blob: 5f604f269dfa87e26d24b97dbb945b333697f675 [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001/*
Jens Axboe0fe23472006-09-04 15:41:16 +02002 * Copyright (C) 2001 Jens Axboe <axboe@kernel.dk>
Linus Torvalds1da177e2005-04-16 15:20:36 -07003 *
4 * This program is free software; you can redistribute it and/or modify
5 * it under the terms of the GNU General Public License version 2 as
6 * published by the Free Software Foundation.
7 *
8 * This program is distributed in the hope that it will be useful,
9 * but WITHOUT ANY WARRANTY; without even the implied warranty of
10 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11 * GNU General Public License for more details.
12 *
13 * You should have received a copy of the GNU General Public Licens
14 * along with this program; if not, write to the Free Software
15 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-
16 *
17 */
18#include <linux/mm.h>
19#include <linux/swap.h>
20#include <linux/bio.h>
21#include <linux/blkdev.h>
22#include <linux/slab.h>
23#include <linux/init.h>
24#include <linux/kernel.h>
25#include <linux/module.h>
26#include <linux/mempool.h>
27#include <linux/workqueue.h>
Jens Axboe2056a782006-03-23 20:00:26 +010028#include <linux/blktrace_api.h>
James Bottomley f1970ba2005-06-20 14:06:52 +020029#include <scsi/sg.h> /* for struct sg_iovec */
Linus Torvalds1da177e2005-04-16 15:20:36 -070030
Jens Axboe59725112007-04-02 10:06:42 +020031#define BIO_POOL_SIZE 2
Linus Torvalds1da177e2005-04-16 15:20:36 -070032
Christoph Lametere18b8902006-12-06 20:33:20 -080033static struct kmem_cache *bio_slab __read_mostly;
Linus Torvalds1da177e2005-04-16 15:20:36 -070034
35#define BIOVEC_NR_POOLS 6
36
37/*
38 * a small number of entries is fine, not going to be performance critical.
39 * basically we just need to survive
40 */
Jens Axboe59725112007-04-02 10:06:42 +020041#define BIO_SPLIT_ENTRIES 2
Eric Dumazetfa3536c2006-03-26 01:37:24 -080042mempool_t *bio_split_pool __read_mostly;
Linus Torvalds1da177e2005-04-16 15:20:36 -070043
44struct biovec_slab {
45 int nr_vecs;
46 char *name;
Christoph Lametere18b8902006-12-06 20:33:20 -080047 struct kmem_cache *slab;
Linus Torvalds1da177e2005-04-16 15:20:36 -070048};
49
50/*
51 * if you change this list, also change bvec_alloc or things will
52 * break badly! cannot be bigger than what you can fit into an
53 * unsigned short
54 */
55
56#define BV(x) { .nr_vecs = x, .name = "biovec-"__stringify(x) }
Christoph Lameter6c036522005-07-07 17:56:59 -070057static struct biovec_slab bvec_slabs[BIOVEC_NR_POOLS] __read_mostly = {
Linus Torvalds1da177e2005-04-16 15:20:36 -070058 BV(1), BV(4), BV(16), BV(64), BV(128), BV(BIO_MAX_PAGES),
59};
60#undef BV
61
62/*
63 * bio_set is used to allow other portions of the IO system to
64 * allocate their own private memory pools for bio and iovec structures.
65 * These memory pools in turn all allocate from the bio_slab
66 * and the bvec_slabs[].
67 */
68struct bio_set {
69 mempool_t *bio_pool;
70 mempool_t *bvec_pools[BIOVEC_NR_POOLS];
71};
72
73/*
74 * fs_bio_set is the bio_set containing bio and iovec memory pools used by
75 * IO code that does not need private memory pools.
76 */
77static struct bio_set *fs_bio_set;
78
Al Virodd0fc662005-10-07 07:46:04 +010079static inline struct bio_vec *bvec_alloc_bs(gfp_t gfp_mask, int nr, unsigned long *idx, struct bio_set *bs)
Linus Torvalds1da177e2005-04-16 15:20:36 -070080{
81 struct bio_vec *bvl;
Linus Torvalds1da177e2005-04-16 15:20:36 -070082
83 /*
84 * see comment near bvec_array define!
85 */
86 switch (nr) {
87 case 1 : *idx = 0; break;
88 case 2 ... 4: *idx = 1; break;
89 case 5 ... 16: *idx = 2; break;
90 case 17 ... 64: *idx = 3; break;
91 case 65 ... 128: *idx = 4; break;
92 case 129 ... BIO_MAX_PAGES: *idx = 5; break;
93 default:
94 return NULL;
95 }
96 /*
97 * idx now points to the pool we want to allocate from
98 */
99
Linus Torvalds1da177e2005-04-16 15:20:36 -0700100 bvl = mempool_alloc(bs->bvec_pools[*idx], gfp_mask);
Andreas Mohrbf02c082006-10-11 01:22:24 -0700101 if (bvl) {
102 struct biovec_slab *bp = bvec_slabs + *idx;
103
Linus Torvalds1da177e2005-04-16 15:20:36 -0700104 memset(bvl, 0, bp->nr_vecs * sizeof(struct bio_vec));
Andreas Mohrbf02c082006-10-11 01:22:24 -0700105 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700106
107 return bvl;
108}
109
Peter Osterlund36763472005-09-06 15:16:42 -0700110void bio_free(struct bio *bio, struct bio_set *bio_set)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700111{
112 const int pool_idx = BIO_POOL_IDX(bio);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700113
114 BIO_BUG_ON(pool_idx >= BIOVEC_NR_POOLS);
115
Peter Osterlund36763472005-09-06 15:16:42 -0700116 mempool_free(bio->bi_io_vec, bio_set->bvec_pools[pool_idx]);
117 mempool_free(bio, bio_set->bio_pool);
118}
119
120/*
121 * default destructor for a bio allocated with bio_alloc_bioset()
122 */
123static void bio_fs_destructor(struct bio *bio)
124{
125 bio_free(bio, fs_bio_set);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700126}
127
Arjan van de Ven858119e2006-01-14 13:20:43 -0800128void bio_init(struct bio *bio)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700129{
130 bio->bi_next = NULL;
Jens Axboe0ea60b52006-01-09 14:45:10 +0100131 bio->bi_bdev = NULL;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700132 bio->bi_flags = 1 << BIO_UPTODATE;
133 bio->bi_rw = 0;
134 bio->bi_vcnt = 0;
135 bio->bi_idx = 0;
136 bio->bi_phys_segments = 0;
137 bio->bi_hw_segments = 0;
138 bio->bi_hw_front_size = 0;
139 bio->bi_hw_back_size = 0;
140 bio->bi_size = 0;
141 bio->bi_max_vecs = 0;
142 bio->bi_end_io = NULL;
143 atomic_set(&bio->bi_cnt, 1);
144 bio->bi_private = NULL;
145}
146
147/**
148 * bio_alloc_bioset - allocate a bio for I/O
149 * @gfp_mask: the GFP_ mask given to the slab allocator
150 * @nr_iovecs: number of iovecs to pre-allocate
Martin Waitz67be2dd2005-05-01 08:59:26 -0700151 * @bs: the bio_set to allocate from
Linus Torvalds1da177e2005-04-16 15:20:36 -0700152 *
153 * Description:
154 * bio_alloc_bioset will first try it's on mempool to satisfy the allocation.
155 * If %__GFP_WAIT is set then we will block on the internal pool waiting
156 * for a &struct bio to become free.
157 *
158 * allocate bio and iovecs from the memory pools specified by the
159 * bio_set structure.
160 **/
Al Virodd0fc662005-10-07 07:46:04 +0100161struct bio *bio_alloc_bioset(gfp_t gfp_mask, int nr_iovecs, struct bio_set *bs)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700162{
163 struct bio *bio = mempool_alloc(bs->bio_pool, gfp_mask);
164
165 if (likely(bio)) {
166 struct bio_vec *bvl = NULL;
167
168 bio_init(bio);
169 if (likely(nr_iovecs)) {
Andreas Mohrbf02c082006-10-11 01:22:24 -0700170 unsigned long idx = 0; /* shut up gcc */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700171
172 bvl = bvec_alloc_bs(gfp_mask, nr_iovecs, &idx, bs);
173 if (unlikely(!bvl)) {
174 mempool_free(bio, bs->bio_pool);
175 bio = NULL;
176 goto out;
177 }
178 bio->bi_flags |= idx << BIO_POOL_OFFSET;
179 bio->bi_max_vecs = bvec_slabs[idx].nr_vecs;
180 }
181 bio->bi_io_vec = bvl;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700182 }
183out:
184 return bio;
185}
186
Al Virodd0fc662005-10-07 07:46:04 +0100187struct bio *bio_alloc(gfp_t gfp_mask, int nr_iovecs)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700188{
Peter Osterlund36763472005-09-06 15:16:42 -0700189 struct bio *bio = bio_alloc_bioset(gfp_mask, nr_iovecs, fs_bio_set);
190
191 if (bio)
192 bio->bi_destructor = bio_fs_destructor;
193
194 return bio;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700195}
196
197void zero_fill_bio(struct bio *bio)
198{
199 unsigned long flags;
200 struct bio_vec *bv;
201 int i;
202
203 bio_for_each_segment(bv, bio, i) {
204 char *data = bvec_kmap_irq(bv, &flags);
205 memset(data, 0, bv->bv_len);
206 flush_dcache_page(bv->bv_page);
207 bvec_kunmap_irq(data, &flags);
208 }
209}
210EXPORT_SYMBOL(zero_fill_bio);
211
212/**
213 * bio_put - release a reference to a bio
214 * @bio: bio to release reference to
215 *
216 * Description:
217 * Put a reference to a &struct bio, either one you have gotten with
218 * bio_alloc or bio_get. The last put of a bio will free it.
219 **/
220void bio_put(struct bio *bio)
221{
222 BIO_BUG_ON(!atomic_read(&bio->bi_cnt));
223
224 /*
225 * last put frees it
226 */
227 if (atomic_dec_and_test(&bio->bi_cnt)) {
228 bio->bi_next = NULL;
229 bio->bi_destructor(bio);
230 }
231}
232
Jens Axboe165125e2007-07-24 09:28:11 +0200233inline int bio_phys_segments(struct request_queue *q, struct bio *bio)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700234{
235 if (unlikely(!bio_flagged(bio, BIO_SEG_VALID)))
236 blk_recount_segments(q, bio);
237
238 return bio->bi_phys_segments;
239}
240
Jens Axboe165125e2007-07-24 09:28:11 +0200241inline int bio_hw_segments(struct request_queue *q, struct bio *bio)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700242{
243 if (unlikely(!bio_flagged(bio, BIO_SEG_VALID)))
244 blk_recount_segments(q, bio);
245
246 return bio->bi_hw_segments;
247}
248
249/**
250 * __bio_clone - clone a bio
251 * @bio: destination bio
252 * @bio_src: bio to clone
253 *
254 * Clone a &bio. Caller will own the returned bio, but not
255 * the actual data it points to. Reference count of returned
256 * bio will be one.
257 */
Arjan van de Ven858119e2006-01-14 13:20:43 -0800258void __bio_clone(struct bio *bio, struct bio *bio_src)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700259{
Jens Axboe165125e2007-07-24 09:28:11 +0200260 struct request_queue *q = bdev_get_queue(bio_src->bi_bdev);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700261
Andrew Mortone525e152005-08-07 09:42:12 -0700262 memcpy(bio->bi_io_vec, bio_src->bi_io_vec,
263 bio_src->bi_max_vecs * sizeof(struct bio_vec));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700264
265 bio->bi_sector = bio_src->bi_sector;
266 bio->bi_bdev = bio_src->bi_bdev;
267 bio->bi_flags |= 1 << BIO_CLONED;
268 bio->bi_rw = bio_src->bi_rw;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700269 bio->bi_vcnt = bio_src->bi_vcnt;
270 bio->bi_size = bio_src->bi_size;
Andrew Mortona5453be2005-07-28 01:07:18 -0700271 bio->bi_idx = bio_src->bi_idx;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700272 bio_phys_segments(q, bio);
273 bio_hw_segments(q, bio);
274}
275
276/**
277 * bio_clone - clone a bio
278 * @bio: bio to clone
279 * @gfp_mask: allocation priority
280 *
281 * Like __bio_clone, only also allocates the returned bio
282 */
Al Virodd0fc662005-10-07 07:46:04 +0100283struct bio *bio_clone(struct bio *bio, gfp_t gfp_mask)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700284{
285 struct bio *b = bio_alloc_bioset(gfp_mask, bio->bi_max_vecs, fs_bio_set);
286
Peter Osterlund36763472005-09-06 15:16:42 -0700287 if (b) {
288 b->bi_destructor = bio_fs_destructor;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700289 __bio_clone(b, bio);
Peter Osterlund36763472005-09-06 15:16:42 -0700290 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700291
292 return b;
293}
294
295/**
296 * bio_get_nr_vecs - return approx number of vecs
297 * @bdev: I/O target
298 *
299 * Return the approximate number of pages we can send to this target.
300 * There's no guarantee that you will be able to fit this number of pages
301 * into a bio, it does not account for dynamic restrictions that vary
302 * on offset.
303 */
304int bio_get_nr_vecs(struct block_device *bdev)
305{
Jens Axboe165125e2007-07-24 09:28:11 +0200306 struct request_queue *q = bdev_get_queue(bdev);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700307 int nr_pages;
308
309 nr_pages = ((q->max_sectors << 9) + PAGE_SIZE - 1) >> PAGE_SHIFT;
310 if (nr_pages > q->max_phys_segments)
311 nr_pages = q->max_phys_segments;
312 if (nr_pages > q->max_hw_segments)
313 nr_pages = q->max_hw_segments;
314
315 return nr_pages;
316}
317
Jens Axboe165125e2007-07-24 09:28:11 +0200318static int __bio_add_page(struct request_queue *q, struct bio *bio, struct page
Mike Christiedefd94b2005-12-05 02:37:06 -0600319 *page, unsigned int len, unsigned int offset,
320 unsigned short max_sectors)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700321{
322 int retried_segments = 0;
323 struct bio_vec *bvec;
324
325 /*
326 * cloned bio must not modify vec list
327 */
328 if (unlikely(bio_flagged(bio, BIO_CLONED)))
329 return 0;
330
Jens Axboe80cfd542006-01-06 09:43:28 +0100331 if (((bio->bi_size + len) >> 9) > max_sectors)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700332 return 0;
333
Jens Axboe80cfd542006-01-06 09:43:28 +0100334 /*
335 * For filesystems with a blocksize smaller than the pagesize
336 * we will often be called with the same page as last time and
337 * a consecutive offset. Optimize this special case.
338 */
339 if (bio->bi_vcnt > 0) {
340 struct bio_vec *prev = &bio->bi_io_vec[bio->bi_vcnt - 1];
341
342 if (page == prev->bv_page &&
343 offset == prev->bv_offset + prev->bv_len) {
344 prev->bv_len += len;
345 if (q->merge_bvec_fn &&
346 q->merge_bvec_fn(q, bio, prev) < len) {
347 prev->bv_len -= len;
348 return 0;
349 }
350
351 goto done;
352 }
353 }
354
355 if (bio->bi_vcnt >= bio->bi_max_vecs)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700356 return 0;
357
358 /*
359 * we might lose a segment or two here, but rather that than
360 * make this too complex.
361 */
362
363 while (bio->bi_phys_segments >= q->max_phys_segments
364 || bio->bi_hw_segments >= q->max_hw_segments
365 || BIOVEC_VIRT_OVERSIZE(bio->bi_size)) {
366
367 if (retried_segments)
368 return 0;
369
370 retried_segments = 1;
371 blk_recount_segments(q, bio);
372 }
373
374 /*
375 * setup the new entry, we might clear it again later if we
376 * cannot add the page
377 */
378 bvec = &bio->bi_io_vec[bio->bi_vcnt];
379 bvec->bv_page = page;
380 bvec->bv_len = len;
381 bvec->bv_offset = offset;
382
383 /*
384 * if queue has other restrictions (eg varying max sector size
385 * depending on offset), it can specify a merge_bvec_fn in the
386 * queue to get further control
387 */
388 if (q->merge_bvec_fn) {
389 /*
390 * merge_bvec_fn() returns number of bytes it can accept
391 * at this offset
392 */
393 if (q->merge_bvec_fn(q, bio, bvec) < len) {
394 bvec->bv_page = NULL;
395 bvec->bv_len = 0;
396 bvec->bv_offset = 0;
397 return 0;
398 }
399 }
400
401 /* If we may be able to merge these biovecs, force a recount */
402 if (bio->bi_vcnt && (BIOVEC_PHYS_MERGEABLE(bvec-1, bvec) ||
403 BIOVEC_VIRT_MERGEABLE(bvec-1, bvec)))
404 bio->bi_flags &= ~(1 << BIO_SEG_VALID);
405
406 bio->bi_vcnt++;
407 bio->bi_phys_segments++;
408 bio->bi_hw_segments++;
Jens Axboe80cfd542006-01-06 09:43:28 +0100409 done:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700410 bio->bi_size += len;
411 return len;
412}
413
414/**
Mike Christie6e68af62005-11-11 05:30:27 -0600415 * bio_add_pc_page - attempt to add page to bio
Jens Axboefddfdea2006-01-31 15:24:34 +0100416 * @q: the target queue
Mike Christie6e68af62005-11-11 05:30:27 -0600417 * @bio: destination bio
418 * @page: page to add
419 * @len: vec entry length
420 * @offset: vec entry offset
421 *
422 * Attempt to add a page to the bio_vec maplist. This can fail for a
423 * number of reasons, such as the bio being full or target block
424 * device limitations. The target block device must allow bio's
425 * smaller than PAGE_SIZE, so it is always possible to add a single
426 * page to an empty bio. This should only be used by REQ_PC bios.
427 */
Jens Axboe165125e2007-07-24 09:28:11 +0200428int bio_add_pc_page(struct request_queue *q, struct bio *bio, struct page *page,
Mike Christie6e68af62005-11-11 05:30:27 -0600429 unsigned int len, unsigned int offset)
430{
Mike Christiedefd94b2005-12-05 02:37:06 -0600431 return __bio_add_page(q, bio, page, len, offset, q->max_hw_sectors);
Mike Christie6e68af62005-11-11 05:30:27 -0600432}
433
434/**
Linus Torvalds1da177e2005-04-16 15:20:36 -0700435 * bio_add_page - attempt to add page to bio
436 * @bio: destination bio
437 * @page: page to add
438 * @len: vec entry length
439 * @offset: vec entry offset
440 *
441 * Attempt to add a page to the bio_vec maplist. This can fail for a
442 * number of reasons, such as the bio being full or target block
443 * device limitations. The target block device must allow bio's
444 * smaller than PAGE_SIZE, so it is always possible to add a single
445 * page to an empty bio.
446 */
447int bio_add_page(struct bio *bio, struct page *page, unsigned int len,
448 unsigned int offset)
449{
Mike Christiedefd94b2005-12-05 02:37:06 -0600450 struct request_queue *q = bdev_get_queue(bio->bi_bdev);
451 return __bio_add_page(q, bio, page, len, offset, q->max_sectors);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700452}
453
454struct bio_map_data {
455 struct bio_vec *iovecs;
456 void __user *userptr;
457};
458
459static void bio_set_map_data(struct bio_map_data *bmd, struct bio *bio)
460{
461 memcpy(bmd->iovecs, bio->bi_io_vec, sizeof(struct bio_vec) * bio->bi_vcnt);
462 bio->bi_private = bmd;
463}
464
465static void bio_free_map_data(struct bio_map_data *bmd)
466{
467 kfree(bmd->iovecs);
468 kfree(bmd);
469}
470
471static struct bio_map_data *bio_alloc_map_data(int nr_segs)
472{
473 struct bio_map_data *bmd = kmalloc(sizeof(*bmd), GFP_KERNEL);
474
475 if (!bmd)
476 return NULL;
477
478 bmd->iovecs = kmalloc(sizeof(struct bio_vec) * nr_segs, GFP_KERNEL);
479 if (bmd->iovecs)
480 return bmd;
481
482 kfree(bmd);
483 return NULL;
484}
485
486/**
487 * bio_uncopy_user - finish previously mapped bio
488 * @bio: bio being terminated
489 *
490 * Free pages allocated from bio_copy_user() and write back data
491 * to user space in case of a read.
492 */
493int bio_uncopy_user(struct bio *bio)
494{
495 struct bio_map_data *bmd = bio->bi_private;
496 const int read = bio_data_dir(bio) == READ;
497 struct bio_vec *bvec;
498 int i, ret = 0;
499
500 __bio_for_each_segment(bvec, bio, i, 0) {
501 char *addr = page_address(bvec->bv_page);
502 unsigned int len = bmd->iovecs[i].bv_len;
503
504 if (read && !ret && copy_to_user(bmd->userptr, addr, len))
505 ret = -EFAULT;
506
507 __free_page(bvec->bv_page);
508 bmd->userptr += len;
509 }
510 bio_free_map_data(bmd);
511 bio_put(bio);
512 return ret;
513}
514
515/**
516 * bio_copy_user - copy user data to bio
517 * @q: destination block queue
518 * @uaddr: start of user address
519 * @len: length in bytes
520 * @write_to_vm: bool indicating writing to pages or not
521 *
522 * Prepares and returns a bio for indirect user io, bouncing data
523 * to/from kernel pages as necessary. Must be paired with
524 * call bio_uncopy_user() on io completion.
525 */
Jens Axboe165125e2007-07-24 09:28:11 +0200526struct bio *bio_copy_user(struct request_queue *q, unsigned long uaddr,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700527 unsigned int len, int write_to_vm)
528{
529 unsigned long end = (uaddr + len + PAGE_SIZE - 1) >> PAGE_SHIFT;
530 unsigned long start = uaddr >> PAGE_SHIFT;
531 struct bio_map_data *bmd;
532 struct bio_vec *bvec;
533 struct page *page;
534 struct bio *bio;
535 int i, ret;
536
537 bmd = bio_alloc_map_data(end - start);
538 if (!bmd)
539 return ERR_PTR(-ENOMEM);
540
541 bmd->userptr = (void __user *) uaddr;
542
543 ret = -ENOMEM;
544 bio = bio_alloc(GFP_KERNEL, end - start);
545 if (!bio)
546 goto out_bmd;
547
548 bio->bi_rw |= (!write_to_vm << BIO_RW);
549
550 ret = 0;
551 while (len) {
552 unsigned int bytes = PAGE_SIZE;
553
554 if (bytes > len)
555 bytes = len;
556
557 page = alloc_page(q->bounce_gfp | GFP_KERNEL);
558 if (!page) {
559 ret = -ENOMEM;
560 break;
561 }
562
Mike Christie0e75f902006-12-01 10:40:55 +0100563 if (bio_add_pc_page(q, bio, page, bytes, 0) < bytes)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700564 break;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700565
566 len -= bytes;
567 }
568
569 if (ret)
570 goto cleanup;
571
572 /*
573 * success
574 */
575 if (!write_to_vm) {
576 char __user *p = (char __user *) uaddr;
577
578 /*
579 * for a write, copy in data to kernel pages
580 */
581 ret = -EFAULT;
582 bio_for_each_segment(bvec, bio, i) {
583 char *addr = page_address(bvec->bv_page);
584
585 if (copy_from_user(addr, p, bvec->bv_len))
586 goto cleanup;
587 p += bvec->bv_len;
588 }
589 }
590
591 bio_set_map_data(bmd, bio);
592 return bio;
593cleanup:
594 bio_for_each_segment(bvec, bio, i)
595 __free_page(bvec->bv_page);
596
597 bio_put(bio);
598out_bmd:
599 bio_free_map_data(bmd);
600 return ERR_PTR(ret);
601}
602
Jens Axboe165125e2007-07-24 09:28:11 +0200603static struct bio *__bio_map_user_iov(struct request_queue *q,
James Bottomley f1970ba2005-06-20 14:06:52 +0200604 struct block_device *bdev,
605 struct sg_iovec *iov, int iov_count,
606 int write_to_vm)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700607{
James Bottomley f1970ba2005-06-20 14:06:52 +0200608 int i, j;
609 int nr_pages = 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700610 struct page **pages;
611 struct bio *bio;
James Bottomley f1970ba2005-06-20 14:06:52 +0200612 int cur_page = 0;
613 int ret, offset;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700614
James Bottomley f1970ba2005-06-20 14:06:52 +0200615 for (i = 0; i < iov_count; i++) {
616 unsigned long uaddr = (unsigned long)iov[i].iov_base;
617 unsigned long len = iov[i].iov_len;
618 unsigned long end = (uaddr + len + PAGE_SIZE - 1) >> PAGE_SHIFT;
619 unsigned long start = uaddr >> PAGE_SHIFT;
620
621 nr_pages += end - start;
622 /*
Mike Christiead2d7222006-12-01 10:40:20 +0100623 * buffer must be aligned to at least hardsector size for now
James Bottomley f1970ba2005-06-20 14:06:52 +0200624 */
Mike Christiead2d7222006-12-01 10:40:20 +0100625 if (uaddr & queue_dma_alignment(q))
James Bottomley f1970ba2005-06-20 14:06:52 +0200626 return ERR_PTR(-EINVAL);
627 }
628
629 if (!nr_pages)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700630 return ERR_PTR(-EINVAL);
631
632 bio = bio_alloc(GFP_KERNEL, nr_pages);
633 if (!bio)
634 return ERR_PTR(-ENOMEM);
635
636 ret = -ENOMEM;
Oliver Neukum11b0b5a2006-03-25 03:08:13 -0800637 pages = kcalloc(nr_pages, sizeof(struct page *), GFP_KERNEL);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700638 if (!pages)
639 goto out;
640
James Bottomley f1970ba2005-06-20 14:06:52 +0200641 for (i = 0; i < iov_count; i++) {
642 unsigned long uaddr = (unsigned long)iov[i].iov_base;
643 unsigned long len = iov[i].iov_len;
644 unsigned long end = (uaddr + len + PAGE_SIZE - 1) >> PAGE_SHIFT;
645 unsigned long start = uaddr >> PAGE_SHIFT;
646 const int local_nr_pages = end - start;
647 const int page_limit = cur_page + local_nr_pages;
648
649 down_read(&current->mm->mmap_sem);
650 ret = get_user_pages(current, current->mm, uaddr,
651 local_nr_pages,
652 write_to_vm, 0, &pages[cur_page], NULL);
653 up_read(&current->mm->mmap_sem);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700654
Jens Axboe99172152006-06-16 13:02:29 +0200655 if (ret < local_nr_pages) {
656 ret = -EFAULT;
James Bottomley f1970ba2005-06-20 14:06:52 +0200657 goto out_unmap;
Jens Axboe99172152006-06-16 13:02:29 +0200658 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700659
James Bottomley f1970ba2005-06-20 14:06:52 +0200660 offset = uaddr & ~PAGE_MASK;
661 for (j = cur_page; j < page_limit; j++) {
662 unsigned int bytes = PAGE_SIZE - offset;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700663
James Bottomley f1970ba2005-06-20 14:06:52 +0200664 if (len <= 0)
665 break;
666
667 if (bytes > len)
668 bytes = len;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700669
James Bottomley f1970ba2005-06-20 14:06:52 +0200670 /*
671 * sorry...
672 */
Mike Christiedefd94b2005-12-05 02:37:06 -0600673 if (bio_add_pc_page(q, bio, pages[j], bytes, offset) <
674 bytes)
James Bottomley f1970ba2005-06-20 14:06:52 +0200675 break;
676
677 len -= bytes;
678 offset = 0;
679 }
680
681 cur_page = j;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700682 /*
James Bottomley f1970ba2005-06-20 14:06:52 +0200683 * release the pages we didn't map into the bio, if any
Linus Torvalds1da177e2005-04-16 15:20:36 -0700684 */
James Bottomley f1970ba2005-06-20 14:06:52 +0200685 while (j < page_limit)
686 page_cache_release(pages[j++]);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700687 }
688
Linus Torvalds1da177e2005-04-16 15:20:36 -0700689 kfree(pages);
690
691 /*
692 * set data direction, and check if mapped pages need bouncing
693 */
694 if (!write_to_vm)
695 bio->bi_rw |= (1 << BIO_RW);
696
James Bottomley f1970ba2005-06-20 14:06:52 +0200697 bio->bi_bdev = bdev;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700698 bio->bi_flags |= (1 << BIO_USER_MAPPED);
699 return bio;
James Bottomley f1970ba2005-06-20 14:06:52 +0200700
701 out_unmap:
702 for (i = 0; i < nr_pages; i++) {
703 if(!pages[i])
704 break;
705 page_cache_release(pages[i]);
706 }
707 out:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700708 kfree(pages);
709 bio_put(bio);
710 return ERR_PTR(ret);
711}
712
713/**
714 * bio_map_user - map user address into bio
Jens Axboe165125e2007-07-24 09:28:11 +0200715 * @q: the struct request_queue for the bio
Linus Torvalds1da177e2005-04-16 15:20:36 -0700716 * @bdev: destination block device
717 * @uaddr: start of user address
718 * @len: length in bytes
719 * @write_to_vm: bool indicating writing to pages or not
720 *
721 * Map the user space address into a bio suitable for io to a block
722 * device. Returns an error pointer in case of error.
723 */
Jens Axboe165125e2007-07-24 09:28:11 +0200724struct bio *bio_map_user(struct request_queue *q, struct block_device *bdev,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700725 unsigned long uaddr, unsigned int len, int write_to_vm)
726{
James Bottomley f1970ba2005-06-20 14:06:52 +0200727 struct sg_iovec iov;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700728
viro@ZenIV.linux.org.uk3f703532005-09-09 16:53:56 +0100729 iov.iov_base = (void __user *)uaddr;
James Bottomley f1970ba2005-06-20 14:06:52 +0200730 iov.iov_len = len;
731
732 return bio_map_user_iov(q, bdev, &iov, 1, write_to_vm);
733}
734
735/**
736 * bio_map_user_iov - map user sg_iovec table into bio
Jens Axboe165125e2007-07-24 09:28:11 +0200737 * @q: the struct request_queue for the bio
James Bottomley f1970ba2005-06-20 14:06:52 +0200738 * @bdev: destination block device
739 * @iov: the iovec.
740 * @iov_count: number of elements in the iovec
741 * @write_to_vm: bool indicating writing to pages or not
742 *
743 * Map the user space address into a bio suitable for io to a block
744 * device. Returns an error pointer in case of error.
745 */
Jens Axboe165125e2007-07-24 09:28:11 +0200746struct bio *bio_map_user_iov(struct request_queue *q, struct block_device *bdev,
James Bottomley f1970ba2005-06-20 14:06:52 +0200747 struct sg_iovec *iov, int iov_count,
748 int write_to_vm)
749{
750 struct bio *bio;
James Bottomley f1970ba2005-06-20 14:06:52 +0200751
752 bio = __bio_map_user_iov(q, bdev, iov, iov_count, write_to_vm);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700753
754 if (IS_ERR(bio))
755 return bio;
756
757 /*
758 * subtle -- if __bio_map_user() ended up bouncing a bio,
759 * it would normally disappear when its bi_end_io is run.
760 * however, we need it for the unmap, so grab an extra
761 * reference to it
762 */
763 bio_get(bio);
764
Mike Christie0e75f902006-12-01 10:40:55 +0100765 return bio;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700766}
767
768static void __bio_unmap_user(struct bio *bio)
769{
770 struct bio_vec *bvec;
771 int i;
772
773 /*
774 * make sure we dirty pages we wrote to
775 */
776 __bio_for_each_segment(bvec, bio, i, 0) {
777 if (bio_data_dir(bio) == READ)
778 set_page_dirty_lock(bvec->bv_page);
779
780 page_cache_release(bvec->bv_page);
781 }
782
783 bio_put(bio);
784}
785
786/**
787 * bio_unmap_user - unmap a bio
788 * @bio: the bio being unmapped
789 *
790 * Unmap a bio previously mapped by bio_map_user(). Must be called with
791 * a process context.
792 *
793 * bio_unmap_user() may sleep.
794 */
795void bio_unmap_user(struct bio *bio)
796{
797 __bio_unmap_user(bio);
798 bio_put(bio);
799}
800
NeilBrown6712ecf2007-09-27 12:47:43 +0200801static void bio_map_kern_endio(struct bio *bio, int err)
Jens Axboeb8238252005-06-20 14:05:27 +0200802{
Jens Axboeb8238252005-06-20 14:05:27 +0200803 bio_put(bio);
Jens Axboeb8238252005-06-20 14:05:27 +0200804}
805
806
Jens Axboe165125e2007-07-24 09:28:11 +0200807static struct bio *__bio_map_kern(struct request_queue *q, void *data,
Al Viro27496a82005-10-21 03:20:48 -0400808 unsigned int len, gfp_t gfp_mask)
Mike Christie df46b9a2005-06-20 14:04:44 +0200809{
810 unsigned long kaddr = (unsigned long)data;
811 unsigned long end = (kaddr + len + PAGE_SIZE - 1) >> PAGE_SHIFT;
812 unsigned long start = kaddr >> PAGE_SHIFT;
813 const int nr_pages = end - start;
814 int offset, i;
815 struct bio *bio;
816
817 bio = bio_alloc(gfp_mask, nr_pages);
818 if (!bio)
819 return ERR_PTR(-ENOMEM);
820
821 offset = offset_in_page(kaddr);
822 for (i = 0; i < nr_pages; i++) {
823 unsigned int bytes = PAGE_SIZE - offset;
824
825 if (len <= 0)
826 break;
827
828 if (bytes > len)
829 bytes = len;
830
Mike Christiedefd94b2005-12-05 02:37:06 -0600831 if (bio_add_pc_page(q, bio, virt_to_page(data), bytes,
832 offset) < bytes)
Mike Christie df46b9a2005-06-20 14:04:44 +0200833 break;
834
835 data += bytes;
836 len -= bytes;
837 offset = 0;
838 }
839
Jens Axboeb8238252005-06-20 14:05:27 +0200840 bio->bi_end_io = bio_map_kern_endio;
Mike Christie df46b9a2005-06-20 14:04:44 +0200841 return bio;
842}
843
844/**
845 * bio_map_kern - map kernel address into bio
Jens Axboe165125e2007-07-24 09:28:11 +0200846 * @q: the struct request_queue for the bio
Mike Christie df46b9a2005-06-20 14:04:44 +0200847 * @data: pointer to buffer to map
848 * @len: length in bytes
849 * @gfp_mask: allocation flags for bio allocation
850 *
851 * Map the kernel address into a bio suitable for io to a block
852 * device. Returns an error pointer in case of error.
853 */
Jens Axboe165125e2007-07-24 09:28:11 +0200854struct bio *bio_map_kern(struct request_queue *q, void *data, unsigned int len,
Al Viro27496a82005-10-21 03:20:48 -0400855 gfp_t gfp_mask)
Mike Christie df46b9a2005-06-20 14:04:44 +0200856{
857 struct bio *bio;
858
859 bio = __bio_map_kern(q, data, len, gfp_mask);
860 if (IS_ERR(bio))
861 return bio;
862
863 if (bio->bi_size == len)
864 return bio;
865
866 /*
867 * Don't support partial mappings.
868 */
869 bio_put(bio);
870 return ERR_PTR(-EINVAL);
871}
872
Linus Torvalds1da177e2005-04-16 15:20:36 -0700873/*
874 * bio_set_pages_dirty() and bio_check_pages_dirty() are support functions
875 * for performing direct-IO in BIOs.
876 *
877 * The problem is that we cannot run set_page_dirty() from interrupt context
878 * because the required locks are not interrupt-safe. So what we can do is to
879 * mark the pages dirty _before_ performing IO. And in interrupt context,
880 * check that the pages are still dirty. If so, fine. If not, redirty them
881 * in process context.
882 *
883 * We special-case compound pages here: normally this means reads into hugetlb
884 * pages. The logic in here doesn't really work right for compound pages
885 * because the VM does not uniformly chase down the head page in all cases.
886 * But dirtiness of compound pages is pretty meaningless anyway: the VM doesn't
887 * handle them at all. So we skip compound pages here at an early stage.
888 *
889 * Note that this code is very hard to test under normal circumstances because
890 * direct-io pins the pages with get_user_pages(). This makes
891 * is_page_cache_freeable return false, and the VM will not clean the pages.
892 * But other code (eg, pdflush) could clean the pages if they are mapped
893 * pagecache.
894 *
895 * Simply disabling the call to bio_set_pages_dirty() is a good way to test the
896 * deferred bio dirtying paths.
897 */
898
899/*
900 * bio_set_pages_dirty() will mark all the bio's pages as dirty.
901 */
902void bio_set_pages_dirty(struct bio *bio)
903{
904 struct bio_vec *bvec = bio->bi_io_vec;
905 int i;
906
907 for (i = 0; i < bio->bi_vcnt; i++) {
908 struct page *page = bvec[i].bv_page;
909
910 if (page && !PageCompound(page))
911 set_page_dirty_lock(page);
912 }
913}
914
Chen, Kenneth We61c9012006-12-13 00:34:36 -0800915void bio_release_pages(struct bio *bio)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700916{
917 struct bio_vec *bvec = bio->bi_io_vec;
918 int i;
919
920 for (i = 0; i < bio->bi_vcnt; i++) {
921 struct page *page = bvec[i].bv_page;
922
923 if (page)
924 put_page(page);
925 }
926}
927
928/*
929 * bio_check_pages_dirty() will check that all the BIO's pages are still dirty.
930 * If they are, then fine. If, however, some pages are clean then they must
931 * have been written out during the direct-IO read. So we take another ref on
932 * the BIO and the offending pages and re-dirty the pages in process context.
933 *
934 * It is expected that bio_check_pages_dirty() will wholly own the BIO from
935 * here on. It will run one page_cache_release() against each page and will
936 * run one bio_put() against the BIO.
937 */
938
David Howells65f27f32006-11-22 14:55:48 +0000939static void bio_dirty_fn(struct work_struct *work);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700940
David Howells65f27f32006-11-22 14:55:48 +0000941static DECLARE_WORK(bio_dirty_work, bio_dirty_fn);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700942static DEFINE_SPINLOCK(bio_dirty_lock);
943static struct bio *bio_dirty_list;
944
945/*
946 * This runs in process context
947 */
David Howells65f27f32006-11-22 14:55:48 +0000948static void bio_dirty_fn(struct work_struct *work)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700949{
950 unsigned long flags;
951 struct bio *bio;
952
953 spin_lock_irqsave(&bio_dirty_lock, flags);
954 bio = bio_dirty_list;
955 bio_dirty_list = NULL;
956 spin_unlock_irqrestore(&bio_dirty_lock, flags);
957
958 while (bio) {
959 struct bio *next = bio->bi_private;
960
961 bio_set_pages_dirty(bio);
962 bio_release_pages(bio);
963 bio_put(bio);
964 bio = next;
965 }
966}
967
968void bio_check_pages_dirty(struct bio *bio)
969{
970 struct bio_vec *bvec = bio->bi_io_vec;
971 int nr_clean_pages = 0;
972 int i;
973
974 for (i = 0; i < bio->bi_vcnt; i++) {
975 struct page *page = bvec[i].bv_page;
976
977 if (PageDirty(page) || PageCompound(page)) {
978 page_cache_release(page);
979 bvec[i].bv_page = NULL;
980 } else {
981 nr_clean_pages++;
982 }
983 }
984
985 if (nr_clean_pages) {
986 unsigned long flags;
987
988 spin_lock_irqsave(&bio_dirty_lock, flags);
989 bio->bi_private = bio_dirty_list;
990 bio_dirty_list = bio;
991 spin_unlock_irqrestore(&bio_dirty_lock, flags);
992 schedule_work(&bio_dirty_work);
993 } else {
994 bio_put(bio);
995 }
996}
997
998/**
999 * bio_endio - end I/O on a bio
1000 * @bio: bio
Linus Torvalds1da177e2005-04-16 15:20:36 -07001001 * @error: error, if any
1002 *
1003 * Description:
NeilBrown6712ecf2007-09-27 12:47:43 +02001004 * bio_endio() will end I/O on the whole bio. bio_endio() is the
NeilBrown5bb23a62007-09-27 12:46:13 +02001005 * preferred way to end I/O on a bio, it takes care of clearing
1006 * BIO_UPTODATE on error. @error is 0 on success, and and one of the
1007 * established -Exxxx (-EIO, for instance) error values in case
1008 * something went wrong. Noone should call bi_end_io() directly on a
1009 * bio unless they own it and thus know that it has an end_io
1010 * function.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001011 **/
NeilBrown6712ecf2007-09-27 12:47:43 +02001012void bio_endio(struct bio *bio, int error)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001013{
1014 if (error)
1015 clear_bit(BIO_UPTODATE, &bio->bi_flags);
NeilBrown9cc54d42007-09-27 12:46:12 +02001016 else if (!test_bit(BIO_UPTODATE, &bio->bi_flags))
1017 error = -EIO;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001018
NeilBrown5bb23a62007-09-27 12:46:13 +02001019 if (bio->bi_end_io)
NeilBrown6712ecf2007-09-27 12:47:43 +02001020 bio->bi_end_io(bio, error);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001021}
1022
1023void bio_pair_release(struct bio_pair *bp)
1024{
1025 if (atomic_dec_and_test(&bp->cnt)) {
1026 struct bio *master = bp->bio1.bi_private;
1027
NeilBrown6712ecf2007-09-27 12:47:43 +02001028 bio_endio(master, bp->error);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001029 mempool_free(bp, bp->bio2.bi_private);
1030 }
1031}
1032
NeilBrown6712ecf2007-09-27 12:47:43 +02001033static void bio_pair_end_1(struct bio *bi, int err)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001034{
1035 struct bio_pair *bp = container_of(bi, struct bio_pair, bio1);
1036
1037 if (err)
1038 bp->error = err;
1039
Linus Torvalds1da177e2005-04-16 15:20:36 -07001040 bio_pair_release(bp);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001041}
1042
NeilBrown6712ecf2007-09-27 12:47:43 +02001043static void bio_pair_end_2(struct bio *bi, int err)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001044{
1045 struct bio_pair *bp = container_of(bi, struct bio_pair, bio2);
1046
1047 if (err)
1048 bp->error = err;
1049
Linus Torvalds1da177e2005-04-16 15:20:36 -07001050 bio_pair_release(bp);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001051}
1052
1053/*
1054 * split a bio - only worry about a bio with a single page
1055 * in it's iovec
1056 */
1057struct bio_pair *bio_split(struct bio *bi, mempool_t *pool, int first_sectors)
1058{
1059 struct bio_pair *bp = mempool_alloc(pool, GFP_NOIO);
1060
1061 if (!bp)
1062 return bp;
1063
Jens Axboe2056a782006-03-23 20:00:26 +01001064 blk_add_trace_pdu_int(bdev_get_queue(bi->bi_bdev), BLK_TA_SPLIT, bi,
1065 bi->bi_sector + first_sectors);
1066
Linus Torvalds1da177e2005-04-16 15:20:36 -07001067 BUG_ON(bi->bi_vcnt != 1);
1068 BUG_ON(bi->bi_idx != 0);
1069 atomic_set(&bp->cnt, 3);
1070 bp->error = 0;
1071 bp->bio1 = *bi;
1072 bp->bio2 = *bi;
1073 bp->bio2.bi_sector += first_sectors;
1074 bp->bio2.bi_size -= first_sectors << 9;
1075 bp->bio1.bi_size = first_sectors << 9;
1076
1077 bp->bv1 = bi->bi_io_vec[0];
1078 bp->bv2 = bi->bi_io_vec[0];
1079 bp->bv2.bv_offset += first_sectors << 9;
1080 bp->bv2.bv_len -= first_sectors << 9;
1081 bp->bv1.bv_len = first_sectors << 9;
1082
1083 bp->bio1.bi_io_vec = &bp->bv1;
1084 bp->bio2.bi_io_vec = &bp->bv2;
1085
NeilBrowna2eb0c12006-05-22 22:35:27 -07001086 bp->bio1.bi_max_vecs = 1;
1087 bp->bio2.bi_max_vecs = 1;
1088
Linus Torvalds1da177e2005-04-16 15:20:36 -07001089 bp->bio1.bi_end_io = bio_pair_end_1;
1090 bp->bio2.bi_end_io = bio_pair_end_2;
1091
1092 bp->bio1.bi_private = bi;
1093 bp->bio2.bi_private = pool;
1094
1095 return bp;
1096}
1097
Linus Torvalds1da177e2005-04-16 15:20:36 -07001098
1099/*
1100 * create memory pools for biovec's in a bio_set.
1101 * use the global biovec slabs created for general use.
1102 */
Jens Axboe59725112007-04-02 10:06:42 +02001103static int biovec_create_pools(struct bio_set *bs, int pool_entries)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001104{
1105 int i;
1106
1107 for (i = 0; i < BIOVEC_NR_POOLS; i++) {
1108 struct biovec_slab *bp = bvec_slabs + i;
1109 mempool_t **bvp = bs->bvec_pools + i;
1110
Matthew Dobson93d23412006-03-26 01:37:50 -08001111 *bvp = mempool_create_slab_pool(pool_entries, bp->slab);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001112 if (!*bvp)
1113 return -ENOMEM;
1114 }
1115 return 0;
1116}
1117
1118static void biovec_free_pools(struct bio_set *bs)
1119{
1120 int i;
1121
1122 for (i = 0; i < BIOVEC_NR_POOLS; i++) {
1123 mempool_t *bvp = bs->bvec_pools[i];
1124
1125 if (bvp)
1126 mempool_destroy(bvp);
1127 }
1128
1129}
1130
1131void bioset_free(struct bio_set *bs)
1132{
1133 if (bs->bio_pool)
1134 mempool_destroy(bs->bio_pool);
1135
1136 biovec_free_pools(bs);
1137
1138 kfree(bs);
1139}
1140
Jens Axboe59725112007-04-02 10:06:42 +02001141struct bio_set *bioset_create(int bio_pool_size, int bvec_pool_size)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001142{
Oliver Neukum11b0b5a2006-03-25 03:08:13 -08001143 struct bio_set *bs = kzalloc(sizeof(*bs), GFP_KERNEL);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001144
1145 if (!bs)
1146 return NULL;
1147
Matthew Dobson93d23412006-03-26 01:37:50 -08001148 bs->bio_pool = mempool_create_slab_pool(bio_pool_size, bio_slab);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001149 if (!bs->bio_pool)
1150 goto bad;
1151
Jens Axboe59725112007-04-02 10:06:42 +02001152 if (!biovec_create_pools(bs, bvec_pool_size))
Linus Torvalds1da177e2005-04-16 15:20:36 -07001153 return bs;
1154
1155bad:
1156 bioset_free(bs);
1157 return NULL;
1158}
1159
1160static void __init biovec_init_slabs(void)
1161{
1162 int i;
1163
1164 for (i = 0; i < BIOVEC_NR_POOLS; i++) {
1165 int size;
1166 struct biovec_slab *bvs = bvec_slabs + i;
1167
1168 size = bvs->nr_vecs * sizeof(struct bio_vec);
1169 bvs->slab = kmem_cache_create(bvs->name, size, 0,
Paul Mundt20c2df82007-07-20 10:11:58 +09001170 SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001171 }
1172}
1173
1174static int __init init_bio(void)
1175{
Christoph Lameter0a31bd52007-05-06 14:49:57 -07001176 bio_slab = KMEM_CACHE(bio, SLAB_HWCACHE_ALIGN|SLAB_PANIC);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001177
1178 biovec_init_slabs();
1179
Jens Axboe59725112007-04-02 10:06:42 +02001180 fs_bio_set = bioset_create(BIO_POOL_SIZE, 2);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001181 if (!fs_bio_set)
1182 panic("bio: can't allocate bios\n");
1183
Matthew Dobson0eaae62a2006-03-26 01:37:47 -08001184 bio_split_pool = mempool_create_kmalloc_pool(BIO_SPLIT_ENTRIES,
1185 sizeof(struct bio_pair));
Linus Torvalds1da177e2005-04-16 15:20:36 -07001186 if (!bio_split_pool)
1187 panic("bio: can't create split pool\n");
1188
1189 return 0;
1190}
1191
1192subsys_initcall(init_bio);
1193
1194EXPORT_SYMBOL(bio_alloc);
1195EXPORT_SYMBOL(bio_put);
Peter Osterlund36763472005-09-06 15:16:42 -07001196EXPORT_SYMBOL(bio_free);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001197EXPORT_SYMBOL(bio_endio);
1198EXPORT_SYMBOL(bio_init);
1199EXPORT_SYMBOL(__bio_clone);
1200EXPORT_SYMBOL(bio_clone);
1201EXPORT_SYMBOL(bio_phys_segments);
1202EXPORT_SYMBOL(bio_hw_segments);
1203EXPORT_SYMBOL(bio_add_page);
Mike Christie6e68af62005-11-11 05:30:27 -06001204EXPORT_SYMBOL(bio_add_pc_page);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001205EXPORT_SYMBOL(bio_get_nr_vecs);
Mike Christie df46b9a2005-06-20 14:04:44 +02001206EXPORT_SYMBOL(bio_map_kern);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001207EXPORT_SYMBOL(bio_pair_release);
1208EXPORT_SYMBOL(bio_split);
1209EXPORT_SYMBOL(bio_split_pool);
1210EXPORT_SYMBOL(bio_copy_user);
1211EXPORT_SYMBOL(bio_uncopy_user);
1212EXPORT_SYMBOL(bioset_create);
1213EXPORT_SYMBOL(bioset_free);
1214EXPORT_SYMBOL(bio_alloc_bioset);