blob: 7b214fd672a2d923e56d8263d8c183d7dcf98b0e [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001/*
2 * 2.5 block I/O model
3 *
4 * Copyright (C) 2001 Jens Axboe <axboe@suse.de>
5 *
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.
9 *
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 * GNU General Public License for more details.
15 *
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-
19 */
20#ifndef __LINUX_BIO_H
21#define __LINUX_BIO_H
22
23#include <linux/highmem.h>
24#include <linux/mempool.h>
Jens Axboe22e2c502005-06-27 10:55:12 +020025#include <linux/ioprio.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070026
David Howells02a5e0a2007-08-11 22:34:32 +020027#ifdef CONFIG_BLOCK
28
Linus Torvalds1da177e2005-04-16 15:20:36 -070029#include <asm/io.h>
30
Linus Torvalds1da177e2005-04-16 15:20:36 -070031#define BIO_DEBUG
32
33#ifdef BIO_DEBUG
34#define BIO_BUG_ON BUG_ON
35#else
36#define BIO_BUG_ON
37#endif
38
Alexey Dobriyand84a8472006-06-25 05:49:32 -070039#define BIO_MAX_PAGES 256
Linus Torvalds1da177e2005-04-16 15:20:36 -070040#define BIO_MAX_SIZE (BIO_MAX_PAGES << PAGE_CACHE_SHIFT)
41#define BIO_MAX_SECTORS (BIO_MAX_SIZE >> 9)
42
43/*
44 * was unsigned short, but we might as well be ready for > 64kB I/O pages
45 */
46struct bio_vec {
47 struct page *bv_page;
48 unsigned int bv_len;
49 unsigned int bv_offset;
50};
51
52struct bio_set;
53struct bio;
Martin K. Petersen7ba1ba12008-06-30 20:04:41 +020054struct bio_integrity_payload;
NeilBrown6712ecf2007-09-27 12:47:43 +020055typedef void (bio_end_io_t) (struct bio *, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -070056typedef void (bio_destructor_t) (struct bio *);
57
58/*
59 * main unit of I/O for the block layer and lower layers (ie drivers and
60 * stacking drivers)
61 */
62struct bio {
Roger Gammans2c2345c2006-10-04 13:37:45 +020063 sector_t bi_sector; /* device address in 512 byte
64 sectors */
Linus Torvalds1da177e2005-04-16 15:20:36 -070065 struct bio *bi_next; /* request queue link */
66 struct block_device *bi_bdev;
67 unsigned long bi_flags; /* status, command, etc */
68 unsigned long bi_rw; /* bottom bits READ/WRITE,
69 * top bits priority
70 */
71
72 unsigned short bi_vcnt; /* how many bio_vec's */
73 unsigned short bi_idx; /* current index into bvl_vec */
74
75 /* Number of segments in this BIO after
76 * physical address coalescing is performed.
77 */
Jens Axboe5b99c2f2008-08-15 10:56:11 +020078 unsigned int bi_phys_segments;
Linus Torvalds1da177e2005-04-16 15:20:36 -070079
Linus Torvalds1da177e2005-04-16 15:20:36 -070080 unsigned int bi_size; /* residual I/O count */
81
FUJITA Tomonori86771422008-10-13 14:19:05 +020082 /*
83 * To keep track of the max segment size, we account for the
84 * sizes of the first and last mergeable segments in this bio.
85 */
86 unsigned int bi_seg_front_size;
87 unsigned int bi_seg_back_size;
88
Linus Torvalds1da177e2005-04-16 15:20:36 -070089 unsigned int bi_max_vecs; /* max bvl_vecs we can hold */
90
Jens Axboec7c22e42008-09-13 20:26:01 +020091 unsigned int bi_comp_cpu; /* completion CPU */
92
Richard Kennedyba744d52008-12-03 12:41:40 +010093 atomic_t bi_cnt; /* pin count */
94
Linus Torvalds1da177e2005-04-16 15:20:36 -070095 struct bio_vec *bi_io_vec; /* the actual vec list */
96
97 bio_end_io_t *bi_end_io;
Linus Torvalds1da177e2005-04-16 15:20:36 -070098
99 void *bi_private;
Martin K. Petersen7ba1ba12008-06-30 20:04:41 +0200100#if defined(CONFIG_BLK_DEV_INTEGRITY)
101 struct bio_integrity_payload *bi_integrity; /* data integrity */
102#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700103
104 bio_destructor_t *bi_destructor; /* destructor */
Jens Axboe392ddc32008-12-23 12:42:54 +0100105
106 /*
107 * We can inline a number of vecs at the end of the bio, to avoid
108 * double allocations for a small number of bio_vecs. This member
109 * MUST obviously be kept at the very end of the bio.
110 */
111 struct bio_vec bi_inline_vecs[0];
Linus Torvalds1da177e2005-04-16 15:20:36 -0700112};
113
114/*
115 * bio flags
116 */
117#define BIO_UPTODATE 0 /* ok after I/O completion */
118#define BIO_RW_BLOCK 1 /* RW_AHEAD set, and read/write would block */
119#define BIO_EOF 2 /* out-out-bounds error */
Mikulas Patocka5df97b92008-08-15 10:20:02 +0200120#define BIO_SEG_VALID 3 /* bi_phys_segments valid */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700121#define BIO_CLONED 4 /* doesn't own data */
122#define BIO_BOUNCED 5 /* bio is a bounce bio */
123#define BIO_USER_MAPPED 6 /* contains user pages */
124#define BIO_EOPNOTSUPP 7 /* not supported */
Jens Axboec7c22e42008-09-13 20:26:01 +0200125#define BIO_CPU_AFFINE 8 /* complete bio on same CPU as submitted */
FUJITA Tomonori81882762008-09-02 16:20:19 +0900126#define BIO_NULL_MAPPED 9 /* contains invalid user pages */
Martin K. Petersen74aa8c22008-10-01 03:38:37 -0400127#define BIO_FS_INTEGRITY 10 /* fs owns integrity data, not block layer */
Keith Mannthey08bafc02008-11-25 10:24:35 +0100128#define BIO_QUIET 11 /* Make BIO Quiet */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700129#define bio_flagged(bio, flag) ((bio)->bi_flags & (1 << (flag)))
130
131/*
132 * top 4 bits of bio flags indicate the pool this bio came from
133 */
134#define BIO_POOL_BITS (4)
Tejun Heo451a9eb2009-04-15 19:50:51 +0200135#define BIO_POOL_NONE ((1UL << BIO_POOL_BITS) - 1)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700136#define BIO_POOL_OFFSET (BITS_PER_LONG - BIO_POOL_BITS)
137#define BIO_POOL_MASK (1UL << BIO_POOL_OFFSET)
138#define BIO_POOL_IDX(bio) ((bio)->bi_flags >> BIO_POOL_OFFSET)
139
140/*
141 * bio bi_rw flags
142 *
Jens Axboeaf563942008-10-09 09:01:10 +0200143 * bit 0 -- data direction
144 * If not set, bio is a read from device. If set, it's a write to device.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700145 * bit 1 -- rw-ahead when set
146 * bit 2 -- barrier
Jens Axboeaf563942008-10-09 09:01:10 +0200147 * Insert a serialization point in the IO queue, forcing previously
Alberto Bertogli16642eb2009-01-05 10:18:53 +0100148 * submitted IO to be completed before this one is issued.
Jens Axboeaeb6faf2009-04-06 14:48:07 +0200149 * bit 3 -- synchronous I/O hint.
150 * bit 4 -- Unplug the device immediately after submitting this bio.
151 * bit 5 -- metadata request
Jens Axboeaf563942008-10-09 09:01:10 +0200152 * Used for tracing to differentiate metadata and data IO. May also
153 * get some preferential treatment in the IO scheduler
Jens Axboeaeb6faf2009-04-06 14:48:07 +0200154 * bit 6 -- discard sectors
Jens Axboeaf563942008-10-09 09:01:10 +0200155 * Informs the lower level device that this range of sectors is no longer
156 * used by the file system and may thus be freed by the device. Used
157 * for flash based storage.
Jens Axboeaeb6faf2009-04-06 14:48:07 +0200158 * bit 7 -- fail fast device errors
159 * bit 8 -- fail fast transport errors
160 * bit 9 -- fail fast driver errors
Mike Christie6000a362008-08-19 18:45:30 -0500161 * Don't want driver retries for any fast fail whatever the reason.
Jens Axboeaeb6faf2009-04-06 14:48:07 +0200162 * bit 10 -- Tell the IO scheduler not to wait for more requests after this
163 one has been submitted, even if it is a SYNC request.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700164 */
David Woodhoused628eae2008-08-09 16:22:17 +0100165#define BIO_RW 0 /* Must match RW in req flags (blkdev.h) */
166#define BIO_RW_AHEAD 1 /* Must match FAILFAST in req flags */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700167#define BIO_RW_BARRIER 2
Jens Axboe213d9412009-01-06 09:16:05 +0100168#define BIO_RW_SYNCIO 3
169#define BIO_RW_UNPLUG 4
170#define BIO_RW_META 5
171#define BIO_RW_DISCARD 6
172#define BIO_RW_FAILFAST_DEV 7
173#define BIO_RW_FAILFAST_TRANSPORT 8
174#define BIO_RW_FAILFAST_DRIVER 9
Jens Axboeaeb6faf2009-04-06 14:48:07 +0200175#define BIO_RW_NOIDLE 10
Jens Axboe213d9412009-01-06 09:16:05 +0100176
Jens Axboe1dfa17f2009-01-06 09:21:49 +0100177#define bio_rw_flagged(bio, flag) ((bio)->bi_rw & (1 << (flag)))
178
179/*
180 * Old defines, these should eventually be replaced by direct usage of
181 * bio_rw_flagged()
182 */
183#define bio_barrier(bio) bio_rw_flagged(bio, BIO_RW_BARRIER)
184#define bio_sync(bio) bio_rw_flagged(bio, BIO_RW_SYNCIO)
185#define bio_unplug(bio) bio_rw_flagged(bio, BIO_RW_UNPLUG)
186#define bio_failfast_dev(bio) bio_rw_flagged(bio, BIO_RW_FAILFAST_DEV)
187#define bio_failfast_transport(bio) \
188 bio_rw_flagged(bio, BIO_RW_FAILFAST_TRANSPORT)
189#define bio_failfast_driver(bio) \
190 bio_rw_flagged(bio, BIO_RW_FAILFAST_DRIVER)
191#define bio_rw_ahead(bio) bio_rw_flagged(bio, BIO_RW_AHEAD)
192#define bio_rw_meta(bio) bio_rw_flagged(bio, BIO_RW_META)
193#define bio_discard(bio) bio_rw_flagged(bio, BIO_RW_DISCARD)
Jens Axboeaeb6faf2009-04-06 14:48:07 +0200194#define bio_noidle(bio) bio_rw_flagged(bio, BIO_RW_NOIDLE)
Jens Axboe1dfa17f2009-01-06 09:21:49 +0100195
Linus Torvalds1da177e2005-04-16 15:20:36 -0700196/*
Jens Axboe22e2c502005-06-27 10:55:12 +0200197 * upper 16 bits of bi_rw define the io priority of this bio
198 */
199#define BIO_PRIO_SHIFT (8 * sizeof(unsigned long) - IOPRIO_BITS)
200#define bio_prio(bio) ((bio)->bi_rw >> BIO_PRIO_SHIFT)
201#define bio_prio_valid(bio) ioprio_valid(bio_prio(bio))
202
203#define bio_set_prio(bio, prio) do { \
204 WARN_ON(prio >= (1 << IOPRIO_BITS)); \
205 (bio)->bi_rw &= ((1UL << BIO_PRIO_SHIFT) - 1); \
206 (bio)->bi_rw |= ((unsigned long) (prio) << BIO_PRIO_SHIFT); \
207} while (0)
208
209/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700210 * various member access, note that bio_data should of course not be used
211 * on highmem page vectors
212 */
213#define bio_iovec_idx(bio, idx) (&((bio)->bi_io_vec[(idx)]))
214#define bio_iovec(bio) bio_iovec_idx((bio), (bio)->bi_idx)
215#define bio_page(bio) bio_iovec((bio))->bv_page
216#define bio_offset(bio) bio_iovec((bio))->bv_offset
217#define bio_segments(bio) ((bio)->bi_vcnt - (bio)->bi_idx)
218#define bio_sectors(bio) ((bio)->bi_size >> 9)
David Woodhousee17fc0a2008-08-09 16:42:20 +0100219#define bio_empty_barrier(bio) (bio_barrier(bio) && !bio_has_data(bio) && !bio_discard(bio))
Jens Axboebf2de6f2007-09-27 13:01:25 +0200220
221static inline unsigned int bio_cur_sectors(struct bio *bio)
222{
223 if (bio->bi_vcnt)
224 return bio_iovec(bio)->bv_len >> 9;
David Woodhousefb2dce82008-08-05 18:01:53 +0100225 else /* dataless requests such as discard */
226 return bio->bi_size >> 9;
Jens Axboebf2de6f2007-09-27 13:01:25 +0200227}
228
229static inline void *bio_data(struct bio *bio)
230{
231 if (bio->bi_vcnt)
232 return page_address(bio_page(bio)) + bio_offset(bio);
233
234 return NULL;
235}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700236
Jens Axboe392ddc32008-12-23 12:42:54 +0100237static inline int bio_has_allocated_vec(struct bio *bio)
238{
239 return bio->bi_io_vec && bio->bi_io_vec != bio->bi_inline_vecs;
240}
241
Linus Torvalds1da177e2005-04-16 15:20:36 -0700242/*
243 * will die
244 */
245#define bio_to_phys(bio) (page_to_phys(bio_page((bio))) + (unsigned long) bio_offset((bio)))
246#define bvec_to_phys(bv) (page_to_phys((bv)->bv_page) + (unsigned long) (bv)->bv_offset)
247
248/*
249 * queues that have highmem support enabled may still need to revert to
250 * PIO transfers occasionally and thus map high pages temporarily. For
251 * permanent PIO fall back, user is probably better off disabling highmem
252 * I/O completely on that queue (see ide-dma for example)
253 */
254#define __bio_kmap_atomic(bio, idx, kmtype) \
255 (kmap_atomic(bio_iovec_idx((bio), (idx))->bv_page, kmtype) + \
256 bio_iovec_idx((bio), (idx))->bv_offset)
257
258#define __bio_kunmap_atomic(addr, kmtype) kunmap_atomic(addr, kmtype)
259
260/*
261 * merge helpers etc
262 */
263
264#define __BVEC_END(bio) bio_iovec_idx((bio), (bio)->bi_vcnt - 1)
265#define __BVEC_START(bio) bio_iovec_idx((bio), (bio)->bi_idx)
266
Jeremy Fitzhardingef92131c2008-10-29 14:10:51 +0100267/* Default implementation of BIOVEC_PHYS_MERGEABLE */
268#define __BIOVEC_PHYS_MERGEABLE(vec1, vec2) \
269 ((bvec_to_phys((vec1)) + (vec1)->bv_len) == bvec_to_phys((vec2)))
270
Linus Torvalds1da177e2005-04-16 15:20:36 -0700271/*
272 * allow arch override, for eg virtualized architectures (put in asm/io.h)
273 */
274#ifndef BIOVEC_PHYS_MERGEABLE
275#define BIOVEC_PHYS_MERGEABLE(vec1, vec2) \
Jeremy Fitzhardingef92131c2008-10-29 14:10:51 +0100276 __BIOVEC_PHYS_MERGEABLE(vec1, vec2)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700277#endif
278
Linus Torvalds1da177e2005-04-16 15:20:36 -0700279#define __BIO_SEG_BOUNDARY(addr1, addr2, mask) \
280 (((addr1) | (mask)) == (((addr2) - 1) | (mask)))
281#define BIOVEC_SEG_BOUNDARY(q, b1, b2) \
282 __BIO_SEG_BOUNDARY(bvec_to_phys((b1)), bvec_to_phys((b2)) + (b2)->bv_len, (q)->seg_boundary_mask)
283#define BIO_SEG_BOUNDARY(q, b1, b2) \
284 BIOVEC_SEG_BOUNDARY((q), __BVEC_END((b1)), __BVEC_START((b2)))
285
NeilBrown6712ecf2007-09-27 12:47:43 +0200286#define bio_io_error(bio) bio_endio((bio), -EIO)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700287
288/*
289 * drivers should not use the __ version unless they _really_ want to
290 * run through the entire bio and not just pending pieces
291 */
292#define __bio_for_each_segment(bvl, bio, i, start_idx) \
293 for (bvl = bio_iovec_idx((bio), (start_idx)), i = (start_idx); \
294 i < (bio)->bi_vcnt; \
295 bvl++, i++)
296
297#define bio_for_each_segment(bvl, bio, i) \
298 __bio_for_each_segment(bvl, bio, i, (bio)->bi_idx)
299
300/*
301 * get a reference to a bio, so it won't disappear. the intended use is
302 * something like:
303 *
304 * bio_get(bio);
305 * submit_bio(rw, bio);
306 * if (bio->bi_flags ...)
307 * do_something
308 * bio_put(bio);
309 *
310 * without the bio_get(), it could potentially complete I/O before submit_bio
311 * returns. and then bio would be freed memory when if (bio->bi_flags ...)
312 * runs
313 */
314#define bio_get(bio) atomic_inc(&(bio)->bi_cnt)
315
Martin K. Petersen7ba1ba12008-06-30 20:04:41 +0200316#if defined(CONFIG_BLK_DEV_INTEGRITY)
317/*
318 * bio integrity payload
319 */
320struct bio_integrity_payload {
321 struct bio *bip_bio; /* parent bio */
322 struct bio_vec *bip_vec; /* integrity data vector */
323
324 sector_t bip_sector; /* virtual start sector */
325
326 void *bip_buf; /* generated integrity data */
327 bio_end_io_t *bip_end_io; /* saved I/O completion fn */
328
Martin K. Petersen7ba1ba12008-06-30 20:04:41 +0200329 unsigned int bip_size;
330
331 unsigned short bip_pool; /* pool the ivec came from */
332 unsigned short bip_vcnt; /* # of integrity bio_vecs */
333 unsigned short bip_idx; /* current bip_vec index */
334
335 struct work_struct bip_work; /* I/O completion */
336};
337#endif /* CONFIG_BLK_DEV_INTEGRITY */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700338
339/*
340 * A bio_pair is used when we need to split a bio.
341 * This can only happen for a bio that refers to just one
342 * page of data, and in the unusual situation when the
343 * page crosses a chunk/device boundary
344 *
345 * The address of the master bio is stored in bio1.bi_private
346 * The address of the pool the pair was allocated from is stored
347 * in bio2.bi_private
348 */
349struct bio_pair {
Martin K. Petersen7ba1ba12008-06-30 20:04:41 +0200350 struct bio bio1, bio2;
351 struct bio_vec bv1, bv2;
352#if defined(CONFIG_BLK_DEV_INTEGRITY)
353 struct bio_integrity_payload bip1, bip2;
354 struct bio_vec iv1, iv2;
355#endif
356 atomic_t cnt;
357 int error;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700358};
Denis ChengRq6feef532008-10-09 08:57:05 +0200359extern struct bio_pair *bio_split(struct bio *bi, int first_sectors);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700360extern void bio_pair_release(struct bio_pair *dbio);
361
Jens Axboebb799ca2008-12-10 15:35:05 +0100362extern struct bio_set *bioset_create(unsigned int, unsigned int);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700363extern void bioset_free(struct bio_set *);
364
Al Virodd0fc662005-10-07 07:46:04 +0100365extern struct bio *bio_alloc(gfp_t, int);
Jens Axboe0a0d96b2008-09-11 13:17:37 +0200366extern struct bio *bio_kmalloc(gfp_t, int);
Al Virodd0fc662005-10-07 07:46:04 +0100367extern struct bio *bio_alloc_bioset(gfp_t, int, struct bio_set *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700368extern void bio_put(struct bio *);
Peter Osterlund36763472005-09-06 15:16:42 -0700369extern void bio_free(struct bio *, struct bio_set *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700370
NeilBrown6712ecf2007-09-27 12:47:43 +0200371extern void bio_endio(struct bio *, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700372struct request_queue;
373extern int bio_phys_segments(struct request_queue *, struct bio *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700374
375extern void __bio_clone(struct bio *, struct bio *);
Al Virodd0fc662005-10-07 07:46:04 +0100376extern struct bio *bio_clone(struct bio *, gfp_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700377
378extern void bio_init(struct bio *);
379
380extern int bio_add_page(struct bio *, struct page *, unsigned int,unsigned int);
Mike Christie6e68af62005-11-11 05:30:27 -0600381extern int bio_add_pc_page(struct request_queue *, struct bio *, struct page *,
382 unsigned int, unsigned int);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700383extern int bio_get_nr_vecs(struct block_device *);
Martin K. Petersenad3316b2008-10-01 22:42:53 -0400384extern sector_t bio_sector_offset(struct bio *, unsigned short, unsigned int);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700385extern struct bio *bio_map_user(struct request_queue *, struct block_device *,
FUJITA Tomonoria3bce902008-08-28 16:17:05 +0900386 unsigned long, unsigned int, int, gfp_t);
James Bottomley f1970ba2005-06-20 14:06:52 +0200387struct sg_iovec;
FUJITA Tomonori152e2832008-08-28 16:17:06 +0900388struct rq_map_data;
James Bottomley f1970ba2005-06-20 14:06:52 +0200389extern struct bio *bio_map_user_iov(struct request_queue *,
390 struct block_device *,
FUJITA Tomonoria3bce902008-08-28 16:17:05 +0900391 struct sg_iovec *, int, int, gfp_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700392extern void bio_unmap_user(struct bio *);
Mike Christie df46b9a2005-06-20 14:04:44 +0200393extern struct bio *bio_map_kern(struct request_queue *, void *, unsigned int,
Al Viro27496a82005-10-21 03:20:48 -0400394 gfp_t);
FUJITA Tomonori68154e92008-04-25 12:47:50 +0200395extern struct bio *bio_copy_kern(struct request_queue *, void *, unsigned int,
396 gfp_t, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700397extern void bio_set_pages_dirty(struct bio *bio);
398extern void bio_check_pages_dirty(struct bio *bio);
FUJITA Tomonori152e2832008-08-28 16:17:06 +0900399extern struct bio *bio_copy_user(struct request_queue *, struct rq_map_data *,
400 unsigned long, unsigned int, int, gfp_t);
401extern struct bio *bio_copy_user_iov(struct request_queue *,
402 struct rq_map_data *, struct sg_iovec *,
FUJITA Tomonoria3bce902008-08-28 16:17:05 +0900403 int, int, gfp_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700404extern int bio_uncopy_user(struct bio *);
405void zero_fill_bio(struct bio *bio);
Martin K. Petersen51d654e2008-06-17 18:59:56 +0200406extern struct bio_vec *bvec_alloc_bs(gfp_t, int, unsigned long *, struct bio_set *);
Jens Axboebb799ca2008-12-10 15:35:05 +0100407extern void bvec_free_bs(struct bio_set *, struct bio_vec *, unsigned int);
Martin K. Petersen7ba1ba12008-06-30 20:04:41 +0200408extern unsigned int bvec_nr_vecs(unsigned short idx);
Martin K. Petersen51d654e2008-06-17 18:59:56 +0200409
410/*
Jens Axboec7c22e42008-09-13 20:26:01 +0200411 * Allow queuer to specify a completion CPU for this bio
412 */
413static inline void bio_set_completion_cpu(struct bio *bio, unsigned int cpu)
414{
415 bio->bi_comp_cpu = cpu;
416}
417
418/*
Martin K. Petersen51d654e2008-06-17 18:59:56 +0200419 * bio_set is used to allow other portions of the IO system to
420 * allocate their own private memory pools for bio and iovec structures.
421 * These memory pools in turn all allocate from the bio_slab
422 * and the bvec_slabs[].
423 */
424#define BIO_POOL_SIZE 2
425#define BIOVEC_NR_POOLS 6
Jens Axboe7ff93452008-12-11 11:53:43 +0100426#define BIOVEC_MAX_IDX (BIOVEC_NR_POOLS - 1)
Martin K. Petersen51d654e2008-06-17 18:59:56 +0200427
428struct bio_set {
Jens Axboe1b434492008-10-22 20:32:58 +0200429 struct kmem_cache *bio_slab;
Jens Axboebb799ca2008-12-10 15:35:05 +0100430 unsigned int front_pad;
431
Martin K. Petersen51d654e2008-06-17 18:59:56 +0200432 mempool_t *bio_pool;
Jens Axboe7ff93452008-12-11 11:53:43 +0100433 mempool_t *bvec_pool;
Martin K. Petersen51d654e2008-06-17 18:59:56 +0200434};
435
436struct biovec_slab {
437 int nr_vecs;
438 char *name;
439 struct kmem_cache *slab;
440};
441
442extern struct bio_set *fs_bio_set;
Jens Axboebb799ca2008-12-10 15:35:05 +0100443extern struct biovec_slab bvec_slabs[BIOVEC_NR_POOLS] __read_mostly;
Martin K. Petersen51d654e2008-06-17 18:59:56 +0200444
445/*
446 * a small number of entries is fine, not going to be performance critical.
447 * basically we just need to survive
448 */
449#define BIO_SPLIT_ENTRIES 2
Linus Torvalds1da177e2005-04-16 15:20:36 -0700450
451#ifdef CONFIG_HIGHMEM
452/*
Alberto Bertogli20b636b2009-02-02 12:41:07 +0100453 * remember never ever reenable interrupts between a bvec_kmap_irq and
454 * bvec_kunmap_irq!
Linus Torvalds1da177e2005-04-16 15:20:36 -0700455 *
456 * This function MUST be inlined - it plays with the CPU interrupt flags.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700457 */
Alberto Bertoglic52440a2009-02-02 12:41:07 +0100458static __always_inline char *bvec_kmap_irq(struct bio_vec *bvec,
459 unsigned long *flags)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700460{
461 unsigned long addr;
462
463 /*
464 * might not be a highmem page, but the preempt/irq count
465 * balancing is a lot nicer this way
466 */
467 local_irq_save(*flags);
468 addr = (unsigned long) kmap_atomic(bvec->bv_page, KM_BIO_SRC_IRQ);
469
470 BUG_ON(addr & ~PAGE_MASK);
471
472 return (char *) addr + bvec->bv_offset;
473}
474
Alberto Bertoglic52440a2009-02-02 12:41:07 +0100475static __always_inline void bvec_kunmap_irq(char *buffer,
476 unsigned long *flags)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700477{
478 unsigned long ptr = (unsigned long) buffer & PAGE_MASK;
479
480 kunmap_atomic((void *) ptr, KM_BIO_SRC_IRQ);
481 local_irq_restore(*flags);
482}
483
484#else
485#define bvec_kmap_irq(bvec, flags) (page_address((bvec)->bv_page) + (bvec)->bv_offset)
486#define bvec_kunmap_irq(buf, flags) do { *(flags) = 0; } while (0)
487#endif
488
Adrian Bunkc2d08da2005-09-10 00:27:18 -0700489static inline char *__bio_kmap_irq(struct bio *bio, unsigned short idx,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700490 unsigned long *flags)
491{
492 return bvec_kmap_irq(bio_iovec_idx(bio, idx), flags);
493}
494#define __bio_kunmap_irq(buf, flags) bvec_kunmap_irq(buf, flags)
495
496#define bio_kmap_irq(bio, flags) \
497 __bio_kmap_irq((bio), (bio)->bi_idx, (flags))
498#define bio_kunmap_irq(buf,flags) __bio_kunmap_irq(buf, flags)
499
Jens Axboe7a67f632008-08-08 11:17:12 +0200500/*
501 * Check whether this bio carries any data or not. A NULL bio is allowed.
502 */
503static inline int bio_has_data(struct bio *bio)
504{
505 return bio && bio->bi_io_vec != NULL;
506}
507
Christoph Hellwig8f3d8ba2009-04-07 19:55:13 +0200508/*
509 * BIO list managment for use by remapping drivers (e.g. DM or MD).
510 *
511 * A bio_list anchors a singly-linked list of bios chained through the bi_next
512 * member of the bio. The bio_list also caches the last list member to allow
513 * fast access to the tail.
514 */
515struct bio_list {
516 struct bio *head;
517 struct bio *tail;
518};
519
520static inline int bio_list_empty(const struct bio_list *bl)
521{
522 return bl->head == NULL;
523}
524
525static inline void bio_list_init(struct bio_list *bl)
526{
527 bl->head = bl->tail = NULL;
528}
529
530#define bio_list_for_each(bio, bl) \
531 for (bio = (bl)->head; bio; bio = bio->bi_next)
532
533static inline unsigned bio_list_size(const struct bio_list *bl)
534{
535 unsigned sz = 0;
536 struct bio *bio;
537
538 bio_list_for_each(bio, bl)
539 sz++;
540
541 return sz;
542}
543
544static inline void bio_list_add(struct bio_list *bl, struct bio *bio)
545{
546 bio->bi_next = NULL;
547
548 if (bl->tail)
549 bl->tail->bi_next = bio;
550 else
551 bl->head = bio;
552
553 bl->tail = bio;
554}
555
556static inline void bio_list_add_head(struct bio_list *bl, struct bio *bio)
557{
558 bio->bi_next = bl->head;
559
560 bl->head = bio;
561
562 if (!bl->tail)
563 bl->tail = bio;
564}
565
566static inline void bio_list_merge(struct bio_list *bl, struct bio_list *bl2)
567{
568 if (!bl2->head)
569 return;
570
571 if (bl->tail)
572 bl->tail->bi_next = bl2->head;
573 else
574 bl->head = bl2->head;
575
576 bl->tail = bl2->tail;
577}
578
579static inline void bio_list_merge_head(struct bio_list *bl,
580 struct bio_list *bl2)
581{
582 if (!bl2->head)
583 return;
584
585 if (bl->head)
586 bl2->tail->bi_next = bl->head;
587 else
588 bl->tail = bl2->tail;
589
590 bl->head = bl2->head;
591}
592
593static inline struct bio *bio_list_pop(struct bio_list *bl)
594{
595 struct bio *bio = bl->head;
596
597 if (bio) {
598 bl->head = bl->head->bi_next;
599 if (!bl->head)
600 bl->tail = NULL;
601
602 bio->bi_next = NULL;
603 }
604
605 return bio;
606}
607
608static inline struct bio *bio_list_get(struct bio_list *bl)
609{
610 struct bio *bio = bl->head;
611
612 bl->head = bl->tail = NULL;
613
614 return bio;
615}
616
Martin K. Petersen7ba1ba12008-06-30 20:04:41 +0200617#if defined(CONFIG_BLK_DEV_INTEGRITY)
618
619#define bip_vec_idx(bip, idx) (&(bip->bip_vec[(idx)]))
620#define bip_vec(bip) bip_vec_idx(bip, 0)
621
622#define __bip_for_each_vec(bvl, bip, i, start_idx) \
623 for (bvl = bip_vec_idx((bip), (start_idx)), i = (start_idx); \
624 i < (bip)->bip_vcnt; \
625 bvl++, i++)
626
627#define bip_for_each_vec(bvl, bip, i) \
628 __bip_for_each_vec(bvl, bip, i, (bip)->bip_idx)
629
Alberto Bertogli8deaf722008-10-02 12:46:53 +0200630#define bio_integrity(bio) (bio->bi_integrity != NULL)
Martin K. Petersen7ba1ba12008-06-30 20:04:41 +0200631
Martin K. Petersen7ba1ba12008-06-30 20:04:41 +0200632extern struct bio_integrity_payload *bio_integrity_alloc(struct bio *, gfp_t, unsigned int);
Martin K. Petersen6d2a78e2009-03-10 08:27:39 +0100633extern void bio_integrity_free(struct bio *);
Martin K. Petersen7ba1ba12008-06-30 20:04:41 +0200634extern int bio_integrity_add_page(struct bio *, struct page *, unsigned int, unsigned int);
635extern int bio_integrity_enabled(struct bio *bio);
636extern int bio_integrity_set_tag(struct bio *, void *, unsigned int);
637extern int bio_integrity_get_tag(struct bio *, void *, unsigned int);
638extern int bio_integrity_prep(struct bio *);
639extern void bio_integrity_endio(struct bio *, int);
640extern void bio_integrity_advance(struct bio *, unsigned int);
641extern void bio_integrity_trim(struct bio *, unsigned int, unsigned int);
642extern void bio_integrity_split(struct bio *, struct bio_pair *, int);
Martin K. Petersen6d2a78e2009-03-10 08:27:39 +0100643extern int bio_integrity_clone(struct bio *, struct bio *, gfp_t);
Martin K. Petersen7ba1ba12008-06-30 20:04:41 +0200644
645#else /* CONFIG_BLK_DEV_INTEGRITY */
646
647#define bio_integrity(a) (0)
Martin K. Petersen7ba1ba12008-06-30 20:04:41 +0200648#define bio_integrity_prep(a) (0)
649#define bio_integrity_enabled(a) (0)
Martin K. Petersen6d2a78e2009-03-10 08:27:39 +0100650#define bio_integrity_clone(a, b, c) (0)
651#define bio_integrity_free(a) do { } while (0)
Martin K. Petersen7ba1ba12008-06-30 20:04:41 +0200652#define bio_integrity_endio(a, b) do { } while (0)
653#define bio_integrity_advance(a, b) do { } while (0)
654#define bio_integrity_trim(a, b, c) do { } while (0)
655#define bio_integrity_split(a, b, c) do { } while (0)
656#define bio_integrity_set_tag(a, b, c) do { } while (0)
657#define bio_integrity_get_tag(a, b, c) do { } while (0)
Martin K. Petersen7ba1ba12008-06-30 20:04:41 +0200658
659#endif /* CONFIG_BLK_DEV_INTEGRITY */
660
David Howells02a5e0a2007-08-11 22:34:32 +0200661#endif /* CONFIG_BLOCK */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700662#endif /* __LINUX_BIO_H */