blob: 4d85262b4fa4e8a4e9eb984e668f57363fd09183 [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
27/* Platforms may set this to teach the BIO layer about IOMMU hardware. */
28#include <asm/io.h>
29
30#if defined(BIO_VMERGE_MAX_SIZE) && defined(BIO_VMERGE_BOUNDARY)
31#define BIOVEC_VIRT_START_SIZE(x) (bvec_to_phys(x) & (BIO_VMERGE_BOUNDARY - 1))
32#define BIOVEC_VIRT_OVERSIZE(x) ((x) > BIO_VMERGE_MAX_SIZE)
33#else
34#define BIOVEC_VIRT_START_SIZE(x) 0
35#define BIOVEC_VIRT_OVERSIZE(x) 0
36#endif
37
38#ifndef BIO_VMERGE_BOUNDARY
39#define BIO_VMERGE_BOUNDARY 0
40#endif
41
42#define BIO_DEBUG
43
44#ifdef BIO_DEBUG
45#define BIO_BUG_ON BUG_ON
46#else
47#define BIO_BUG_ON
48#endif
49
Alexey Dobriyand84a8472006-06-25 05:49:32 -070050#define BIO_MAX_PAGES 256
Linus Torvalds1da177e2005-04-16 15:20:36 -070051#define BIO_MAX_SIZE (BIO_MAX_PAGES << PAGE_CACHE_SHIFT)
52#define BIO_MAX_SECTORS (BIO_MAX_SIZE >> 9)
53
54/*
55 * was unsigned short, but we might as well be ready for > 64kB I/O pages
56 */
57struct bio_vec {
58 struct page *bv_page;
59 unsigned int bv_len;
60 unsigned int bv_offset;
61};
62
63struct bio_set;
64struct bio;
65typedef int (bio_end_io_t) (struct bio *, unsigned int, int);
66typedef void (bio_destructor_t) (struct bio *);
67
68/*
69 * main unit of I/O for the block layer and lower layers (ie drivers and
70 * stacking drivers)
71 */
72struct bio {
Roger Gammans2c2345c2006-10-04 13:37:45 +020073 sector_t bi_sector; /* device address in 512 byte
74 sectors */
Linus Torvalds1da177e2005-04-16 15:20:36 -070075 struct bio *bi_next; /* request queue link */
76 struct block_device *bi_bdev;
77 unsigned long bi_flags; /* status, command, etc */
78 unsigned long bi_rw; /* bottom bits READ/WRITE,
79 * top bits priority
80 */
81
82 unsigned short bi_vcnt; /* how many bio_vec's */
83 unsigned short bi_idx; /* current index into bvl_vec */
84
85 /* Number of segments in this BIO after
86 * physical address coalescing is performed.
87 */
88 unsigned short bi_phys_segments;
89
90 /* Number of segments after physical and DMA remapping
91 * hardware coalescing is performed.
92 */
93 unsigned short bi_hw_segments;
94
95 unsigned int bi_size; /* residual I/O count */
96
97 /*
98 * To keep track of the max hw size, we account for the
99 * sizes of the first and last virtually mergeable segments
100 * in this bio
101 */
102 unsigned int bi_hw_front_size;
103 unsigned int bi_hw_back_size;
104
105 unsigned int bi_max_vecs; /* max bvl_vecs we can hold */
106
107 struct bio_vec *bi_io_vec; /* the actual vec list */
108
109 bio_end_io_t *bi_end_io;
110 atomic_t bi_cnt; /* pin count */
111
112 void *bi_private;
113
114 bio_destructor_t *bi_destructor; /* destructor */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700115};
116
117/*
118 * bio flags
119 */
120#define BIO_UPTODATE 0 /* ok after I/O completion */
121#define BIO_RW_BLOCK 1 /* RW_AHEAD set, and read/write would block */
122#define BIO_EOF 2 /* out-out-bounds error */
123#define BIO_SEG_VALID 3 /* nr_hw_seg valid */
124#define BIO_CLONED 4 /* doesn't own data */
125#define BIO_BOUNCED 5 /* bio is a bounce bio */
126#define BIO_USER_MAPPED 6 /* contains user pages */
127#define BIO_EOPNOTSUPP 7 /* not supported */
128#define bio_flagged(bio, flag) ((bio)->bi_flags & (1 << (flag)))
129
130/*
131 * top 4 bits of bio flags indicate the pool this bio came from
132 */
133#define BIO_POOL_BITS (4)
134#define BIO_POOL_OFFSET (BITS_PER_LONG - BIO_POOL_BITS)
135#define BIO_POOL_MASK (1UL << BIO_POOL_OFFSET)
136#define BIO_POOL_IDX(bio) ((bio)->bi_flags >> BIO_POOL_OFFSET)
137
138/*
139 * bio bi_rw flags
140 *
141 * bit 0 -- read (not set) or write (set)
142 * bit 1 -- rw-ahead when set
143 * bit 2 -- barrier
144 * bit 3 -- fail fast, don't want low level driver retries
145 * bit 4 -- synchronous I/O hint: the block layer will unplug immediately
146 */
147#define BIO_RW 0
148#define BIO_RW_AHEAD 1
149#define BIO_RW_BARRIER 2
150#define BIO_RW_FAILFAST 3
151#define BIO_RW_SYNC 4
Jens Axboe5404bc72006-08-10 09:01:02 +0200152#define BIO_RW_META 5
Linus Torvalds1da177e2005-04-16 15:20:36 -0700153
154/*
Jens Axboe22e2c502005-06-27 10:55:12 +0200155 * upper 16 bits of bi_rw define the io priority of this bio
156 */
157#define BIO_PRIO_SHIFT (8 * sizeof(unsigned long) - IOPRIO_BITS)
158#define bio_prio(bio) ((bio)->bi_rw >> BIO_PRIO_SHIFT)
159#define bio_prio_valid(bio) ioprio_valid(bio_prio(bio))
160
161#define bio_set_prio(bio, prio) do { \
162 WARN_ON(prio >= (1 << IOPRIO_BITS)); \
163 (bio)->bi_rw &= ((1UL << BIO_PRIO_SHIFT) - 1); \
164 (bio)->bi_rw |= ((unsigned long) (prio) << BIO_PRIO_SHIFT); \
165} while (0)
166
167/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700168 * various member access, note that bio_data should of course not be used
169 * on highmem page vectors
170 */
171#define bio_iovec_idx(bio, idx) (&((bio)->bi_io_vec[(idx)]))
172#define bio_iovec(bio) bio_iovec_idx((bio), (bio)->bi_idx)
173#define bio_page(bio) bio_iovec((bio))->bv_page
174#define bio_offset(bio) bio_iovec((bio))->bv_offset
175#define bio_segments(bio) ((bio)->bi_vcnt - (bio)->bi_idx)
176#define bio_sectors(bio) ((bio)->bi_size >> 9)
177#define bio_cur_sectors(bio) (bio_iovec(bio)->bv_len >> 9)
178#define bio_data(bio) (page_address(bio_page((bio))) + bio_offset((bio)))
179#define bio_barrier(bio) ((bio)->bi_rw & (1 << BIO_RW_BARRIER))
180#define bio_sync(bio) ((bio)->bi_rw & (1 << BIO_RW_SYNC))
181#define bio_failfast(bio) ((bio)->bi_rw & (1 << BIO_RW_FAILFAST))
182#define bio_rw_ahead(bio) ((bio)->bi_rw & (1 << BIO_RW_AHEAD))
Jens Axboe5404bc72006-08-10 09:01:02 +0200183#define bio_rw_meta(bio) ((bio)->bi_rw & (1 << BIO_RW_META))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700184
185/*
186 * will die
187 */
188#define bio_to_phys(bio) (page_to_phys(bio_page((bio))) + (unsigned long) bio_offset((bio)))
189#define bvec_to_phys(bv) (page_to_phys((bv)->bv_page) + (unsigned long) (bv)->bv_offset)
190
191/*
192 * queues that have highmem support enabled may still need to revert to
193 * PIO transfers occasionally and thus map high pages temporarily. For
194 * permanent PIO fall back, user is probably better off disabling highmem
195 * I/O completely on that queue (see ide-dma for example)
196 */
197#define __bio_kmap_atomic(bio, idx, kmtype) \
198 (kmap_atomic(bio_iovec_idx((bio), (idx))->bv_page, kmtype) + \
199 bio_iovec_idx((bio), (idx))->bv_offset)
200
201#define __bio_kunmap_atomic(addr, kmtype) kunmap_atomic(addr, kmtype)
202
203/*
204 * merge helpers etc
205 */
206
207#define __BVEC_END(bio) bio_iovec_idx((bio), (bio)->bi_vcnt - 1)
208#define __BVEC_START(bio) bio_iovec_idx((bio), (bio)->bi_idx)
209
210/*
211 * allow arch override, for eg virtualized architectures (put in asm/io.h)
212 */
213#ifndef BIOVEC_PHYS_MERGEABLE
214#define BIOVEC_PHYS_MERGEABLE(vec1, vec2) \
215 ((bvec_to_phys((vec1)) + (vec1)->bv_len) == bvec_to_phys((vec2)))
216#endif
217
218#define BIOVEC_VIRT_MERGEABLE(vec1, vec2) \
219 ((((bvec_to_phys((vec1)) + (vec1)->bv_len) | bvec_to_phys((vec2))) & (BIO_VMERGE_BOUNDARY - 1)) == 0)
220#define __BIO_SEG_BOUNDARY(addr1, addr2, mask) \
221 (((addr1) | (mask)) == (((addr2) - 1) | (mask)))
222#define BIOVEC_SEG_BOUNDARY(q, b1, b2) \
223 __BIO_SEG_BOUNDARY(bvec_to_phys((b1)), bvec_to_phys((b2)) + (b2)->bv_len, (q)->seg_boundary_mask)
224#define BIO_SEG_BOUNDARY(q, b1, b2) \
225 BIOVEC_SEG_BOUNDARY((q), __BVEC_END((b1)), __BVEC_START((b2)))
226
227#define bio_io_error(bio, bytes) bio_endio((bio), (bytes), -EIO)
228
229/*
230 * drivers should not use the __ version unless they _really_ want to
231 * run through the entire bio and not just pending pieces
232 */
233#define __bio_for_each_segment(bvl, bio, i, start_idx) \
234 for (bvl = bio_iovec_idx((bio), (start_idx)), i = (start_idx); \
235 i < (bio)->bi_vcnt; \
236 bvl++, i++)
237
238#define bio_for_each_segment(bvl, bio, i) \
239 __bio_for_each_segment(bvl, bio, i, (bio)->bi_idx)
240
241/*
242 * get a reference to a bio, so it won't disappear. the intended use is
243 * something like:
244 *
245 * bio_get(bio);
246 * submit_bio(rw, bio);
247 * if (bio->bi_flags ...)
248 * do_something
249 * bio_put(bio);
250 *
251 * without the bio_get(), it could potentially complete I/O before submit_bio
252 * returns. and then bio would be freed memory when if (bio->bi_flags ...)
253 * runs
254 */
255#define bio_get(bio) atomic_inc(&(bio)->bi_cnt)
256
257
258/*
259 * A bio_pair is used when we need to split a bio.
260 * This can only happen for a bio that refers to just one
261 * page of data, and in the unusual situation when the
262 * page crosses a chunk/device boundary
263 *
264 * The address of the master bio is stored in bio1.bi_private
265 * The address of the pool the pair was allocated from is stored
266 * in bio2.bi_private
267 */
268struct bio_pair {
269 struct bio bio1, bio2;
270 struct bio_vec bv1, bv2;
271 atomic_t cnt;
272 int error;
273};
274extern struct bio_pair *bio_split(struct bio *bi, mempool_t *pool,
275 int first_sectors);
276extern mempool_t *bio_split_pool;
277extern void bio_pair_release(struct bio_pair *dbio);
278
Jens Axboe59725112007-04-02 10:06:42 +0200279extern struct bio_set *bioset_create(int, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700280extern void bioset_free(struct bio_set *);
281
Al Virodd0fc662005-10-07 07:46:04 +0100282extern struct bio *bio_alloc(gfp_t, int);
283extern struct bio *bio_alloc_bioset(gfp_t, int, struct bio_set *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700284extern void bio_put(struct bio *);
Peter Osterlund36763472005-09-06 15:16:42 -0700285extern void bio_free(struct bio *, struct bio_set *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700286
287extern void bio_endio(struct bio *, unsigned int, int);
288struct request_queue;
289extern int bio_phys_segments(struct request_queue *, struct bio *);
290extern int bio_hw_segments(struct request_queue *, struct bio *);
291
292extern void __bio_clone(struct bio *, struct bio *);
Al Virodd0fc662005-10-07 07:46:04 +0100293extern struct bio *bio_clone(struct bio *, gfp_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700294
295extern void bio_init(struct bio *);
296
297extern int bio_add_page(struct bio *, struct page *, unsigned int,unsigned int);
Mike Christie6e68af62005-11-11 05:30:27 -0600298extern int bio_add_pc_page(struct request_queue *, struct bio *, struct page *,
299 unsigned int, unsigned int);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700300extern int bio_get_nr_vecs(struct block_device *);
301extern struct bio *bio_map_user(struct request_queue *, struct block_device *,
302 unsigned long, unsigned int, int);
James Bottomley f1970ba2005-06-20 14:06:52 +0200303struct sg_iovec;
304extern struct bio *bio_map_user_iov(struct request_queue *,
305 struct block_device *,
306 struct sg_iovec *, int, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700307extern void bio_unmap_user(struct bio *);
Mike Christie df46b9a2005-06-20 14:04:44 +0200308extern struct bio *bio_map_kern(struct request_queue *, void *, unsigned int,
Al Viro27496a82005-10-21 03:20:48 -0400309 gfp_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700310extern void bio_set_pages_dirty(struct bio *bio);
311extern void bio_check_pages_dirty(struct bio *bio);
Chen, Kenneth We61c9012006-12-13 00:34:36 -0800312extern void bio_release_pages(struct bio *bio);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700313extern struct bio *bio_copy_user(struct request_queue *, unsigned long, unsigned int, int);
314extern int bio_uncopy_user(struct bio *);
315void zero_fill_bio(struct bio *bio);
316
317#ifdef CONFIG_HIGHMEM
318/*
319 * remember to add offset! and never ever reenable interrupts between a
320 * bvec_kmap_irq and bvec_kunmap_irq!!
321 *
322 * This function MUST be inlined - it plays with the CPU interrupt flags.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700323 */
Adrian Bunkc2d08da2005-09-10 00:27:18 -0700324static inline char *bvec_kmap_irq(struct bio_vec *bvec, unsigned long *flags)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700325{
326 unsigned long addr;
327
328 /*
329 * might not be a highmem page, but the preempt/irq count
330 * balancing is a lot nicer this way
331 */
332 local_irq_save(*flags);
333 addr = (unsigned long) kmap_atomic(bvec->bv_page, KM_BIO_SRC_IRQ);
334
335 BUG_ON(addr & ~PAGE_MASK);
336
337 return (char *) addr + bvec->bv_offset;
338}
339
Adrian Bunkc2d08da2005-09-10 00:27:18 -0700340static inline void bvec_kunmap_irq(char *buffer, unsigned long *flags)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700341{
342 unsigned long ptr = (unsigned long) buffer & PAGE_MASK;
343
344 kunmap_atomic((void *) ptr, KM_BIO_SRC_IRQ);
345 local_irq_restore(*flags);
346}
347
348#else
349#define bvec_kmap_irq(bvec, flags) (page_address((bvec)->bv_page) + (bvec)->bv_offset)
350#define bvec_kunmap_irq(buf, flags) do { *(flags) = 0; } while (0)
351#endif
352
Adrian Bunkc2d08da2005-09-10 00:27:18 -0700353static inline char *__bio_kmap_irq(struct bio *bio, unsigned short idx,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700354 unsigned long *flags)
355{
356 return bvec_kmap_irq(bio_iovec_idx(bio, idx), flags);
357}
358#define __bio_kunmap_irq(buf, flags) bvec_kunmap_irq(buf, flags)
359
360#define bio_kmap_irq(bio, flags) \
361 __bio_kmap_irq((bio), (bio)->bi_idx, (flags))
362#define bio_kunmap_irq(buf,flags) __bio_kunmap_irq(buf, flags)
363
364#endif /* __LINUX_BIO_H */