]>
Commit | Line | Data |
---|---|---|
1 | /* | |
2 | * bio-integrity.c - bio data integrity extensions | |
3 | * | |
4 | * Copyright (C) 2007, 2008, 2009 Oracle Corporation | |
5 | * Written by: Martin K. Petersen <martin.petersen@oracle.com> | |
6 | * | |
7 | * This program is free software; you can redistribute it and/or | |
8 | * modify it under the terms of the GNU General Public License version | |
9 | * 2 as published by the Free Software Foundation. | |
10 | * | |
11 | * This program is distributed in the hope that it will be useful, but | |
12 | * WITHOUT ANY WARRANTY; without even the implied warranty of | |
13 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU | |
14 | * General Public License for more details. | |
15 | * | |
16 | * You should have received a copy of the GNU General Public License | |
17 | * along with this program; see the file COPYING. If not, write to | |
18 | * the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, | |
19 | * USA. | |
20 | * | |
21 | */ | |
22 | ||
23 | #include <linux/blkdev.h> | |
24 | #include <linux/mempool.h> | |
25 | #include <linux/export.h> | |
26 | #include <linux/bio.h> | |
27 | #include <linux/workqueue.h> | |
28 | #include <linux/slab.h> | |
29 | ||
30 | #define BIP_INLINE_VECS 4 | |
31 | ||
32 | static struct kmem_cache *bip_slab; | |
33 | static struct workqueue_struct *kintegrityd_wq; | |
34 | ||
35 | /** | |
36 | * bio_integrity_alloc - Allocate integrity payload and attach it to bio | |
37 | * @bio: bio to attach integrity metadata to | |
38 | * @gfp_mask: Memory allocation mask | |
39 | * @nr_vecs: Number of integrity metadata scatter-gather elements | |
40 | * | |
41 | * Description: This function prepares a bio for attaching integrity | |
42 | * metadata. nr_vecs specifies the maximum number of pages containing | |
43 | * integrity metadata that can be attached. | |
44 | */ | |
45 | struct bio_integrity_payload *bio_integrity_alloc(struct bio *bio, | |
46 | gfp_t gfp_mask, | |
47 | unsigned int nr_vecs) | |
48 | { | |
49 | struct bio_integrity_payload *bip; | |
50 | struct bio_set *bs = bio->bi_pool; | |
51 | unsigned long idx = BIO_POOL_NONE; | |
52 | unsigned inline_vecs; | |
53 | ||
54 | if (!bs) { | |
55 | bip = kmalloc(sizeof(struct bio_integrity_payload) + | |
56 | sizeof(struct bio_vec) * nr_vecs, gfp_mask); | |
57 | inline_vecs = nr_vecs; | |
58 | } else { | |
59 | bip = mempool_alloc(bs->bio_integrity_pool, gfp_mask); | |
60 | inline_vecs = BIP_INLINE_VECS; | |
61 | } | |
62 | ||
63 | if (unlikely(!bip)) | |
64 | return NULL; | |
65 | ||
66 | memset(bip, 0, sizeof(*bip)); | |
67 | ||
68 | if (nr_vecs > inline_vecs) { | |
69 | bip->bip_vec = bvec_alloc(gfp_mask, nr_vecs, &idx, | |
70 | bs->bvec_integrity_pool); | |
71 | if (!bip->bip_vec) | |
72 | goto err; | |
73 | bip->bip_max_vcnt = bvec_nr_vecs(idx); | |
74 | } else { | |
75 | bip->bip_vec = bip->bip_inline_vecs; | |
76 | bip->bip_max_vcnt = inline_vecs; | |
77 | } | |
78 | ||
79 | bip->bip_slab = idx; | |
80 | bip->bip_bio = bio; | |
81 | bio->bi_integrity = bip; | |
82 | bio->bi_rw |= REQ_INTEGRITY; | |
83 | ||
84 | return bip; | |
85 | err: | |
86 | mempool_free(bip, bs->bio_integrity_pool); | |
87 | return NULL; | |
88 | } | |
89 | EXPORT_SYMBOL(bio_integrity_alloc); | |
90 | ||
91 | /** | |
92 | * bio_integrity_free - Free bio integrity payload | |
93 | * @bio: bio containing bip to be freed | |
94 | * | |
95 | * Description: Used to free the integrity portion of a bio. Usually | |
96 | * called from bio_free(). | |
97 | */ | |
98 | void bio_integrity_free(struct bio *bio) | |
99 | { | |
100 | struct bio_integrity_payload *bip = bio_integrity(bio); | |
101 | struct bio_set *bs = bio->bi_pool; | |
102 | ||
103 | if (bip->bip_owns_buf) | |
104 | kfree(page_address(bip->bip_vec->bv_page) + | |
105 | bip->bip_vec->bv_offset); | |
106 | ||
107 | if (bs) { | |
108 | if (bip->bip_slab != BIO_POOL_NONE) | |
109 | bvec_free(bs->bvec_integrity_pool, bip->bip_vec, | |
110 | bip->bip_slab); | |
111 | ||
112 | mempool_free(bip, bs->bio_integrity_pool); | |
113 | } else { | |
114 | kfree(bip); | |
115 | } | |
116 | ||
117 | bio->bi_integrity = NULL; | |
118 | } | |
119 | EXPORT_SYMBOL(bio_integrity_free); | |
120 | ||
121 | /** | |
122 | * bio_integrity_add_page - Attach integrity metadata | |
123 | * @bio: bio to update | |
124 | * @page: page containing integrity metadata | |
125 | * @len: number of bytes of integrity metadata in page | |
126 | * @offset: start offset within page | |
127 | * | |
128 | * Description: Attach a page containing integrity metadata to bio. | |
129 | */ | |
130 | int bio_integrity_add_page(struct bio *bio, struct page *page, | |
131 | unsigned int len, unsigned int offset) | |
132 | { | |
133 | struct bio_integrity_payload *bip = bio_integrity(bio); | |
134 | struct bio_vec *iv; | |
135 | ||
136 | if (bip->bip_vcnt >= bip->bip_max_vcnt) { | |
137 | printk(KERN_ERR "%s: bip_vec full\n", __func__); | |
138 | return 0; | |
139 | } | |
140 | ||
141 | iv = bip->bip_vec + bip->bip_vcnt; | |
142 | ||
143 | iv->bv_page = page; | |
144 | iv->bv_len = len; | |
145 | iv->bv_offset = offset; | |
146 | bip->bip_vcnt++; | |
147 | ||
148 | return len; | |
149 | } | |
150 | EXPORT_SYMBOL(bio_integrity_add_page); | |
151 | ||
152 | /** | |
153 | * bio_integrity_enabled - Check whether integrity can be passed | |
154 | * @bio: bio to check | |
155 | * | |
156 | * Description: Determines whether bio_integrity_prep() can be called | |
157 | * on this bio or not. bio data direction and target device must be | |
158 | * set prior to calling. The functions honors the write_generate and | |
159 | * read_verify flags in sysfs. | |
160 | */ | |
161 | bool bio_integrity_enabled(struct bio *bio) | |
162 | { | |
163 | struct blk_integrity *bi = bdev_get_integrity(bio->bi_bdev); | |
164 | ||
165 | if (!bio_is_rw(bio)) | |
166 | return false; | |
167 | ||
168 | /* Already protected? */ | |
169 | if (bio_integrity(bio)) | |
170 | return false; | |
171 | ||
172 | if (bi == NULL) | |
173 | return false; | |
174 | ||
175 | if (bio_data_dir(bio) == READ && bi->verify_fn != NULL && | |
176 | (bi->flags & INTEGRITY_FLAG_READ)) | |
177 | return true; | |
178 | ||
179 | if (bio_data_dir(bio) == WRITE && bi->generate_fn != NULL && | |
180 | (bi->flags & INTEGRITY_FLAG_WRITE)) | |
181 | return true; | |
182 | ||
183 | return false; | |
184 | } | |
185 | EXPORT_SYMBOL(bio_integrity_enabled); | |
186 | ||
187 | /** | |
188 | * bio_integrity_intervals - Return number of integrity intervals for a bio | |
189 | * @bi: blk_integrity profile for device | |
190 | * @sectors: Size of the bio in 512-byte sectors | |
191 | * | |
192 | * Description: The block layer calculates everything in 512 byte | |
193 | * sectors but integrity metadata is done in terms of the data integrity | |
194 | * interval size of the storage device. Convert the block layer sectors | |
195 | * to the appropriate number of integrity intervals. | |
196 | */ | |
197 | static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi, | |
198 | unsigned int sectors) | |
199 | { | |
200 | return sectors >> (ilog2(bi->interval) - 9); | |
201 | } | |
202 | ||
203 | static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi, | |
204 | unsigned int sectors) | |
205 | { | |
206 | return bio_integrity_intervals(bi, sectors) * bi->tuple_size; | |
207 | } | |
208 | ||
209 | /** | |
210 | * bio_integrity_generate_verify - Generate/verify integrity metadata for a bio | |
211 | * @bio: bio to generate/verify integrity metadata for | |
212 | * @operate: operate number, 1 for generate, 0 for verify | |
213 | */ | |
214 | static int bio_integrity_generate_verify(struct bio *bio, int operate) | |
215 | { | |
216 | struct blk_integrity *bi = bdev_get_integrity(bio->bi_bdev); | |
217 | struct blk_integrity_exchg bix; | |
218 | struct bio_vec *bv; | |
219 | struct bio_integrity_payload *bip = bio_integrity(bio); | |
220 | sector_t seed; | |
221 | unsigned int intervals, ret = 0, i; | |
222 | void *prot_buf = page_address(bip->bip_vec->bv_page) + | |
223 | bip->bip_vec->bv_offset; | |
224 | ||
225 | if (operate) | |
226 | seed = bio->bi_iter.bi_sector; | |
227 | else | |
228 | seed = bip->bip_iter.bi_sector; | |
229 | ||
230 | bix.disk_name = bio->bi_bdev->bd_disk->disk_name; | |
231 | bix.interval = bi->interval; | |
232 | ||
233 | bio_for_each_segment_all(bv, bio, i) { | |
234 | void *kaddr = kmap_atomic(bv->bv_page); | |
235 | bix.data_buf = kaddr + bv->bv_offset; | |
236 | bix.data_size = bv->bv_len; | |
237 | bix.prot_buf = prot_buf; | |
238 | bix.seed = seed; | |
239 | ||
240 | if (operate) | |
241 | bi->generate_fn(&bix); | |
242 | else { | |
243 | ret = bi->verify_fn(&bix); | |
244 | if (ret) { | |
245 | kunmap_atomic(kaddr); | |
246 | return ret; | |
247 | } | |
248 | } | |
249 | ||
250 | intervals = bv->bv_len / bi->interval; | |
251 | seed += intervals; | |
252 | prot_buf += intervals * bi->tuple_size; | |
253 | ||
254 | kunmap_atomic(kaddr); | |
255 | } | |
256 | return ret; | |
257 | } | |
258 | ||
259 | /** | |
260 | * bio_integrity_generate - Generate integrity metadata for a bio | |
261 | * @bio: bio to generate integrity metadata for | |
262 | * | |
263 | * Description: Generates integrity metadata for a bio by calling the | |
264 | * block device's generation callback function. The bio must have a | |
265 | * bip attached with enough room to accommodate the generated | |
266 | * integrity metadata. | |
267 | */ | |
268 | static void bio_integrity_generate(struct bio *bio) | |
269 | { | |
270 | bio_integrity_generate_verify(bio, 1); | |
271 | } | |
272 | ||
273 | /** | |
274 | * bio_integrity_prep - Prepare bio for integrity I/O | |
275 | * @bio: bio to prepare | |
276 | * | |
277 | * Description: Allocates a buffer for integrity metadata, maps the | |
278 | * pages and attaches them to a bio. The bio must have data | |
279 | * direction, target device and start sector set priot to calling. In | |
280 | * the WRITE case, integrity metadata will be generated using the | |
281 | * block device's integrity function. In the READ case, the buffer | |
282 | * will be prepared for DMA and a suitable end_io handler set up. | |
283 | */ | |
284 | int bio_integrity_prep(struct bio *bio) | |
285 | { | |
286 | struct bio_integrity_payload *bip; | |
287 | struct blk_integrity *bi; | |
288 | struct request_queue *q; | |
289 | void *buf; | |
290 | unsigned long start, end; | |
291 | unsigned int len, nr_pages; | |
292 | unsigned int bytes, offset, i; | |
293 | unsigned int intervals; | |
294 | ||
295 | bi = bdev_get_integrity(bio->bi_bdev); | |
296 | q = bdev_get_queue(bio->bi_bdev); | |
297 | BUG_ON(bi == NULL); | |
298 | BUG_ON(bio_integrity(bio)); | |
299 | ||
300 | intervals = bio_integrity_intervals(bi, bio_sectors(bio)); | |
301 | ||
302 | /* Allocate kernel buffer for protection data */ | |
303 | len = intervals * bi->tuple_size; | |
304 | buf = kmalloc(len, GFP_NOIO | q->bounce_gfp); | |
305 | if (unlikely(buf == NULL)) { | |
306 | printk(KERN_ERR "could not allocate integrity buffer\n"); | |
307 | return -ENOMEM; | |
308 | } | |
309 | ||
310 | end = (((unsigned long) buf) + len + PAGE_SIZE - 1) >> PAGE_SHIFT; | |
311 | start = ((unsigned long) buf) >> PAGE_SHIFT; | |
312 | nr_pages = end - start; | |
313 | ||
314 | /* Allocate bio integrity payload and integrity vectors */ | |
315 | bip = bio_integrity_alloc(bio, GFP_NOIO, nr_pages); | |
316 | if (unlikely(bip == NULL)) { | |
317 | printk(KERN_ERR "could not allocate data integrity bioset\n"); | |
318 | kfree(buf); | |
319 | return -EIO; | |
320 | } | |
321 | ||
322 | bip->bip_owns_buf = 1; | |
323 | bip->bip_iter.bi_size = len; | |
324 | bip->bip_iter.bi_sector = bio->bi_iter.bi_sector; | |
325 | ||
326 | /* Map it */ | |
327 | offset = offset_in_page(buf); | |
328 | for (i = 0 ; i < nr_pages ; i++) { | |
329 | int ret; | |
330 | bytes = PAGE_SIZE - offset; | |
331 | ||
332 | if (len <= 0) | |
333 | break; | |
334 | ||
335 | if (bytes > len) | |
336 | bytes = len; | |
337 | ||
338 | ret = bio_integrity_add_page(bio, virt_to_page(buf), | |
339 | bytes, offset); | |
340 | ||
341 | if (ret == 0) | |
342 | return 0; | |
343 | ||
344 | if (ret < bytes) | |
345 | break; | |
346 | ||
347 | buf += bytes; | |
348 | len -= bytes; | |
349 | offset = 0; | |
350 | } | |
351 | ||
352 | /* Install custom I/O completion handler if read verify is enabled */ | |
353 | if (bio_data_dir(bio) == READ) { | |
354 | bip->bip_end_io = bio->bi_end_io; | |
355 | bio->bi_end_io = bio_integrity_endio; | |
356 | } | |
357 | ||
358 | /* Auto-generate integrity metadata if this is a write */ | |
359 | if (bio_data_dir(bio) == WRITE) | |
360 | bio_integrity_generate(bio); | |
361 | ||
362 | return 0; | |
363 | } | |
364 | EXPORT_SYMBOL(bio_integrity_prep); | |
365 | ||
366 | /** | |
367 | * bio_integrity_verify - Verify integrity metadata for a bio | |
368 | * @bio: bio to verify | |
369 | * | |
370 | * Description: This function is called to verify the integrity of a | |
371 | * bio. The data in the bio io_vec is compared to the integrity | |
372 | * metadata returned by the HBA. | |
373 | */ | |
374 | static int bio_integrity_verify(struct bio *bio) | |
375 | { | |
376 | return bio_integrity_generate_verify(bio, 0); | |
377 | } | |
378 | ||
379 | /** | |
380 | * bio_integrity_verify_fn - Integrity I/O completion worker | |
381 | * @work: Work struct stored in bio to be verified | |
382 | * | |
383 | * Description: This workqueue function is called to complete a READ | |
384 | * request. The function verifies the transferred integrity metadata | |
385 | * and then calls the original bio end_io function. | |
386 | */ | |
387 | static void bio_integrity_verify_fn(struct work_struct *work) | |
388 | { | |
389 | struct bio_integrity_payload *bip = | |
390 | container_of(work, struct bio_integrity_payload, bip_work); | |
391 | struct bio *bio = bip->bip_bio; | |
392 | int error; | |
393 | ||
394 | error = bio_integrity_verify(bio); | |
395 | ||
396 | /* Restore original bio completion handler */ | |
397 | bio->bi_end_io = bip->bip_end_io; | |
398 | bio_endio_nodec(bio, error); | |
399 | } | |
400 | ||
401 | /** | |
402 | * bio_integrity_endio - Integrity I/O completion function | |
403 | * @bio: Protected bio | |
404 | * @error: Pointer to errno | |
405 | * | |
406 | * Description: Completion for integrity I/O | |
407 | * | |
408 | * Normally I/O completion is done in interrupt context. However, | |
409 | * verifying I/O integrity is a time-consuming task which must be run | |
410 | * in process context. This function postpones completion | |
411 | * accordingly. | |
412 | */ | |
413 | void bio_integrity_endio(struct bio *bio, int error) | |
414 | { | |
415 | struct bio_integrity_payload *bip = bio_integrity(bio); | |
416 | ||
417 | BUG_ON(bip->bip_bio != bio); | |
418 | ||
419 | /* In case of an I/O error there is no point in verifying the | |
420 | * integrity metadata. Restore original bio end_io handler | |
421 | * and run it. | |
422 | */ | |
423 | if (error) { | |
424 | bio->bi_end_io = bip->bip_end_io; | |
425 | bio_endio_nodec(bio, error); | |
426 | ||
427 | return; | |
428 | } | |
429 | ||
430 | INIT_WORK(&bip->bip_work, bio_integrity_verify_fn); | |
431 | queue_work(kintegrityd_wq, &bip->bip_work); | |
432 | } | |
433 | EXPORT_SYMBOL(bio_integrity_endio); | |
434 | ||
435 | /** | |
436 | * bio_integrity_advance - Advance integrity vector | |
437 | * @bio: bio whose integrity vector to update | |
438 | * @bytes_done: number of data bytes that have been completed | |
439 | * | |
440 | * Description: This function calculates how many integrity bytes the | |
441 | * number of completed data bytes correspond to and advances the | |
442 | * integrity vector accordingly. | |
443 | */ | |
444 | void bio_integrity_advance(struct bio *bio, unsigned int bytes_done) | |
445 | { | |
446 | struct bio_integrity_payload *bip = bio_integrity(bio); | |
447 | struct blk_integrity *bi = bdev_get_integrity(bio->bi_bdev); | |
448 | unsigned bytes = bio_integrity_bytes(bi, bytes_done >> 9); | |
449 | ||
450 | bvec_iter_advance(bip->bip_vec, &bip->bip_iter, bytes); | |
451 | } | |
452 | EXPORT_SYMBOL(bio_integrity_advance); | |
453 | ||
454 | /** | |
455 | * bio_integrity_trim - Trim integrity vector | |
456 | * @bio: bio whose integrity vector to update | |
457 | * @offset: offset to first data sector | |
458 | * @sectors: number of data sectors | |
459 | * | |
460 | * Description: Used to trim the integrity vector in a cloned bio. | |
461 | * The ivec will be advanced corresponding to 'offset' data sectors | |
462 | * and the length will be truncated corresponding to 'len' data | |
463 | * sectors. | |
464 | */ | |
465 | void bio_integrity_trim(struct bio *bio, unsigned int offset, | |
466 | unsigned int sectors) | |
467 | { | |
468 | struct bio_integrity_payload *bip = bio_integrity(bio); | |
469 | struct blk_integrity *bi = bdev_get_integrity(bio->bi_bdev); | |
470 | ||
471 | bio_integrity_advance(bio, offset << 9); | |
472 | bip->bip_iter.bi_size = bio_integrity_bytes(bi, sectors); | |
473 | } | |
474 | EXPORT_SYMBOL(bio_integrity_trim); | |
475 | ||
476 | /** | |
477 | * bio_integrity_clone - Callback for cloning bios with integrity metadata | |
478 | * @bio: New bio | |
479 | * @bio_src: Original bio | |
480 | * @gfp_mask: Memory allocation mask | |
481 | * | |
482 | * Description: Called to allocate a bip when cloning a bio | |
483 | */ | |
484 | int bio_integrity_clone(struct bio *bio, struct bio *bio_src, | |
485 | gfp_t gfp_mask) | |
486 | { | |
487 | struct bio_integrity_payload *bip_src = bio_integrity(bio_src); | |
488 | struct bio_integrity_payload *bip; | |
489 | ||
490 | BUG_ON(bip_src == NULL); | |
491 | ||
492 | bip = bio_integrity_alloc(bio, gfp_mask, bip_src->bip_vcnt); | |
493 | ||
494 | if (bip == NULL) | |
495 | return -EIO; | |
496 | ||
497 | memcpy(bip->bip_vec, bip_src->bip_vec, | |
498 | bip_src->bip_vcnt * sizeof(struct bio_vec)); | |
499 | ||
500 | bip->bip_vcnt = bip_src->bip_vcnt; | |
501 | bip->bip_iter = bip_src->bip_iter; | |
502 | ||
503 | return 0; | |
504 | } | |
505 | EXPORT_SYMBOL(bio_integrity_clone); | |
506 | ||
507 | int bioset_integrity_create(struct bio_set *bs, int pool_size) | |
508 | { | |
509 | if (bs->bio_integrity_pool) | |
510 | return 0; | |
511 | ||
512 | bs->bio_integrity_pool = mempool_create_slab_pool(pool_size, bip_slab); | |
513 | if (!bs->bio_integrity_pool) | |
514 | return -1; | |
515 | ||
516 | bs->bvec_integrity_pool = biovec_create_pool(pool_size); | |
517 | if (!bs->bvec_integrity_pool) { | |
518 | mempool_destroy(bs->bio_integrity_pool); | |
519 | return -1; | |
520 | } | |
521 | ||
522 | return 0; | |
523 | } | |
524 | EXPORT_SYMBOL(bioset_integrity_create); | |
525 | ||
526 | void bioset_integrity_free(struct bio_set *bs) | |
527 | { | |
528 | if (bs->bio_integrity_pool) | |
529 | mempool_destroy(bs->bio_integrity_pool); | |
530 | ||
531 | if (bs->bvec_integrity_pool) | |
532 | mempool_destroy(bs->bvec_integrity_pool); | |
533 | } | |
534 | EXPORT_SYMBOL(bioset_integrity_free); | |
535 | ||
536 | void __init bio_integrity_init(void) | |
537 | { | |
538 | /* | |
539 | * kintegrityd won't block much but may burn a lot of CPU cycles. | |
540 | * Make it highpri CPU intensive wq with max concurrency of 1. | |
541 | */ | |
542 | kintegrityd_wq = alloc_workqueue("kintegrityd", WQ_MEM_RECLAIM | | |
543 | WQ_HIGHPRI | WQ_CPU_INTENSIVE, 1); | |
544 | if (!kintegrityd_wq) | |
545 | panic("Failed to create kintegrityd\n"); | |
546 | ||
547 | bip_slab = kmem_cache_create("bio_integrity_payload", | |
548 | sizeof(struct bio_integrity_payload) + | |
549 | sizeof(struct bio_vec) * BIP_INLINE_VECS, | |
550 | 0, SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL); | |
551 | if (!bip_slab) | |
552 | panic("Failed to create slab\n"); | |
553 | } |