]> git.proxmox.com Git - mirror_ubuntu-bionic-kernel.git/blob - fs/bio.c
[PATCH] slab: remove kmem_cache_t
[mirror_ubuntu-bionic-kernel.git] / fs / bio.c
1 /*
2 * Copyright (C) 2001 Jens Axboe <axboe@kernel.dk>
3 *
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>
28 #include <linux/blktrace_api.h>
29 #include <scsi/sg.h> /* for struct sg_iovec */
30
31 #define BIO_POOL_SIZE 256
32
33 static struct kmem_cache *bio_slab __read_mostly;
34
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 */
41 #define BIO_SPLIT_ENTRIES 8
42 mempool_t *bio_split_pool __read_mostly;
43
44 struct biovec_slab {
45 int nr_vecs;
46 char *name;
47 struct kmem_cache *slab;
48 };
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) }
57 static struct biovec_slab bvec_slabs[BIOVEC_NR_POOLS] __read_mostly = {
58 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 */
68 struct 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 */
77 static struct bio_set *fs_bio_set;
78
79 static inline struct bio_vec *bvec_alloc_bs(gfp_t gfp_mask, int nr, unsigned long *idx, struct bio_set *bs)
80 {
81 struct bio_vec *bvl;
82
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
100 bvl = mempool_alloc(bs->bvec_pools[*idx], gfp_mask);
101 if (bvl) {
102 struct biovec_slab *bp = bvec_slabs + *idx;
103
104 memset(bvl, 0, bp->nr_vecs * sizeof(struct bio_vec));
105 }
106
107 return bvl;
108 }
109
110 void bio_free(struct bio *bio, struct bio_set *bio_set)
111 {
112 const int pool_idx = BIO_POOL_IDX(bio);
113
114 BIO_BUG_ON(pool_idx >= BIOVEC_NR_POOLS);
115
116 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 */
123 static void bio_fs_destructor(struct bio *bio)
124 {
125 bio_free(bio, fs_bio_set);
126 }
127
128 void bio_init(struct bio *bio)
129 {
130 bio->bi_next = NULL;
131 bio->bi_bdev = NULL;
132 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
151 * @bs: the bio_set to allocate from
152 *
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 **/
161 struct bio *bio_alloc_bioset(gfp_t gfp_mask, int nr_iovecs, struct bio_set *bs)
162 {
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)) {
170 unsigned long idx = 0; /* shut up gcc */
171
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;
182 }
183 out:
184 return bio;
185 }
186
187 struct bio *bio_alloc(gfp_t gfp_mask, int nr_iovecs)
188 {
189 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;
195 }
196
197 void 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 }
210 EXPORT_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 **/
220 void 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
233 inline int bio_phys_segments(request_queue_t *q, struct bio *bio)
234 {
235 if (unlikely(!bio_flagged(bio, BIO_SEG_VALID)))
236 blk_recount_segments(q, bio);
237
238 return bio->bi_phys_segments;
239 }
240
241 inline int bio_hw_segments(request_queue_t *q, struct bio *bio)
242 {
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 */
258 void __bio_clone(struct bio *bio, struct bio *bio_src)
259 {
260 request_queue_t *q = bdev_get_queue(bio_src->bi_bdev);
261
262 memcpy(bio->bi_io_vec, bio_src->bi_io_vec,
263 bio_src->bi_max_vecs * sizeof(struct bio_vec));
264
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;
269 bio->bi_vcnt = bio_src->bi_vcnt;
270 bio->bi_size = bio_src->bi_size;
271 bio->bi_idx = bio_src->bi_idx;
272 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 */
283 struct bio *bio_clone(struct bio *bio, gfp_t gfp_mask)
284 {
285 struct bio *b = bio_alloc_bioset(gfp_mask, bio->bi_max_vecs, fs_bio_set);
286
287 if (b) {
288 b->bi_destructor = bio_fs_destructor;
289 __bio_clone(b, bio);
290 }
291
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 */
304 int bio_get_nr_vecs(struct block_device *bdev)
305 {
306 request_queue_t *q = bdev_get_queue(bdev);
307 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
318 static int __bio_add_page(request_queue_t *q, struct bio *bio, struct page
319 *page, unsigned int len, unsigned int offset,
320 unsigned short max_sectors)
321 {
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
331 if (((bio->bi_size + len) >> 9) > max_sectors)
332 return 0;
333
334 /*
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)
356 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++;
409 done:
410 bio->bi_size += len;
411 return len;
412 }
413
414 /**
415 * bio_add_pc_page - attempt to add page to bio
416 * @q: the target queue
417 * @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 */
428 int bio_add_pc_page(request_queue_t *q, struct bio *bio, struct page *page,
429 unsigned int len, unsigned int offset)
430 {
431 return __bio_add_page(q, bio, page, len, offset, q->max_hw_sectors);
432 }
433
434 /**
435 * 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 */
447 int bio_add_page(struct bio *bio, struct page *page, unsigned int len,
448 unsigned int offset)
449 {
450 struct request_queue *q = bdev_get_queue(bio->bi_bdev);
451 return __bio_add_page(q, bio, page, len, offset, q->max_sectors);
452 }
453
454 struct bio_map_data {
455 struct bio_vec *iovecs;
456 void __user *userptr;
457 };
458
459 static 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
465 static void bio_free_map_data(struct bio_map_data *bmd)
466 {
467 kfree(bmd->iovecs);
468 kfree(bmd);
469 }
470
471 static 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 */
493 int 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 */
526 struct bio *bio_copy_user(request_queue_t *q, unsigned long uaddr,
527 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
563 if (bio_add_pc_page(q, bio, page, bytes, 0) < bytes)
564 break;
565
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;
593 cleanup:
594 bio_for_each_segment(bvec, bio, i)
595 __free_page(bvec->bv_page);
596
597 bio_put(bio);
598 out_bmd:
599 bio_free_map_data(bmd);
600 return ERR_PTR(ret);
601 }
602
603 static struct bio *__bio_map_user_iov(request_queue_t *q,
604 struct block_device *bdev,
605 struct sg_iovec *iov, int iov_count,
606 int write_to_vm)
607 {
608 int i, j;
609 int nr_pages = 0;
610 struct page **pages;
611 struct bio *bio;
612 int cur_page = 0;
613 int ret, offset;
614
615 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 /*
623 * buffer must be aligned to at least hardsector size for now
624 */
625 if (uaddr & queue_dma_alignment(q))
626 return ERR_PTR(-EINVAL);
627 }
628
629 if (!nr_pages)
630 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;
637 pages = kcalloc(nr_pages, sizeof(struct page *), GFP_KERNEL);
638 if (!pages)
639 goto out;
640
641 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);
654
655 if (ret < local_nr_pages) {
656 ret = -EFAULT;
657 goto out_unmap;
658 }
659
660 offset = uaddr & ~PAGE_MASK;
661 for (j = cur_page; j < page_limit; j++) {
662 unsigned int bytes = PAGE_SIZE - offset;
663
664 if (len <= 0)
665 break;
666
667 if (bytes > len)
668 bytes = len;
669
670 /*
671 * sorry...
672 */
673 if (bio_add_pc_page(q, bio, pages[j], bytes, offset) <
674 bytes)
675 break;
676
677 len -= bytes;
678 offset = 0;
679 }
680
681 cur_page = j;
682 /*
683 * release the pages we didn't map into the bio, if any
684 */
685 while (j < page_limit)
686 page_cache_release(pages[j++]);
687 }
688
689 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
697 bio->bi_bdev = bdev;
698 bio->bi_flags |= (1 << BIO_USER_MAPPED);
699 return bio;
700
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:
708 kfree(pages);
709 bio_put(bio);
710 return ERR_PTR(ret);
711 }
712
713 /**
714 * bio_map_user - map user address into bio
715 * @q: the request_queue_t for the bio
716 * @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 */
724 struct bio *bio_map_user(request_queue_t *q, struct block_device *bdev,
725 unsigned long uaddr, unsigned int len, int write_to_vm)
726 {
727 struct sg_iovec iov;
728
729 iov.iov_base = (void __user *)uaddr;
730 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
737 * @q: the request_queue_t for the bio
738 * @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 */
746 struct bio *bio_map_user_iov(request_queue_t *q, struct block_device *bdev,
747 struct sg_iovec *iov, int iov_count,
748 int write_to_vm)
749 {
750 struct bio *bio;
751
752 bio = __bio_map_user_iov(q, bdev, iov, iov_count, write_to_vm);
753
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
765 return bio;
766 }
767
768 static 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 */
795 void bio_unmap_user(struct bio *bio)
796 {
797 __bio_unmap_user(bio);
798 bio_put(bio);
799 }
800
801 static int bio_map_kern_endio(struct bio *bio, unsigned int bytes_done, int err)
802 {
803 if (bio->bi_size)
804 return 1;
805
806 bio_put(bio);
807 return 0;
808 }
809
810
811 static struct bio *__bio_map_kern(request_queue_t *q, void *data,
812 unsigned int len, gfp_t gfp_mask)
813 {
814 unsigned long kaddr = (unsigned long)data;
815 unsigned long end = (kaddr + len + PAGE_SIZE - 1) >> PAGE_SHIFT;
816 unsigned long start = kaddr >> PAGE_SHIFT;
817 const int nr_pages = end - start;
818 int offset, i;
819 struct bio *bio;
820
821 bio = bio_alloc(gfp_mask, nr_pages);
822 if (!bio)
823 return ERR_PTR(-ENOMEM);
824
825 offset = offset_in_page(kaddr);
826 for (i = 0; i < nr_pages; i++) {
827 unsigned int bytes = PAGE_SIZE - offset;
828
829 if (len <= 0)
830 break;
831
832 if (bytes > len)
833 bytes = len;
834
835 if (bio_add_pc_page(q, bio, virt_to_page(data), bytes,
836 offset) < bytes)
837 break;
838
839 data += bytes;
840 len -= bytes;
841 offset = 0;
842 }
843
844 bio->bi_end_io = bio_map_kern_endio;
845 return bio;
846 }
847
848 /**
849 * bio_map_kern - map kernel address into bio
850 * @q: the request_queue_t for the bio
851 * @data: pointer to buffer to map
852 * @len: length in bytes
853 * @gfp_mask: allocation flags for bio allocation
854 *
855 * Map the kernel address into a bio suitable for io to a block
856 * device. Returns an error pointer in case of error.
857 */
858 struct bio *bio_map_kern(request_queue_t *q, void *data, unsigned int len,
859 gfp_t gfp_mask)
860 {
861 struct bio *bio;
862
863 bio = __bio_map_kern(q, data, len, gfp_mask);
864 if (IS_ERR(bio))
865 return bio;
866
867 if (bio->bi_size == len)
868 return bio;
869
870 /*
871 * Don't support partial mappings.
872 */
873 bio_put(bio);
874 return ERR_PTR(-EINVAL);
875 }
876
877 /*
878 * bio_set_pages_dirty() and bio_check_pages_dirty() are support functions
879 * for performing direct-IO in BIOs.
880 *
881 * The problem is that we cannot run set_page_dirty() from interrupt context
882 * because the required locks are not interrupt-safe. So what we can do is to
883 * mark the pages dirty _before_ performing IO. And in interrupt context,
884 * check that the pages are still dirty. If so, fine. If not, redirty them
885 * in process context.
886 *
887 * We special-case compound pages here: normally this means reads into hugetlb
888 * pages. The logic in here doesn't really work right for compound pages
889 * because the VM does not uniformly chase down the head page in all cases.
890 * But dirtiness of compound pages is pretty meaningless anyway: the VM doesn't
891 * handle them at all. So we skip compound pages here at an early stage.
892 *
893 * Note that this code is very hard to test under normal circumstances because
894 * direct-io pins the pages with get_user_pages(). This makes
895 * is_page_cache_freeable return false, and the VM will not clean the pages.
896 * But other code (eg, pdflush) could clean the pages if they are mapped
897 * pagecache.
898 *
899 * Simply disabling the call to bio_set_pages_dirty() is a good way to test the
900 * deferred bio dirtying paths.
901 */
902
903 /*
904 * bio_set_pages_dirty() will mark all the bio's pages as dirty.
905 */
906 void bio_set_pages_dirty(struct bio *bio)
907 {
908 struct bio_vec *bvec = bio->bi_io_vec;
909 int i;
910
911 for (i = 0; i < bio->bi_vcnt; i++) {
912 struct page *page = bvec[i].bv_page;
913
914 if (page && !PageCompound(page))
915 set_page_dirty_lock(page);
916 }
917 }
918
919 static void bio_release_pages(struct bio *bio)
920 {
921 struct bio_vec *bvec = bio->bi_io_vec;
922 int i;
923
924 for (i = 0; i < bio->bi_vcnt; i++) {
925 struct page *page = bvec[i].bv_page;
926
927 if (page)
928 put_page(page);
929 }
930 }
931
932 /*
933 * bio_check_pages_dirty() will check that all the BIO's pages are still dirty.
934 * If they are, then fine. If, however, some pages are clean then they must
935 * have been written out during the direct-IO read. So we take another ref on
936 * the BIO and the offending pages and re-dirty the pages in process context.
937 *
938 * It is expected that bio_check_pages_dirty() will wholly own the BIO from
939 * here on. It will run one page_cache_release() against each page and will
940 * run one bio_put() against the BIO.
941 */
942
943 static void bio_dirty_fn(struct work_struct *work);
944
945 static DECLARE_WORK(bio_dirty_work, bio_dirty_fn);
946 static DEFINE_SPINLOCK(bio_dirty_lock);
947 static struct bio *bio_dirty_list;
948
949 /*
950 * This runs in process context
951 */
952 static void bio_dirty_fn(struct work_struct *work)
953 {
954 unsigned long flags;
955 struct bio *bio;
956
957 spin_lock_irqsave(&bio_dirty_lock, flags);
958 bio = bio_dirty_list;
959 bio_dirty_list = NULL;
960 spin_unlock_irqrestore(&bio_dirty_lock, flags);
961
962 while (bio) {
963 struct bio *next = bio->bi_private;
964
965 bio_set_pages_dirty(bio);
966 bio_release_pages(bio);
967 bio_put(bio);
968 bio = next;
969 }
970 }
971
972 void bio_check_pages_dirty(struct bio *bio)
973 {
974 struct bio_vec *bvec = bio->bi_io_vec;
975 int nr_clean_pages = 0;
976 int i;
977
978 for (i = 0; i < bio->bi_vcnt; i++) {
979 struct page *page = bvec[i].bv_page;
980
981 if (PageDirty(page) || PageCompound(page)) {
982 page_cache_release(page);
983 bvec[i].bv_page = NULL;
984 } else {
985 nr_clean_pages++;
986 }
987 }
988
989 if (nr_clean_pages) {
990 unsigned long flags;
991
992 spin_lock_irqsave(&bio_dirty_lock, flags);
993 bio->bi_private = bio_dirty_list;
994 bio_dirty_list = bio;
995 spin_unlock_irqrestore(&bio_dirty_lock, flags);
996 schedule_work(&bio_dirty_work);
997 } else {
998 bio_put(bio);
999 }
1000 }
1001
1002 /**
1003 * bio_endio - end I/O on a bio
1004 * @bio: bio
1005 * @bytes_done: number of bytes completed
1006 * @error: error, if any
1007 *
1008 * Description:
1009 * bio_endio() will end I/O on @bytes_done number of bytes. This may be
1010 * just a partial part of the bio, or it may be the whole bio. bio_endio()
1011 * is the preferred way to end I/O on a bio, it takes care of decrementing
1012 * bi_size and clearing BIO_UPTODATE on error. @error is 0 on success, and
1013 * and one of the established -Exxxx (-EIO, for instance) error values in
1014 * case something went wrong. Noone should call bi_end_io() directly on
1015 * a bio unless they own it and thus know that it has an end_io function.
1016 **/
1017 void bio_endio(struct bio *bio, unsigned int bytes_done, int error)
1018 {
1019 if (error)
1020 clear_bit(BIO_UPTODATE, &bio->bi_flags);
1021
1022 if (unlikely(bytes_done > bio->bi_size)) {
1023 printk("%s: want %u bytes done, only %u left\n", __FUNCTION__,
1024 bytes_done, bio->bi_size);
1025 bytes_done = bio->bi_size;
1026 }
1027
1028 bio->bi_size -= bytes_done;
1029 bio->bi_sector += (bytes_done >> 9);
1030
1031 if (bio->bi_end_io)
1032 bio->bi_end_io(bio, bytes_done, error);
1033 }
1034
1035 void bio_pair_release(struct bio_pair *bp)
1036 {
1037 if (atomic_dec_and_test(&bp->cnt)) {
1038 struct bio *master = bp->bio1.bi_private;
1039
1040 bio_endio(master, master->bi_size, bp->error);
1041 mempool_free(bp, bp->bio2.bi_private);
1042 }
1043 }
1044
1045 static int bio_pair_end_1(struct bio * bi, unsigned int done, int err)
1046 {
1047 struct bio_pair *bp = container_of(bi, struct bio_pair, bio1);
1048
1049 if (err)
1050 bp->error = err;
1051
1052 if (bi->bi_size)
1053 return 1;
1054
1055 bio_pair_release(bp);
1056 return 0;
1057 }
1058
1059 static int bio_pair_end_2(struct bio * bi, unsigned int done, int err)
1060 {
1061 struct bio_pair *bp = container_of(bi, struct bio_pair, bio2);
1062
1063 if (err)
1064 bp->error = err;
1065
1066 if (bi->bi_size)
1067 return 1;
1068
1069 bio_pair_release(bp);
1070 return 0;
1071 }
1072
1073 /*
1074 * split a bio - only worry about a bio with a single page
1075 * in it's iovec
1076 */
1077 struct bio_pair *bio_split(struct bio *bi, mempool_t *pool, int first_sectors)
1078 {
1079 struct bio_pair *bp = mempool_alloc(pool, GFP_NOIO);
1080
1081 if (!bp)
1082 return bp;
1083
1084 blk_add_trace_pdu_int(bdev_get_queue(bi->bi_bdev), BLK_TA_SPLIT, bi,
1085 bi->bi_sector + first_sectors);
1086
1087 BUG_ON(bi->bi_vcnt != 1);
1088 BUG_ON(bi->bi_idx != 0);
1089 atomic_set(&bp->cnt, 3);
1090 bp->error = 0;
1091 bp->bio1 = *bi;
1092 bp->bio2 = *bi;
1093 bp->bio2.bi_sector += first_sectors;
1094 bp->bio2.bi_size -= first_sectors << 9;
1095 bp->bio1.bi_size = first_sectors << 9;
1096
1097 bp->bv1 = bi->bi_io_vec[0];
1098 bp->bv2 = bi->bi_io_vec[0];
1099 bp->bv2.bv_offset += first_sectors << 9;
1100 bp->bv2.bv_len -= first_sectors << 9;
1101 bp->bv1.bv_len = first_sectors << 9;
1102
1103 bp->bio1.bi_io_vec = &bp->bv1;
1104 bp->bio2.bi_io_vec = &bp->bv2;
1105
1106 bp->bio1.bi_max_vecs = 1;
1107 bp->bio2.bi_max_vecs = 1;
1108
1109 bp->bio1.bi_end_io = bio_pair_end_1;
1110 bp->bio2.bi_end_io = bio_pair_end_2;
1111
1112 bp->bio1.bi_private = bi;
1113 bp->bio2.bi_private = pool;
1114
1115 return bp;
1116 }
1117
1118
1119 /*
1120 * create memory pools for biovec's in a bio_set.
1121 * use the global biovec slabs created for general use.
1122 */
1123 static int biovec_create_pools(struct bio_set *bs, int pool_entries, int scale)
1124 {
1125 int i;
1126
1127 for (i = 0; i < BIOVEC_NR_POOLS; i++) {
1128 struct biovec_slab *bp = bvec_slabs + i;
1129 mempool_t **bvp = bs->bvec_pools + i;
1130
1131 if (pool_entries > 1 && i >= scale)
1132 pool_entries >>= 1;
1133
1134 *bvp = mempool_create_slab_pool(pool_entries, bp->slab);
1135 if (!*bvp)
1136 return -ENOMEM;
1137 }
1138 return 0;
1139 }
1140
1141 static void biovec_free_pools(struct bio_set *bs)
1142 {
1143 int i;
1144
1145 for (i = 0; i < BIOVEC_NR_POOLS; i++) {
1146 mempool_t *bvp = bs->bvec_pools[i];
1147
1148 if (bvp)
1149 mempool_destroy(bvp);
1150 }
1151
1152 }
1153
1154 void bioset_free(struct bio_set *bs)
1155 {
1156 if (bs->bio_pool)
1157 mempool_destroy(bs->bio_pool);
1158
1159 biovec_free_pools(bs);
1160
1161 kfree(bs);
1162 }
1163
1164 struct bio_set *bioset_create(int bio_pool_size, int bvec_pool_size, int scale)
1165 {
1166 struct bio_set *bs = kzalloc(sizeof(*bs), GFP_KERNEL);
1167
1168 if (!bs)
1169 return NULL;
1170
1171 bs->bio_pool = mempool_create_slab_pool(bio_pool_size, bio_slab);
1172 if (!bs->bio_pool)
1173 goto bad;
1174
1175 if (!biovec_create_pools(bs, bvec_pool_size, scale))
1176 return bs;
1177
1178 bad:
1179 bioset_free(bs);
1180 return NULL;
1181 }
1182
1183 static void __init biovec_init_slabs(void)
1184 {
1185 int i;
1186
1187 for (i = 0; i < BIOVEC_NR_POOLS; i++) {
1188 int size;
1189 struct biovec_slab *bvs = bvec_slabs + i;
1190
1191 size = bvs->nr_vecs * sizeof(struct bio_vec);
1192 bvs->slab = kmem_cache_create(bvs->name, size, 0,
1193 SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL, NULL);
1194 }
1195 }
1196
1197 static int __init init_bio(void)
1198 {
1199 int megabytes, bvec_pool_entries;
1200 int scale = BIOVEC_NR_POOLS;
1201
1202 bio_slab = kmem_cache_create("bio", sizeof(struct bio), 0,
1203 SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL, NULL);
1204
1205 biovec_init_slabs();
1206
1207 megabytes = nr_free_pages() >> (20 - PAGE_SHIFT);
1208
1209 /*
1210 * find out where to start scaling
1211 */
1212 if (megabytes <= 16)
1213 scale = 0;
1214 else if (megabytes <= 32)
1215 scale = 1;
1216 else if (megabytes <= 64)
1217 scale = 2;
1218 else if (megabytes <= 96)
1219 scale = 3;
1220 else if (megabytes <= 128)
1221 scale = 4;
1222
1223 /*
1224 * Limit number of entries reserved -- mempools are only used when
1225 * the system is completely unable to allocate memory, so we only
1226 * need enough to make progress.
1227 */
1228 bvec_pool_entries = 1 + scale;
1229
1230 fs_bio_set = bioset_create(BIO_POOL_SIZE, bvec_pool_entries, scale);
1231 if (!fs_bio_set)
1232 panic("bio: can't allocate bios\n");
1233
1234 bio_split_pool = mempool_create_kmalloc_pool(BIO_SPLIT_ENTRIES,
1235 sizeof(struct bio_pair));
1236 if (!bio_split_pool)
1237 panic("bio: can't create split pool\n");
1238
1239 return 0;
1240 }
1241
1242 subsys_initcall(init_bio);
1243
1244 EXPORT_SYMBOL(bio_alloc);
1245 EXPORT_SYMBOL(bio_put);
1246 EXPORT_SYMBOL(bio_free);
1247 EXPORT_SYMBOL(bio_endio);
1248 EXPORT_SYMBOL(bio_init);
1249 EXPORT_SYMBOL(__bio_clone);
1250 EXPORT_SYMBOL(bio_clone);
1251 EXPORT_SYMBOL(bio_phys_segments);
1252 EXPORT_SYMBOL(bio_hw_segments);
1253 EXPORT_SYMBOL(bio_add_page);
1254 EXPORT_SYMBOL(bio_add_pc_page);
1255 EXPORT_SYMBOL(bio_get_nr_vecs);
1256 EXPORT_SYMBOL(bio_map_user);
1257 EXPORT_SYMBOL(bio_unmap_user);
1258 EXPORT_SYMBOL(bio_map_kern);
1259 EXPORT_SYMBOL(bio_pair_release);
1260 EXPORT_SYMBOL(bio_split);
1261 EXPORT_SYMBOL(bio_split_pool);
1262 EXPORT_SYMBOL(bio_copy_user);
1263 EXPORT_SYMBOL(bio_uncopy_user);
1264 EXPORT_SYMBOL(bioset_create);
1265 EXPORT_SYMBOL(bioset_free);
1266 EXPORT_SYMBOL(bio_alloc_bioset);