(bio_page((bio))) + (unsigned long) bio_offset((bio)))
97 #define bvec_to_phys(bv) (page_to_phys((bv)->bv_page) + (unsigned long) (bv)->bv_offset)
100 * queues that have highmem support enabled may still need to revert to
101 * PIO transfers occasionally and thus map high pages temporarily. For
102 * permanent PIO fall back, user is probably better off disabling highmem
103 * I/O completely on that queue (see ide-dma for example)
105 #define __bio_kmap_atomic(bio, idx, kmtype) \
106 (kmap_atomic(bio_iovec_idx((bio), (idx))->bv_page) + \
107 bio_iovec_idx((bio), (idx))->bv_offset)
109 #define __bio_kunmap_atomic(addr, kmtype) kunmap_atomic(addr)
115 #define __BVEC_END(bio) bio_iovec_idx((bio), (bio)->bi_vcnt - 1)
116 #define __BVEC_START(bio) bio_iovec_idx((bio), (bio)->bi_idx)
118 /* Default implementation of BIOVEC_PHYS_MERGEABLE */
119 #define __BIOVEC_PHYS_MERGEABLE(vec1, vec2) \
120 ((bvec_to_phys((vec1)) + (vec1)->bv_len) == bvec_to_phys((vec2)))
123 * allow arch override, for eg virtualized architectures (put in asm/io.h)
125 #ifndef BIOVEC_PHYS_MERGEABLE
126 #define BIOVEC_PHYS_MERGEABLE(vec1, vec2) \
127 __BIOVEC_PHYS_MERGEABLE(vec1, vec2)
130 #define __BIO_SEG_BOUNDARY(addr1, addr2, mask) \
131 (((addr1) | (mask)) == (((addr2) - 1) | (mask)))
132 #define BIOVEC_SEG_BOUNDARY(q, b1, b2) \
133 __BIO_SEG_BOUNDARY(bvec_to_phys((b1)), bvec_to_phys((b2)) + (b2)->bv_len, queue_segment_boundary((q)))
134 #define BIO_SEG_BOUNDARY(q, b1, b2) \
135 BIOVEC_SEG_BOUNDARY((q), __BVEC_END((b1)), __BVEC_START((b2)))
137 #define bio_io_error(bio) bio_endio((bio), -EIO)
140 * drivers should not use the __ version unless they _really_ know what
143 #define __bio_for_each_segment(bvl, bio, i, start_idx) \
144 for (bvl = bio_iovec_idx((bio), (start_idx)), i = (start_idx); \
145 i < (bio)->bi_vcnt; \
149 * drivers should _never_ use the all version - the bio may have been split
150 * before it got to the driver and the driver won't own all of it
152 #define bio_for_each_segment_all(bvl, bio, i) \
154 bvl = bio_iovec_idx((bio), (i)), i < (bio)->bi_vcnt; \
157 #define bio_for_each_segment(bvl, bio, i) \
158 for (i = (bio)->bi_idx; \
159 bvl = bio_iovec_idx((bio), (i)), i < (bio)->bi_vcnt; \
163 * get a reference to a bio, so it won't disappear. the intended use is
167 * submit_bio(rw, bio);
168 * if (bio->bi_flags ...)
172 * without the bio_get(), it could potentially complete I/O before submit_bio
173 * returns. and then bio would be freed memory when if (bio->bi_flags ...)
176 #define bio_get(bio) atomic_inc(&(bio)->bi_cnt)
178 #if defined(CONFIG_BLK_DEV_INTEGRITY)
180 * bio integrity payload
182 struct bio_integrity_payload
{
183 struct bio
*bip_bio
; /* parent bio */
185 sector_t bip_sector
; /* virtual start sector */
187 void *bip_buf
; /* generated integrity data */
188 bio_end_io_t
*bip_end_io
; /* saved I/O completion fn */
190 unsigned int bip_size
;
192 unsigned short bip_slab
; /* slab the bip came from */
193 unsigned short bip_vcnt
; /* # of integrity bio_vecs */
194 unsigned short bip_idx
; /* current bip_vec index */
196 struct work_struct bip_work
; /* I/O completion */
198 struct bio_vec
*bip_vec
;
199 struct bio_vec bip_inline_vecs
[0];/* embedded bvec array */
201 #endif /* CONFIG_BLK_DEV_INTEGRITY */
204 * A bio_pair is used when we need to split a bio.
205 * This can only happen for a bio that refers to just one
206 * page of data, and in the unusual situation when the
207 * page crosses a chunk/device boundary
209 * The address of the master bio is stored in bio1.bi_private
210 * The address of the pool the pair was allocated from is stored
214 struct bio bio1
, bio2
;
215 struct bio_vec bv1
, bv2
;
216 #if defined(CONFIG_BLK_DEV_INTEGRITY)
217 struct bio_integrity_payload bip1
, bip2
;
218 struct bio_vec iv1
, iv2
;
223 extern struct bio_pair
*bio_split(struct bio
*bi
, int first_sectors
);
224 extern void bio_pair_release(struct bio_pair
*dbio
);
226 extern struct bio_set
*bioset_create(unsigned int, unsigned int);
227 extern void bioset_free(struct bio_set
*);
228 extern mempool_t
*biovec_create_pool(struct bio_set
*bs
, int pool_entries
);
230 extern struct bio
*bio_alloc_bioset(gfp_t
, int, struct bio_set
*);
231 extern void bio_put(struct bio
*);
233 extern void __bio_clone(struct bio
*, struct bio
*);
234 extern struct bio
*bio_clone_bioset(struct bio
*, gfp_t
, struct bio_set
*bs
);
236 extern struct bio_set
*fs_bio_set
;
238 static inline struct bio
*bio_alloc(gfp_t gfp_mask
, unsigned int nr_iovecs
)
240 return bio_alloc_bioset(gfp_mask
, nr_iovecs
, fs_bio_set
);
243 static inline struct bio
*bio_clone(struct bio
*bio
, gfp_t gfp_mask
)
245 return bio_clone_bioset(bio
, gfp_mask
, fs_bio_set
);
248 static inline struct bio
*bio_kmalloc(gfp_t gfp_mask
, unsigned int nr_iovecs
)
250 return bio_alloc_bioset(gfp_mask
, nr_iovecs
, NULL
);
253 static inline struct bio
*bio_clone_kmalloc(struct bio
*bio
, gfp_t gfp_mask
)
255 return bio_clone_bioset(bio
, gfp_mask
, NULL
);
259 extern void bio_endio(struct bio
*, int);
260 struct request_queue
;
261 extern int bio_phys_segments(struct request_queue
*, struct bio
*);
263 extern int submit_bio_wait(int rw
, struct bio
*bio
);
264 extern void bio_advance(struct bio
*, unsigned);
266 extern void bio_init(struct bio
*);
267 extern void bio_reset(struct bio
*);
269 extern int bio_add_page(struct bio
*, struct page
*, unsigned int,unsigned int);
270 extern int bio_add_pc_page(struct request_queue
*, struct bio
*, struct page
*,
271 unsigned int, unsigned int);
272 extern int bio_get_nr_vecs(struct block_device
*);
273 extern sector_t
bio_sector_offset(struct bio
*, unsigned short, unsigned int);
274 extern struct bio
*bio_map_user(struct request_queue
*, struct block_device
*,
275 unsigned long, unsigned int, int, gfp_t
);
278 extern struct bio
*bio_map_user_iov(struct request_queue
*,
279 struct block_device
*,
280 struct sg_iovec
*, int, int, gfp_t
);
281 extern void bio_unmap_user(struct bio
*);
282 extern struct bio
*bio_map_kern(struct request_queue
*, void *, unsigned int,
284 extern struct bio
*bio_copy_kern(struct request_queue
*, void *, unsigned int,
286 extern void bio_set_pages_dirty(struct bio
*bio
);
287 extern void bio_check_pages_dirty(struct bio
*bio
);
289 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
290 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
292 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
293 extern void bio_flush_dcache_pages(struct bio
*bi
);
295 static inline void bio_flush_dcache_pages(struct bio
*bi
)
300 extern void bio_copy_data(struct bio
*dst
, struct bio
*src
);
301 extern int bio_alloc_pages(struct bio
*bio
, gfp_t gfp
);
303 extern struct bio
*bio_copy_user(struct request_queue
*, struct rq_map_data
*,
304 unsigned long, unsigned int, int, gfp_t
);
305 extern struct bio
*bio_copy_user_iov(struct request_queue
*,
306 struct rq_map_data
*, struct sg_iovec
*,
308 extern int bio_uncopy_user(struct bio
*);
309 void zero_fill_bio(struct bio
*bio
);
310 extern struct bio_vec
*bvec_alloc(gfp_t
, int, unsigned long *, mempool_t
*);
311 extern void bvec_free(mempool_t
*, struct bio_vec
*, unsigned int);
312 extern unsigned int bvec_nr_vecs(unsigned short idx
);
314 #ifdef CONFIG_BLK_CGROUP
315 int bio_associate_current(struct bio
*bio
);
316 void bio_disassociate_task(struct bio
*bio
);
317 #else /* CONFIG_BLK_CGROUP */
318 static inline int bio_associate_current(struct bio
*bio
) { return -ENOENT
; }
319 static inline void bio_disassociate_task(struct bio
*bio
) { }
320 #endif /* CONFIG_BLK_CGROUP */
322 #ifdef CONFIG_HIGHMEM
324 * remember never ever reenable interrupts between a bvec_kmap_irq and
327 static inline char *bvec_kmap_irq(struct bio_vec
*bvec
, unsigned long *flags
)
332 * might not be a highmem page, but the preempt/irq count
333 * balancing is a lot nicer this way
335 local_irq_save(*flags
);
336 addr
= (unsigned long) kmap_atomic(bvec
->bv_page
);
338 BUG_ON(addr
& ~PAGE_MASK
);
340 return (char *) addr
+ bvec
->bv_offset
;
343 static inline void bvec_kunmap_irq(char *buffer
, unsigned long *flags
)
345 unsigned long ptr
= (unsigned long) buffer
& PAGE_MASK
;
347 kunmap_atomic((void *) ptr
);
348 local_irq_restore(*flags
);
352 static inline char *bvec_kmap_irq(struct bio_vec
*bvec
, unsigned long *flags
)
354 return page_address(bvec
->bv_page
) + bvec
->bv_offset
;
357 static inline void bvec_kunmap_irq(char *buffer
, unsigned long *flags
)
363 static inline char *__bio_kmap_irq(struct bio
*bio
, unsigned short idx
,
364 unsigned long *flags
)
366 return bvec_kmap_irq(bio_iovec_idx(bio
, idx
), flags
);
368 #define __bio_kunmap_irq(buf, flags) bvec_kunmap_irq(buf, flags)
370 #define bio_kmap_irq(bio, flags) \
371 __bio_kmap_irq((bio), (bio)->bi_idx, (flags))
372 #define bio_kunmap_irq(buf,flags) __bio_kunmap_irq(buf, flags)
375 * Check whether this bio carries any data or not. A NULL bio is allowed.
377 static inline bool bio_has_data(struct bio
*bio
)
379 if (bio
&& bio
->bi_vcnt
)
385 static inline bool bio_is_rw(struct bio
*bio
)
387 if (!bio_has_data(bio
))
390 if (bio
->bi_rw
& REQ_WRITE_SAME
)
396 static inline bool bio_mergeable(struct bio
*bio
)
398 if (bio
->bi_rw
& REQ_NOMERGE_FLAGS
)
405 * BIO list management for use by remapping drivers (e.g. DM or MD) and loop.
407 * A bio_list anchors a singly-linked list of bios chained through the bi_next
408 * member of the bio. The bio_list also caches the last list member to allow
409 * fast access to the tail.
416 static inline int bio_list_empty(const struct bio_list
*bl
)
418 return bl
->head
== NULL
;
421 static inline void bio_list_init(struct bio_list
*bl
)
423 bl
->head
= bl
->tail
= NULL
;
426 #define bio_list_for_each(bio, bl) \
427 for (bio = (bl)->head; bio; bio = bio->bi_next)
429 static inline unsigned bio_list_size(const struct bio_list
*bl
)
434 bio_list_for_each(bio
, bl
)
440 static inline void bio_list_add(struct bio_list
*bl
, struct bio
*bio
)
445 bl
->tail
->bi_next
= bio
;
452 static inline void bio_list_add_head(struct bio_list
*bl
, struct bio
*bio
)
454 bio
->bi_next
= bl
->head
;
462 static inline void bio_list_merge(struct bio_list
*bl
, struct bio_list
*bl2
)
468 bl
->tail
->bi_next
= bl2
->head
;
470 bl
->head
= bl2
->head
;
472 bl
->tail
= bl2
->tail
;
475 static inline void bio_list_merge_head(struct bio_list
*bl
,
476 struct bio_list
*bl2
)
482 bl2
->tail
->bi_next
= bl
->head
;
484 bl
->tail
= bl2
->tail
;
486 bl
->head
= bl2
->head
;
489 static inline struct bio
*bio_list_peek(struct bio_list
*bl
)
494 static inline struct bio
*bio_list_pop(struct bio_list
*bl
)
496 struct bio
*bio
= bl
->head
;
499 bl
->head
= bl
->head
->bi_next
;
509 static inline struct bio
*bio_list_get(struct bio_list
*bl
)
511 struct bio
*bio
= bl
->head
;
513 bl
->head
= bl
->tail
= NULL
;
519 * bio_set is used to allow other portions of the IO system to
520 * allocate their own private memory pools for bio and iovec structures.
521 * These memory pools in turn all allocate from the bio_slab
522 * and the bvec_slabs[].
524 #define BIO_POOL_SIZE 2
525 #define BIOVEC_NR_POOLS 6
526 #define BIOVEC_MAX_IDX (BIOVEC_NR_POOLS - 1)
529 struct kmem_cache
*bio_slab
;
530 unsigned int front_pad
;
533 mempool_t
*bvec_pool
;
534 #if defined(CONFIG_BLK_DEV_INTEGRITY)
535 mempool_t
*bio_integrity_pool
;
536 mempool_t
*bvec_integrity_pool
;
540 * Deadlock avoidance for stacking block drivers: see comments in
541 * bio_alloc_bioset() for details
543 spinlock_t rescue_lock
;
544 struct bio_list rescue_list
;
545 struct work_struct rescue_work
;
546 struct workqueue_struct
*rescue_workqueue
;
552 struct kmem_cache
*slab
;
556 * a small number of entries is fine, not going to be performance critical.
557 * basically we just need to survive
559 #define BIO_SPLIT_ENTRIES 2
561 #if defined(CONFIG_BLK_DEV_INTEGRITY)
563 #define bip_vec_idx(bip, idx) (&(bip->bip_vec[(idx)]))
564 #define bip_vec(bip) bip_vec_idx(bip, 0)
566 #define __bip_for_each_vec(bvl, bip, i, start_idx) \
567 for (bvl = bip_vec_idx((bip), (start_idx)), i = (start_idx); \
568 i < (bip)->bip_vcnt; \
571 #define bip_for_each_vec(bvl, bip, i) \
572 __bip_for_each_vec(bvl, bip, i, (bip)->bip_idx)
574 #define bio_for_each_integrity_vec(_bvl, _bio, _iter) \
576 bip_for_each_vec(_bvl, _bio->bi_integrity, _iter)
578 #define bio_integrity(bio) (bio->bi_integrity != NULL)
580 extern struct bio_integrity_payload
*bio_integrity_alloc(struct bio
*, gfp_t
, unsigned int);
581 extern void bio_integrity_free(struct bio
*);
582 extern int bio_integrity_add_page(struct bio
*, struct page
*, unsigned int, unsigned int);
583 extern int bio_integrity_enabled(struct bio
*bio
);
584 extern int bio_integrity_set_tag(struct bio
*, void *, unsigned int);
585 extern int bio_integrity_get_tag(struct bio
*, void *, unsigned int);
586 extern int bio_integrity_prep(struct bio
*);
587 extern void bio_integrity_endio(struct bio
*, int);
588 extern void bio_integrity_advance(struct bio
*, unsigned int);
589 extern void bio_integrity_trim(struct bio
*, unsigned int, unsigned int);
590 extern void bio_integrity_split(struct bio
*, struct bio_pair
*, int);
591 extern int bio_integrity_clone(struct bio
*, struct bio
*, gfp_t
);
592 extern int bioset_integrity_create(struct bio_set
*, int);
593 extern void bioset_integrity_free(struct bio_set
*);
594 extern void bio_integrity_init(void);
596 #else /* CONFIG_BLK_DEV_INTEGRITY */
598 static inline int bio_integrity(struct bio
*bio
)
603 static inline int bio_integrity_enabled(struct bio
*bio
)
608 static inline int bioset_integrity_create(struct bio_set
*bs
, int pool_size
)
613 static inline void bioset_integrity_free (struct bio_set
*bs
)
618 static inline int bio_integrity_prep(struct bio
*bio
)
623 static inline void bio_integrity_free(struct bio
*bio
)
628 static inline int bio_integrity_clone(struct bio
*bio
, struct bio
*bio_src
,
634 static inline void bio_integrity_split(struct bio
*bio
, struct bio_pair
*bp
,
640 static inline void bio_integrity_advance(struct bio
*bio
,
641 unsigned int bytes_done
)
646 static inline void bio_integrity_trim(struct bio
*bio
, unsigned int offset
,
647 unsigned int sectors
)
652 static inline void bio_integrity_init(void)
657 #endif /* CONFIG_BLK_DEV_INTEGRITY */
659 #endif /* CONFIG_BLOCK */
660 #endif /* __LINUX_BIO_H */