]>
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 | #include "blk.h" | |
30 | ||
31 | #define BIP_INLINE_VECS 4 | |
32 | ||
33 | static struct kmem_cache *bip_slab; | |
34 | static struct workqueue_struct *kintegrityd_wq; | |
35 | ||
36 | void blk_flush_integrity(void) | |
37 | { | |
38 | flush_workqueue(kintegrityd_wq); | |
39 | } | |
40 | ||
41 | /** | |
42 | * bio_integrity_alloc - Allocate integrity payload and attach it to bio | |
43 | * @bio: bio to attach integrity metadata to | |
44 | * @gfp_mask: Memory allocation mask | |
45 | * @nr_vecs: Number of integrity metadata scatter-gather elements | |
46 | * | |
47 | * Description: This function prepares a bio for attaching integrity | |
48 | * metadata. nr_vecs specifies the maximum number of pages containing | |
49 | * integrity metadata that can be attached. | |
50 | */ | |
51 | struct bio_integrity_payload *bio_integrity_alloc(struct bio *bio, | |
52 | gfp_t gfp_mask, | |
53 | unsigned int nr_vecs) | |
54 | { | |
55 | struct bio_integrity_payload *bip; | |
56 | struct bio_set *bs = bio->bi_pool; | |
57 | unsigned inline_vecs; | |
58 | ||
59 | if (!bs || !bs->bio_integrity_pool) { | |
60 | bip = kmalloc(sizeof(struct bio_integrity_payload) + | |
61 | sizeof(struct bio_vec) * nr_vecs, gfp_mask); | |
62 | inline_vecs = nr_vecs; | |
63 | } else { | |
64 | bip = mempool_alloc(bs->bio_integrity_pool, gfp_mask); | |
65 | inline_vecs = BIP_INLINE_VECS; | |
66 | } | |
67 | ||
68 | if (unlikely(!bip)) | |
69 | return ERR_PTR(-ENOMEM); | |
70 | ||
71 | memset(bip, 0, sizeof(*bip)); | |
72 | ||
73 | if (nr_vecs > inline_vecs) { | |
74 | unsigned long idx = 0; | |
75 | ||
76 | bip->bip_vec = bvec_alloc(gfp_mask, nr_vecs, &idx, | |
77 | bs->bvec_integrity_pool); | |
78 | if (!bip->bip_vec) | |
79 | goto err; | |
80 | bip->bip_max_vcnt = bvec_nr_vecs(idx); | |
81 | bip->bip_slab = idx; | |
82 | } else { | |
83 | bip->bip_vec = bip->bip_inline_vecs; | |
84 | bip->bip_max_vcnt = inline_vecs; | |
85 | } | |
86 | ||
87 | bip->bip_bio = bio; | |
88 | bio->bi_integrity = bip; | |
89 | bio->bi_opf |= REQ_INTEGRITY; | |
90 | ||
91 | return bip; | |
92 | err: | |
93 | mempool_free(bip, bs->bio_integrity_pool); | |
94 | return ERR_PTR(-ENOMEM); | |
95 | } | |
96 | EXPORT_SYMBOL(bio_integrity_alloc); | |
97 | ||
98 | /** | |
99 | * bio_integrity_free - Free bio integrity payload | |
100 | * @bio: bio containing bip to be freed | |
101 | * | |
102 | * Description: Used to free the integrity portion of a bio. Usually | |
103 | * called from bio_free(). | |
104 | */ | |
105 | static void bio_integrity_free(struct bio *bio) | |
106 | { | |
107 | struct bio_integrity_payload *bip = bio_integrity(bio); | |
108 | struct bio_set *bs = bio->bi_pool; | |
109 | ||
110 | if (bip->bip_flags & BIP_BLOCK_INTEGRITY) | |
111 | kfree(page_address(bip->bip_vec->bv_page) + | |
112 | bip->bip_vec->bv_offset); | |
113 | ||
114 | if (bs && bs->bio_integrity_pool) { | |
115 | bvec_free(bs->bvec_integrity_pool, bip->bip_vec, bip->bip_slab); | |
116 | ||
117 | mempool_free(bip, bs->bio_integrity_pool); | |
118 | } else { | |
119 | kfree(bip); | |
120 | } | |
121 | ||
122 | bio->bi_integrity = NULL; | |
123 | bio->bi_opf &= ~REQ_INTEGRITY; | |
124 | } | |
125 | ||
126 | /** | |
127 | * bio_integrity_add_page - Attach integrity metadata | |
128 | * @bio: bio to update | |
129 | * @page: page containing integrity metadata | |
130 | * @len: number of bytes of integrity metadata in page | |
131 | * @offset: start offset within page | |
132 | * | |
133 | * Description: Attach a page containing integrity metadata to bio. | |
134 | */ | |
135 | int bio_integrity_add_page(struct bio *bio, struct page *page, | |
136 | unsigned int len, unsigned int offset) | |
137 | { | |
138 | struct bio_integrity_payload *bip = bio_integrity(bio); | |
139 | struct bio_vec *iv; | |
140 | ||
141 | if (bip->bip_vcnt >= bip->bip_max_vcnt) { | |
142 | printk(KERN_ERR "%s: bip_vec full\n", __func__); | |
143 | return 0; | |
144 | } | |
145 | ||
146 | iv = bip->bip_vec + bip->bip_vcnt; | |
147 | ||
148 | if (bip->bip_vcnt && | |
149 | bvec_gap_to_prev(bdev_get_queue(bio->bi_bdev), | |
150 | &bip->bip_vec[bip->bip_vcnt - 1], offset)) | |
151 | return 0; | |
152 | ||
153 | iv->bv_page = page; | |
154 | iv->bv_len = len; | |
155 | iv->bv_offset = offset; | |
156 | bip->bip_vcnt++; | |
157 | ||
158 | return len; | |
159 | } | |
160 | EXPORT_SYMBOL(bio_integrity_add_page); | |
161 | ||
162 | /** | |
163 | * bio_integrity_intervals - Return number of integrity intervals for a bio | |
164 | * @bi: blk_integrity profile for device | |
165 | * @sectors: Size of the bio in 512-byte sectors | |
166 | * | |
167 | * Description: The block layer calculates everything in 512 byte | |
168 | * sectors but integrity metadata is done in terms of the data integrity | |
169 | * interval size of the storage device. Convert the block layer sectors | |
170 | * to the appropriate number of integrity intervals. | |
171 | */ | |
172 | static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi, | |
173 | unsigned int sectors) | |
174 | { | |
175 | return sectors >> (bi->interval_exp - 9); | |
176 | } | |
177 | ||
178 | static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi, | |
179 | unsigned int sectors) | |
180 | { | |
181 | return bio_integrity_intervals(bi, sectors) * bi->tuple_size; | |
182 | } | |
183 | ||
184 | /** | |
185 | * bio_integrity_process - Process integrity metadata for a bio | |
186 | * @bio: bio to generate/verify integrity metadata for | |
187 | * @proc_iter: iterator to process | |
188 | * @proc_fn: Pointer to the relevant processing function | |
189 | */ | |
190 | static blk_status_t bio_integrity_process(struct bio *bio, | |
191 | struct bvec_iter *proc_iter, integrity_processing_fn *proc_fn) | |
192 | { | |
193 | struct blk_integrity *bi = bdev_get_integrity(bio->bi_bdev); | |
194 | struct blk_integrity_iter iter; | |
195 | struct bvec_iter bviter; | |
196 | struct bio_vec bv; | |
197 | struct bio_integrity_payload *bip = bio_integrity(bio); | |
198 | blk_status_t ret = BLK_STS_OK; | |
199 | void *prot_buf = page_address(bip->bip_vec->bv_page) + | |
200 | bip->bip_vec->bv_offset; | |
201 | ||
202 | iter.disk_name = bio->bi_bdev->bd_disk->disk_name; | |
203 | iter.interval = 1 << bi->interval_exp; | |
204 | iter.seed = proc_iter->bi_sector; | |
205 | iter.prot_buf = prot_buf; | |
206 | ||
207 | __bio_for_each_segment(bv, bio, bviter, *proc_iter) { | |
208 | void *kaddr = kmap_atomic(bv.bv_page); | |
209 | ||
210 | iter.data_buf = kaddr + bv.bv_offset; | |
211 | iter.data_size = bv.bv_len; | |
212 | ||
213 | ret = proc_fn(&iter); | |
214 | if (ret) { | |
215 | kunmap_atomic(kaddr); | |
216 | return ret; | |
217 | } | |
218 | ||
219 | kunmap_atomic(kaddr); | |
220 | } | |
221 | return ret; | |
222 | } | |
223 | ||
224 | /** | |
225 | * bio_integrity_prep - Prepare bio for integrity I/O | |
226 | * @bio: bio to prepare | |
227 | * | |
228 | * Description: Checks if the bio already has an integrity payload attached. | |
229 | * If it does, the payload has been generated by another kernel subsystem, | |
230 | * and we just pass it through. Otherwise allocates integrity payload. | |
231 | * The bio must have data direction, target device and start sector set priot | |
232 | * to calling. In the WRITE case, integrity metadata will be generated using | |
233 | * the block device's integrity function. In the READ case, the buffer | |
234 | * will be prepared for DMA and a suitable end_io handler set up. | |
235 | */ | |
236 | bool bio_integrity_prep(struct bio *bio) | |
237 | { | |
238 | struct bio_integrity_payload *bip; | |
239 | struct blk_integrity *bi; | |
240 | struct request_queue *q; | |
241 | void *buf; | |
242 | unsigned long start, end; | |
243 | unsigned int len, nr_pages; | |
244 | unsigned int bytes, offset, i; | |
245 | unsigned int intervals; | |
246 | blk_status_t status; | |
247 | ||
248 | bi = bdev_get_integrity(bio->bi_bdev); | |
249 | q = bdev_get_queue(bio->bi_bdev); | |
250 | if (bio_op(bio) != REQ_OP_READ && bio_op(bio) != REQ_OP_WRITE) | |
251 | return true; | |
252 | ||
253 | if (!bio_sectors(bio)) | |
254 | return true; | |
255 | ||
256 | /* Already protected? */ | |
257 | if (bio_integrity(bio)) | |
258 | return true; | |
259 | ||
260 | if (bi == NULL) | |
261 | return true; | |
262 | ||
263 | if (bio_data_dir(bio) == READ) { | |
264 | if (!bi->profile->verify_fn || | |
265 | !(bi->flags & BLK_INTEGRITY_VERIFY)) | |
266 | return true; | |
267 | } else { | |
268 | if (!bi->profile->generate_fn || | |
269 | !(bi->flags & BLK_INTEGRITY_GENERATE)) | |
270 | return true; | |
271 | } | |
272 | intervals = bio_integrity_intervals(bi, bio_sectors(bio)); | |
273 | ||
274 | /* Allocate kernel buffer for protection data */ | |
275 | len = intervals * bi->tuple_size; | |
276 | buf = kmalloc(len, GFP_NOIO | q->bounce_gfp); | |
277 | status = BLK_STS_RESOURCE; | |
278 | if (unlikely(buf == NULL)) { | |
279 | printk(KERN_ERR "could not allocate integrity buffer\n"); | |
280 | goto err_end_io; | |
281 | } | |
282 | ||
283 | end = (((unsigned long) buf) + len + PAGE_SIZE - 1) >> PAGE_SHIFT; | |
284 | start = ((unsigned long) buf) >> PAGE_SHIFT; | |
285 | nr_pages = end - start; | |
286 | ||
287 | /* Allocate bio integrity payload and integrity vectors */ | |
288 | bip = bio_integrity_alloc(bio, GFP_NOIO, nr_pages); | |
289 | if (IS_ERR(bip)) { | |
290 | printk(KERN_ERR "could not allocate data integrity bioset\n"); | |
291 | kfree(buf); | |
292 | status = BLK_STS_RESOURCE; | |
293 | goto err_end_io; | |
294 | } | |
295 | ||
296 | bip->bip_flags |= BIP_BLOCK_INTEGRITY; | |
297 | bip->bip_iter.bi_size = len; | |
298 | bip_set_seed(bip, bio->bi_iter.bi_sector); | |
299 | ||
300 | if (bi->flags & BLK_INTEGRITY_IP_CHECKSUM) | |
301 | bip->bip_flags |= BIP_IP_CHECKSUM; | |
302 | ||
303 | /* Map it */ | |
304 | offset = offset_in_page(buf); | |
305 | for (i = 0 ; i < nr_pages ; i++) { | |
306 | int ret; | |
307 | bytes = PAGE_SIZE - offset; | |
308 | ||
309 | if (len <= 0) | |
310 | break; | |
311 | ||
312 | if (bytes > len) | |
313 | bytes = len; | |
314 | ||
315 | ret = bio_integrity_add_page(bio, virt_to_page(buf), | |
316 | bytes, offset); | |
317 | ||
318 | if (ret == 0) | |
319 | return false; | |
320 | ||
321 | if (ret < bytes) | |
322 | break; | |
323 | ||
324 | buf += bytes; | |
325 | len -= bytes; | |
326 | offset = 0; | |
327 | } | |
328 | ||
329 | /* Auto-generate integrity metadata if this is a write */ | |
330 | if (bio_data_dir(bio) == WRITE) { | |
331 | bio_integrity_process(bio, &bio->bi_iter, | |
332 | bi->profile->generate_fn); | |
333 | } | |
334 | return true; | |
335 | ||
336 | err_end_io: | |
337 | bio->bi_status = status; | |
338 | bio_endio(bio); | |
339 | return false; | |
340 | ||
341 | } | |
342 | EXPORT_SYMBOL(bio_integrity_prep); | |
343 | ||
344 | /** | |
345 | * bio_integrity_verify_fn - Integrity I/O completion worker | |
346 | * @work: Work struct stored in bio to be verified | |
347 | * | |
348 | * Description: This workqueue function is called to complete a READ | |
349 | * request. The function verifies the transferred integrity metadata | |
350 | * and then calls the original bio end_io function. | |
351 | */ | |
352 | static void bio_integrity_verify_fn(struct work_struct *work) | |
353 | { | |
354 | struct bio_integrity_payload *bip = | |
355 | container_of(work, struct bio_integrity_payload, bip_work); | |
356 | struct bio *bio = bip->bip_bio; | |
357 | struct blk_integrity *bi = bdev_get_integrity(bio->bi_bdev); | |
358 | struct bvec_iter iter = bio->bi_iter; | |
359 | ||
360 | /* | |
361 | * At the moment verify is called bio's iterator was advanced | |
362 | * during split and completion, we need to rewind iterator to | |
363 | * it's original position. | |
364 | */ | |
365 | if (bio_rewind_iter(bio, &iter, iter.bi_done)) { | |
366 | bio->bi_status = bio_integrity_process(bio, &iter, | |
367 | bi->profile->verify_fn); | |
368 | } else { | |
369 | bio->bi_status = BLK_STS_IOERR; | |
370 | } | |
371 | ||
372 | bio_integrity_free(bio); | |
373 | bio_endio(bio); | |
374 | } | |
375 | ||
376 | /** | |
377 | * __bio_integrity_endio - Integrity I/O completion function | |
378 | * @bio: Protected bio | |
379 | * @error: Pointer to errno | |
380 | * | |
381 | * Description: Completion for integrity I/O | |
382 | * | |
383 | * Normally I/O completion is done in interrupt context. However, | |
384 | * verifying I/O integrity is a time-consuming task which must be run | |
385 | * in process context. This function postpones completion | |
386 | * accordingly. | |
387 | */ | |
388 | bool __bio_integrity_endio(struct bio *bio) | |
389 | { | |
390 | if (bio_op(bio) == REQ_OP_READ && !bio->bi_status) { | |
391 | struct bio_integrity_payload *bip = bio_integrity(bio); | |
392 | ||
393 | INIT_WORK(&bip->bip_work, bio_integrity_verify_fn); | |
394 | queue_work(kintegrityd_wq, &bip->bip_work); | |
395 | return false; | |
396 | } | |
397 | ||
398 | bio_integrity_free(bio); | |
399 | return true; | |
400 | } | |
401 | ||
402 | /** | |
403 | * bio_integrity_advance - Advance integrity vector | |
404 | * @bio: bio whose integrity vector to update | |
405 | * @bytes_done: number of data bytes that have been completed | |
406 | * | |
407 | * Description: This function calculates how many integrity bytes the | |
408 | * number of completed data bytes correspond to and advances the | |
409 | * integrity vector accordingly. | |
410 | */ | |
411 | void bio_integrity_advance(struct bio *bio, unsigned int bytes_done) | |
412 | { | |
413 | struct bio_integrity_payload *bip = bio_integrity(bio); | |
414 | struct blk_integrity *bi = bdev_get_integrity(bio->bi_bdev); | |
415 | unsigned bytes = bio_integrity_bytes(bi, bytes_done >> 9); | |
416 | ||
417 | bip->bip_iter.bi_sector += bytes_done >> 9; | |
418 | bvec_iter_advance(bip->bip_vec, &bip->bip_iter, bytes); | |
419 | } | |
420 | EXPORT_SYMBOL(bio_integrity_advance); | |
421 | ||
422 | /** | |
423 | * bio_integrity_trim - Trim integrity vector | |
424 | * @bio: bio whose integrity vector to update | |
425 | * | |
426 | * Description: Used to trim the integrity vector in a cloned bio. | |
427 | */ | |
428 | void bio_integrity_trim(struct bio *bio) | |
429 | { | |
430 | struct bio_integrity_payload *bip = bio_integrity(bio); | |
431 | struct blk_integrity *bi = bdev_get_integrity(bio->bi_bdev); | |
432 | ||
433 | bip->bip_iter.bi_size = bio_integrity_bytes(bi, bio_sectors(bio)); | |
434 | } | |
435 | EXPORT_SYMBOL(bio_integrity_trim); | |
436 | ||
437 | /** | |
438 | * bio_integrity_clone - Callback for cloning bios with integrity metadata | |
439 | * @bio: New bio | |
440 | * @bio_src: Original bio | |
441 | * @gfp_mask: Memory allocation mask | |
442 | * | |
443 | * Description: Called to allocate a bip when cloning a bio | |
444 | */ | |
445 | int bio_integrity_clone(struct bio *bio, struct bio *bio_src, | |
446 | gfp_t gfp_mask) | |
447 | { | |
448 | struct bio_integrity_payload *bip_src = bio_integrity(bio_src); | |
449 | struct bio_integrity_payload *bip; | |
450 | ||
451 | BUG_ON(bip_src == NULL); | |
452 | ||
453 | bip = bio_integrity_alloc(bio, gfp_mask, bip_src->bip_vcnt); | |
454 | if (IS_ERR(bip)) | |
455 | return PTR_ERR(bip); | |
456 | ||
457 | memcpy(bip->bip_vec, bip_src->bip_vec, | |
458 | bip_src->bip_vcnt * sizeof(struct bio_vec)); | |
459 | ||
460 | bip->bip_vcnt = bip_src->bip_vcnt; | |
461 | bip->bip_iter = bip_src->bip_iter; | |
462 | ||
463 | return 0; | |
464 | } | |
465 | EXPORT_SYMBOL(bio_integrity_clone); | |
466 | ||
467 | int bioset_integrity_create(struct bio_set *bs, int pool_size) | |
468 | { | |
469 | if (bs->bio_integrity_pool) | |
470 | return 0; | |
471 | ||
472 | bs->bio_integrity_pool = mempool_create_slab_pool(pool_size, bip_slab); | |
473 | if (!bs->bio_integrity_pool) | |
474 | return -1; | |
475 | ||
476 | bs->bvec_integrity_pool = biovec_create_pool(pool_size); | |
477 | if (!bs->bvec_integrity_pool) { | |
478 | mempool_destroy(bs->bio_integrity_pool); | |
479 | return -1; | |
480 | } | |
481 | ||
482 | return 0; | |
483 | } | |
484 | EXPORT_SYMBOL(bioset_integrity_create); | |
485 | ||
486 | void bioset_integrity_free(struct bio_set *bs) | |
487 | { | |
488 | if (bs->bio_integrity_pool) | |
489 | mempool_destroy(bs->bio_integrity_pool); | |
490 | ||
491 | if (bs->bvec_integrity_pool) | |
492 | mempool_destroy(bs->bvec_integrity_pool); | |
493 | } | |
494 | EXPORT_SYMBOL(bioset_integrity_free); | |
495 | ||
496 | void __init bio_integrity_init(void) | |
497 | { | |
498 | /* | |
499 | * kintegrityd won't block much but may burn a lot of CPU cycles. | |
500 | * Make it highpri CPU intensive wq with max concurrency of 1. | |
501 | */ | |
502 | kintegrityd_wq = alloc_workqueue("kintegrityd", WQ_MEM_RECLAIM | | |
503 | WQ_HIGHPRI | WQ_CPU_INTENSIVE, 1); | |
504 | if (!kintegrityd_wq) | |
505 | panic("Failed to create kintegrityd\n"); | |
506 | ||
507 | bip_slab = kmem_cache_create("bio_integrity_payload", | |
508 | sizeof(struct bio_integrity_payload) + | |
509 | sizeof(struct bio_vec) * BIP_INLINE_VECS, | |
510 | 0, SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL); | |
511 | } |