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));
198 #define bio_io_error(bio) bio_endio((bio), -EIO)
201 * drivers should _never_ use the all version - the bio may have been split
202 * before it got to the driver and the driver won't own all of it
204 #define bio_for_each_segment_all(bvl, bio, i) \
205 for (i = 0, bvl = (bio)->bi_io_vec; i < (bio)->bi_vcnt; i++, bvl++)
207 static inline void bvec_iter_advance(struct bio_vec
*bv
, struct bvec_iter
*iter
,
210 WARN_ONCE(bytes
> iter
->bi_size
,
211 "Attempted to advance past end of bvec iter\n");
214 unsigned len
= min(bytes
, bvec_iter_len(bv
, *iter
));
217 iter
->bi_size
-= len
;
218 iter
->bi_bvec_done
+= len
;
220 if (iter
->bi_bvec_done
== __bvec_iter_bvec(bv
, *iter
)->bv_len
) {
221 iter
->bi_bvec_done
= 0;
227 #define for_each_bvec(bvl, bio_vec, iter, start) \
228 for (iter = (start); \
230 ((bvl = bvec_iter_bvec((bio_vec), (iter))), 1); \
231 bvec_iter_advance((bio_vec), &(iter), (bvl).bv_len))
234 static inline void bio_advance_iter(struct bio
*bio
, struct bvec_iter
*iter
,
237 iter
->bi_sector
+= bytes
>> 9;
239 if (bio
->bi_rw
& BIO_NO_ADVANCE_ITER_MASK
)
240 iter
->bi_size
-= bytes
;
242 bvec_iter_advance(bio
->bi_io_vec
, iter
, bytes
);
245 #define __bio_for_each_segment(bvl, bio, iter, start) \
246 for (iter = (start); \
248 ((bvl = bio_iter_iovec((bio), (iter))), 1); \
249 bio_advance_iter((bio), &(iter), (bvl).bv_len))
251 #define bio_for_each_segment(bvl, bio, iter) \
252 __bio_for_each_segment(bvl, bio, iter, (bio)->bi_iter)
254 #define bio_iter_last(bvec, iter) ((iter).bi_size == (bvec).bv_len)
256 static inline unsigned bio_segments(struct bio
*bio
)
260 struct bvec_iter iter
;
263 * We special case discard/write same, because they interpret bi_size
267 if (bio
->bi_rw
& REQ_DISCARD
)
270 if (bio
->bi_rw
& REQ_WRITE_SAME
)
273 bio_for_each_segment(bv
, bio
, iter
)
280 * get a reference to a bio, so it won't disappear. the intended use is
284 * submit_bio(rw, bio);
285 * if (bio->bi_flags ...)
289 * without the bio_get(), it could potentially complete I/O before submit_bio
290 * returns. and then bio would be freed memory when if (bio->bi_flags ...)
293 #define bio_get(bio) atomic_inc(&(bio)->bi_cnt)
296 BIP_BLOCK_INTEGRITY
= 1 << 0, /* block layer owns integrity data */
297 BIP_MAPPED_INTEGRITY
= 1 << 1, /* ref tag has been remapped */
298 BIP_CTRL_NOCHECK
= 1 << 2, /* disable HBA integrity checking */
299 BIP_DISK_NOCHECK
= 1 << 3, /* disable disk integrity checking */
300 BIP_IP_CHECKSUM
= 1 << 4, /* IP checksum */
303 #if defined(CONFIG_BLK_DEV_INTEGRITY)
305 static inline struct bio_integrity_payload
*bio_integrity(struct bio
*bio
)
307 if (bio
->bi_rw
& REQ_INTEGRITY
)
308 return bio
->bi_integrity
;
314 * bio integrity payload
316 struct bio_integrity_payload
{
317 struct bio
*bip_bio
; /* parent bio */
319 struct bvec_iter bip_iter
;
321 bio_end_io_t
*bip_end_io
; /* saved I/O completion fn */
323 unsigned short bip_slab
; /* slab the bip came from */
324 unsigned short bip_vcnt
; /* # of integrity bio_vecs */
325 unsigned short bip_max_vcnt
; /* integrity bio_vec slots */
326 unsigned short bip_flags
; /* control flags */
328 struct work_struct bip_work
; /* I/O completion */
330 struct bio_vec
*bip_vec
;
331 struct bio_vec bip_inline_vecs
[0];/* embedded bvec array */
334 static inline bool bio_integrity_flagged(struct bio
*bio
, enum bip_flags flag
)
336 struct bio_integrity_payload
*bip
= bio_integrity(bio
);
339 return bip
->bip_flags
& flag
;
344 static inline sector_t
bip_get_seed(struct bio_integrity_payload
*bip
)
346 return bip
->bip_iter
.bi_sector
;
349 static inline void bip_set_seed(struct bio_integrity_payload
*bip
,
352 bip
->bip_iter
.bi_sector
= seed
;
355 #endif /* CONFIG_BLK_DEV_INTEGRITY */
357 extern void bio_trim(struct bio
*bio
, int offset
, int size
);
358 extern struct bio
*bio_split(struct bio
*bio
, int sectors
,
359 gfp_t gfp
, struct bio_set
*bs
);
362 * bio_next_split - get next @sectors from a bio, splitting if necessary
364 * @sectors: number of sectors to split from the front of @bio
366 * @bs: bio set to allocate from
368 * Returns a bio representing the next @sectors of @bio - if the bio is smaller
369 * than @sectors, returns the original bio unchanged.
371 static inline struct bio
*bio_next_split(struct bio
*bio
, int sectors
,
372 gfp_t gfp
, struct bio_set
*bs
)
374 if (sectors
>= bio_sectors(bio
))
377 return bio_split(bio
, sectors
, gfp
, bs
);
380 extern struct bio_set
*bioset_create(unsigned int, unsigned int);
381 extern struct bio_set
*bioset_create_nobvec(unsigned int, unsigned int);
382 extern void bioset_free(struct bio_set
*);
383 extern mempool_t
*biovec_create_pool(int pool_entries
);
385 extern struct bio
*bio_alloc_bioset(gfp_t
, int, struct bio_set
*);
386 extern void bio_put(struct bio
*);
388 extern void __bio_clone_fast(struct bio
*, struct bio
*);
389 extern struct bio
*bio_clone_fast(struct bio
*, gfp_t
, struct bio_set
*);
390 extern struct bio
*bio_clone_bioset(struct bio
*, gfp_t
, struct bio_set
*bs
);
392 extern struct bio_set
*fs_bio_set
;
394 static inline struct bio
*bio_alloc(gfp_t gfp_mask
, unsigned int nr_iovecs
)
396 return bio_alloc_bioset(gfp_mask
, nr_iovecs
, fs_bio_set
);
399 static inline struct bio
*bio_clone(struct bio
*bio
, gfp_t gfp_mask
)
401 return bio_clone_bioset(bio
, gfp_mask
, fs_bio_set
);
404 static inline struct bio
*bio_kmalloc(gfp_t gfp_mask
, unsigned int nr_iovecs
)
406 return bio_alloc_bioset(gfp_mask
, nr_iovecs
, NULL
);
409 static inline struct bio
*bio_clone_kmalloc(struct bio
*bio
, gfp_t gfp_mask
)
411 return bio_clone_bioset(bio
, gfp_mask
, NULL
);
415 extern void bio_endio(struct bio
*, int);
416 extern void bio_endio_nodec(struct bio
*, int);
417 struct request_queue
;
418 extern int bio_phys_segments(struct request_queue
*, struct bio
*);
420 extern int submit_bio_wait(int rw
, struct bio
*bio
);
421 extern void bio_advance(struct bio
*, unsigned);
423 extern void bio_init(struct bio
*);
424 extern void bio_reset(struct bio
*);
425 void bio_chain(struct bio
*, struct bio
*);
427 extern int bio_add_page(struct bio
*, struct page
*, unsigned int,unsigned int);
428 extern int bio_add_pc_page(struct request_queue
*, struct bio
*, struct page
*,
429 unsigned int, unsigned int);
430 extern int bio_get_nr_vecs(struct block_device
*);
431 extern struct bio
*bio_map_user(struct request_queue
*, struct block_device
*,
432 unsigned long, unsigned int, int, gfp_t
);
435 extern struct bio
*bio_map_user_iov(struct request_queue
*,
436 struct block_device
*,
437 const struct sg_iovec
*, int, int, gfp_t
);
438 extern void bio_unmap_user(struct bio
*);
439 extern struct bio
*bio_map_kern(struct request_queue
*, void *, unsigned int,
441 extern struct bio
*bio_copy_kern(struct request_queue
*, void *, unsigned int,
443 extern void bio_set_pages_dirty(struct bio
*bio
);
444 extern void bio_check_pages_dirty(struct bio
*bio
);
446 void generic_start_io_acct(int rw
, unsigned long sectors
,
447 struct hd_struct
*part
);
448 void generic_end_io_acct(int rw
, struct hd_struct
*part
,
449 unsigned long start_time
);
451 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
452 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
454 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
455 extern void bio_flush_dcache_pages(struct bio
*bi
);
457 static inline void bio_flush_dcache_pages(struct bio
*bi
)
462 extern void bio_copy_data(struct bio
*dst
, struct bio
*src
);
463 extern int bio_alloc_pages(struct bio
*bio
, gfp_t gfp
);
465 extern struct bio
*bio_copy_user(struct request_queue
*, struct rq_map_data
*,
466 unsigned long, unsigned int, int, gfp_t
);
467 extern struct bio
*bio_copy_user_iov(struct request_queue
*,
468 struct rq_map_data
*,
469 const struct sg_iovec
*,
471 extern int bio_uncopy_user(struct bio
*);
472 void zero_fill_bio(struct bio
*bio
);
473 extern struct bio_vec
*bvec_alloc(gfp_t
, int, unsigned long *, mempool_t
*);
474 extern void bvec_free(mempool_t
*, struct bio_vec
*, unsigned int);
475 extern unsigned int bvec_nr_vecs(unsigned short idx
);
477 #ifdef CONFIG_BLK_CGROUP
478 int bio_associate_current(struct bio
*bio
);
479 void bio_disassociate_task(struct bio
*bio
);
480 #else /* CONFIG_BLK_CGROUP */
481 static inline int bio_associate_current(struct bio
*bio
) { return -ENOENT
; }
482 static inline void bio_disassociate_task(struct bio
*bio
) { }
483 #endif /* CONFIG_BLK_CGROUP */
485 #ifdef CONFIG_HIGHMEM
487 * remember never ever reenable interrupts between a bvec_kmap_irq and
490 static inline char *bvec_kmap_irq(struct bio_vec
*bvec
, unsigned long *flags
)
495 * might not be a highmem page, but the preempt/irq count
496 * balancing is a lot nicer this way
498 local_irq_save(*flags
);
499 addr
= (unsigned long) kmap_atomic(bvec
->bv_page
);
501 BUG_ON(addr
& ~PAGE_MASK
);
503 return (char *) addr
+ bvec
->bv_offset
;
506 static inline void bvec_kunmap_irq(char *buffer
, unsigned long *flags
)
508 unsigned long ptr
= (unsigned long) buffer
& PAGE_MASK
;
510 kunmap_atomic((void *) ptr
);
511 local_irq_restore(*flags
);
515 static inline char *bvec_kmap_irq(struct bio_vec
*bvec
, unsigned long *flags
)
517 return page_address(bvec
->bv_page
) + bvec
->bv_offset
;
520 static inline void bvec_kunmap_irq(char *buffer
, unsigned long *flags
)
526 static inline char *__bio_kmap_irq(struct bio
*bio
, struct bvec_iter iter
,
527 unsigned long *flags
)
529 return bvec_kmap_irq(&bio_iter_iovec(bio
, iter
), flags
);
531 #define __bio_kunmap_irq(buf, flags) bvec_kunmap_irq(buf, flags)
533 #define bio_kmap_irq(bio, flags) \
534 __bio_kmap_irq((bio), (bio)->bi_iter, (flags))
535 #define bio_kunmap_irq(buf,flags) __bio_kunmap_irq(buf, flags)
538 * BIO list management for use by remapping drivers (e.g. DM or MD) and loop.
540 * A bio_list anchors a singly-linked list of bios chained through the bi_next
541 * member of the bio. The bio_list also caches the last list member to allow
542 * fast access to the tail.
549 static inline int bio_list_empty(const struct bio_list
*bl
)
551 return bl
->head
== NULL
;
554 static inline void bio_list_init(struct bio_list
*bl
)
556 bl
->head
= bl
->tail
= NULL
;
559 #define BIO_EMPTY_LIST { NULL, NULL }
561 #define bio_list_for_each(bio, bl) \
562 for (bio = (bl)->head; bio; bio = bio->bi_next)
564 static inline unsigned bio_list_size(const struct bio_list
*bl
)
569 bio_list_for_each(bio
, bl
)
575 static inline void bio_list_add(struct bio_list
*bl
, struct bio
*bio
)
580 bl
->tail
->bi_next
= bio
;
587 static inline void bio_list_add_head(struct bio_list
*bl
, struct bio
*bio
)
589 bio
->bi_next
= bl
->head
;
597 static inline void bio_list_merge(struct bio_list
*bl
, struct bio_list
*bl2
)
603 bl
->tail
->bi_next
= bl2
->head
;
605 bl
->head
= bl2
->head
;
607 bl
->tail
= bl2
->tail
;
610 static inline void bio_list_merge_head(struct bio_list
*bl
,
611 struct bio_list
*bl2
)
617 bl2
->tail
->bi_next
= bl
->head
;
619 bl
->tail
= bl2
->tail
;
621 bl
->head
= bl2
->head
;
624 static inline struct bio
*bio_list_peek(struct bio_list
*bl
)
629 static inline struct bio
*bio_list_pop(struct bio_list
*bl
)
631 struct bio
*bio
= bl
->head
;
634 bl
->head
= bl
->head
->bi_next
;
644 static inline struct bio
*bio_list_get(struct bio_list
*bl
)
646 struct bio
*bio
= bl
->head
;
648 bl
->head
= bl
->tail
= NULL
;
654 * bio_set is used to allow other portions of the IO system to
655 * allocate their own private memory pools for bio and iovec structures.
656 * These memory pools in turn all allocate from the bio_slab
657 * and the bvec_slabs[].
659 #define BIO_POOL_SIZE 2
660 #define BIOVEC_NR_POOLS 6
661 #define BIOVEC_MAX_IDX (BIOVEC_NR_POOLS - 1)
664 struct kmem_cache
*bio_slab
;
665 unsigned int front_pad
;
668 mempool_t
*bvec_pool
;
669 #if defined(CONFIG_BLK_DEV_INTEGRITY)
670 mempool_t
*bio_integrity_pool
;
671 mempool_t
*bvec_integrity_pool
;
675 * Deadlock avoidance for stacking block drivers: see comments in
676 * bio_alloc_bioset() for details
678 spinlock_t rescue_lock
;
679 struct bio_list rescue_list
;
680 struct work_struct rescue_work
;
681 struct workqueue_struct
*rescue_workqueue
;
687 struct kmem_cache
*slab
;
691 * a small number of entries is fine, not going to be performance critical.
692 * basically we just need to survive
694 #define BIO_SPLIT_ENTRIES 2
696 #if defined(CONFIG_BLK_DEV_INTEGRITY)
698 #define bip_for_each_vec(bvl, bip, iter) \
699 for_each_bvec(bvl, (bip)->bip_vec, iter, (bip)->bip_iter)
701 #define bio_for_each_integrity_vec(_bvl, _bio, _iter) \
703 bip_for_each_vec(_bvl, _bio->bi_integrity, _iter)
705 extern struct bio_integrity_payload
*bio_integrity_alloc(struct bio
*, gfp_t
, unsigned int);
706 extern void bio_integrity_free(struct bio
*);
707 extern int bio_integrity_add_page(struct bio
*, struct page
*, unsigned int, unsigned int);
708 extern bool bio_integrity_enabled(struct bio
*bio
);
709 extern int bio_integrity_prep(struct bio
*);
710 extern void bio_integrity_endio(struct bio
*, int);
711 extern void bio_integrity_advance(struct bio
*, unsigned int);
712 extern void bio_integrity_trim(struct bio
*, unsigned int, unsigned int);
713 extern int bio_integrity_clone(struct bio
*, struct bio
*, gfp_t
);
714 extern int bioset_integrity_create(struct bio_set
*, int);
715 extern void bioset_integrity_free(struct bio_set
*);
716 extern void bio_integrity_init(void);
718 #else /* CONFIG_BLK_DEV_INTEGRITY */
720 static inline void *bio_integrity(struct bio
*bio
)
725 static inline bool bio_integrity_enabled(struct bio
*bio
)
730 static inline int bioset_integrity_create(struct bio_set
*bs
, int pool_size
)
735 static inline void bioset_integrity_free (struct bio_set
*bs
)
740 static inline int bio_integrity_prep(struct bio
*bio
)
745 static inline void bio_integrity_free(struct bio
*bio
)
750 static inline int bio_integrity_clone(struct bio
*bio
, struct bio
*bio_src
,
756 static inline void bio_integrity_advance(struct bio
*bio
,
757 unsigned int bytes_done
)
762 static inline void bio_integrity_trim(struct bio
*bio
, unsigned int offset
,
763 unsigned int sectors
)
768 static inline void bio_integrity_init(void)
773 static inline bool bio_integrity_flagged(struct bio
*bio
, enum bip_flags flag
)
778 #endif /* CONFIG_BLK_DEV_INTEGRITY */
780 #endif /* CONFIG_BLOCK */
781 #endif /* __LINUX_BIO_H */
This page took 0.064757 seconds and 6 git commands to generate.