bridge: netlink: make setlink/dellink notifications more accurate
[linux-2.6/btrfs-unstable.git] / block / bio-integrity.c
blob5df32907ff3bf8f7219f7518edbfe05850f46ea1
1 /*
2 * bio-integrity.c - bio data integrity extensions
4 * Copyright (C) 2007, 2008, 2009 Oracle Corporation
5 * Written by: Martin K. Petersen <martin.petersen@oracle.com>
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.
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.
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.
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"
31 #define BIP_INLINE_VECS 4
33 static struct kmem_cache *bip_slab;
34 static struct workqueue_struct *kintegrityd_wq;
36 void blk_flush_integrity(void)
38 flush_workqueue(kintegrityd_wq);
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
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.
51 struct bio_integrity_payload *bio_integrity_alloc(struct bio *bio,
52 gfp_t gfp_mask,
53 unsigned int nr_vecs)
55 struct bio_integrity_payload *bip;
56 struct bio_set *bs = bio->bi_pool;
57 unsigned inline_vecs;
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;
68 if (unlikely(!bip))
69 return ERR_PTR(-ENOMEM);
71 memset(bip, 0, sizeof(*bip));
73 if (nr_vecs > inline_vecs) {
74 unsigned long idx = 0;
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;
87 bip->bip_bio = bio;
88 bio->bi_integrity = bip;
89 bio->bi_opf |= REQ_INTEGRITY;
91 return bip;
92 err:
93 mempool_free(bip, bs->bio_integrity_pool);
94 return ERR_PTR(-ENOMEM);
96 EXPORT_SYMBOL(bio_integrity_alloc);
98 /**
99 * bio_integrity_free - Free bio integrity payload
100 * @bio: bio containing bip to be freed
102 * Description: Used to free the integrity portion of a bio. Usually
103 * called from bio_free().
105 static void bio_integrity_free(struct bio *bio)
107 struct bio_integrity_payload *bip = bio_integrity(bio);
108 struct bio_set *bs = bio->bi_pool;
110 if (bip->bip_flags & BIP_BLOCK_INTEGRITY)
111 kfree(page_address(bip->bip_vec->bv_page) +
112 bip->bip_vec->bv_offset);
114 if (bs && bs->bio_integrity_pool) {
115 bvec_free(bs->bvec_integrity_pool, bip->bip_vec, bip->bip_slab);
117 mempool_free(bip, bs->bio_integrity_pool);
118 } else {
119 kfree(bip);
122 bio->bi_integrity = NULL;
123 bio->bi_opf &= ~REQ_INTEGRITY;
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
133 * Description: Attach a page containing integrity metadata to bio.
135 int bio_integrity_add_page(struct bio *bio, struct page *page,
136 unsigned int len, unsigned int offset)
138 struct bio_integrity_payload *bip = bio_integrity(bio);
139 struct bio_vec *iv;
141 if (bip->bip_vcnt >= bip->bip_max_vcnt) {
142 printk(KERN_ERR "%s: bip_vec full\n", __func__);
143 return 0;
146 iv = bip->bip_vec + bip->bip_vcnt;
148 if (bip->bip_vcnt &&
149 bvec_gap_to_prev(bio->bi_disk->queue,
150 &bip->bip_vec[bip->bip_vcnt - 1], offset))
151 return 0;
153 iv->bv_page = page;
154 iv->bv_len = len;
155 iv->bv_offset = offset;
156 bip->bip_vcnt++;
158 return len;
160 EXPORT_SYMBOL(bio_integrity_add_page);
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
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.
172 static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi,
173 unsigned int sectors)
175 return sectors >> (bi->interval_exp - 9);
178 static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
179 unsigned int sectors)
181 return bio_integrity_intervals(bi, sectors) * bi->tuple_size;
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
190 static blk_status_t bio_integrity_process(struct bio *bio,
191 struct bvec_iter *proc_iter, integrity_processing_fn *proc_fn)
193 struct blk_integrity *bi = blk_get_integrity(bio->bi_disk);
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;
202 iter.disk_name = bio->bi_disk->disk_name;
203 iter.interval = 1 << bi->interval_exp;
204 iter.seed = proc_iter->bi_sector;
205 iter.prot_buf = prot_buf;
207 __bio_for_each_segment(bv, bio, bviter, *proc_iter) {
208 void *kaddr = kmap_atomic(bv.bv_page);
210 iter.data_buf = kaddr + bv.bv_offset;
211 iter.data_size = bv.bv_len;
213 ret = proc_fn(&iter);
214 if (ret) {
215 kunmap_atomic(kaddr);
216 return ret;
219 kunmap_atomic(kaddr);
221 return ret;
225 * bio_integrity_prep - Prepare bio for integrity I/O
226 * @bio: bio to prepare
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.
236 bool bio_integrity_prep(struct bio *bio)
238 struct bio_integrity_payload *bip;
239 struct blk_integrity *bi = blk_get_integrity(bio->bi_disk);
240 struct request_queue *q = bio->bi_disk->queue;
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;
248 if (!bi)
249 return true;
251 if (bio_op(bio) != REQ_OP_READ && bio_op(bio) != REQ_OP_WRITE)
252 return true;
254 if (!bio_sectors(bio))
255 return true;
257 /* Already protected? */
258 if (bio_integrity(bio))
259 return true;
261 if (bio_data_dir(bio) == READ) {
262 if (!bi->profile->verify_fn ||
263 !(bi->flags & BLK_INTEGRITY_VERIFY))
264 return true;
265 } else {
266 if (!bi->profile->generate_fn ||
267 !(bi->flags & BLK_INTEGRITY_GENERATE))
268 return true;
270 intervals = bio_integrity_intervals(bi, bio_sectors(bio));
272 /* Allocate kernel buffer for protection data */
273 len = intervals * bi->tuple_size;
274 buf = kmalloc(len, GFP_NOIO | q->bounce_gfp);
275 status = BLK_STS_RESOURCE;
276 if (unlikely(buf == NULL)) {
277 printk(KERN_ERR "could not allocate integrity buffer\n");
278 goto err_end_io;
281 end = (((unsigned long) buf) + len + PAGE_SIZE - 1) >> PAGE_SHIFT;
282 start = ((unsigned long) buf) >> PAGE_SHIFT;
283 nr_pages = end - start;
285 /* Allocate bio integrity payload and integrity vectors */
286 bip = bio_integrity_alloc(bio, GFP_NOIO, nr_pages);
287 if (IS_ERR(bip)) {
288 printk(KERN_ERR "could not allocate data integrity bioset\n");
289 kfree(buf);
290 status = BLK_STS_RESOURCE;
291 goto err_end_io;
294 bip->bip_flags |= BIP_BLOCK_INTEGRITY;
295 bip->bip_iter.bi_size = len;
296 bip_set_seed(bip, bio->bi_iter.bi_sector);
298 if (bi->flags & BLK_INTEGRITY_IP_CHECKSUM)
299 bip->bip_flags |= BIP_IP_CHECKSUM;
301 /* Map it */
302 offset = offset_in_page(buf);
303 for (i = 0 ; i < nr_pages ; i++) {
304 int ret;
305 bytes = PAGE_SIZE - offset;
307 if (len <= 0)
308 break;
310 if (bytes > len)
311 bytes = len;
313 ret = bio_integrity_add_page(bio, virt_to_page(buf),
314 bytes, offset);
316 if (ret == 0)
317 return false;
319 if (ret < bytes)
320 break;
322 buf += bytes;
323 len -= bytes;
324 offset = 0;
327 /* Auto-generate integrity metadata if this is a write */
328 if (bio_data_dir(bio) == WRITE) {
329 bio_integrity_process(bio, &bio->bi_iter,
330 bi->profile->generate_fn);
332 return true;
334 err_end_io:
335 bio->bi_status = status;
336 bio_endio(bio);
337 return false;
340 EXPORT_SYMBOL(bio_integrity_prep);
343 * bio_integrity_verify_fn - Integrity I/O completion worker
344 * @work: Work struct stored in bio to be verified
346 * Description: This workqueue function is called to complete a READ
347 * request. The function verifies the transferred integrity metadata
348 * and then calls the original bio end_io function.
350 static void bio_integrity_verify_fn(struct work_struct *work)
352 struct bio_integrity_payload *bip =
353 container_of(work, struct bio_integrity_payload, bip_work);
354 struct bio *bio = bip->bip_bio;
355 struct blk_integrity *bi = blk_get_integrity(bio->bi_disk);
356 struct bvec_iter iter = bio->bi_iter;
359 * At the moment verify is called bio's iterator was advanced
360 * during split and completion, we need to rewind iterator to
361 * it's original position.
363 if (bio_rewind_iter(bio, &iter, iter.bi_done)) {
364 bio->bi_status = bio_integrity_process(bio, &iter,
365 bi->profile->verify_fn);
366 } else {
367 bio->bi_status = BLK_STS_IOERR;
370 bio_integrity_free(bio);
371 bio_endio(bio);
375 * __bio_integrity_endio - Integrity I/O completion function
376 * @bio: Protected bio
377 * @error: Pointer to errno
379 * Description: Completion for integrity I/O
381 * Normally I/O completion is done in interrupt context. However,
382 * verifying I/O integrity is a time-consuming task which must be run
383 * in process context. This function postpones completion
384 * accordingly.
386 bool __bio_integrity_endio(struct bio *bio)
388 struct blk_integrity *bi = blk_get_integrity(bio->bi_disk);
389 struct bio_integrity_payload *bip = bio_integrity(bio);
391 if (bio_op(bio) == REQ_OP_READ && !bio->bi_status &&
392 (bip->bip_flags & BIP_BLOCK_INTEGRITY) && bi->profile->verify_fn) {
393 INIT_WORK(&bip->bip_work, bio_integrity_verify_fn);
394 queue_work(kintegrityd_wq, &bip->bip_work);
395 return false;
398 bio_integrity_free(bio);
399 return true;
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
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.
411 void bio_integrity_advance(struct bio *bio, unsigned int bytes_done)
413 struct bio_integrity_payload *bip = bio_integrity(bio);
414 struct blk_integrity *bi = blk_get_integrity(bio->bi_disk);
415 unsigned bytes = bio_integrity_bytes(bi, bytes_done >> 9);
417 bip->bip_iter.bi_sector += bytes_done >> 9;
418 bvec_iter_advance(bip->bip_vec, &bip->bip_iter, bytes);
420 EXPORT_SYMBOL(bio_integrity_advance);
423 * bio_integrity_trim - Trim integrity vector
424 * @bio: bio whose integrity vector to update
426 * Description: Used to trim the integrity vector in a cloned bio.
428 void bio_integrity_trim(struct bio *bio)
430 struct bio_integrity_payload *bip = bio_integrity(bio);
431 struct blk_integrity *bi = blk_get_integrity(bio->bi_disk);
433 bip->bip_iter.bi_size = bio_integrity_bytes(bi, bio_sectors(bio));
435 EXPORT_SYMBOL(bio_integrity_trim);
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
443 * Description: Called to allocate a bip when cloning a bio
445 int bio_integrity_clone(struct bio *bio, struct bio *bio_src,
446 gfp_t gfp_mask)
448 struct bio_integrity_payload *bip_src = bio_integrity(bio_src);
449 struct bio_integrity_payload *bip;
451 BUG_ON(bip_src == NULL);
453 bip = bio_integrity_alloc(bio, gfp_mask, bip_src->bip_vcnt);
454 if (IS_ERR(bip))
455 return PTR_ERR(bip);
457 memcpy(bip->bip_vec, bip_src->bip_vec,
458 bip_src->bip_vcnt * sizeof(struct bio_vec));
460 bip->bip_vcnt = bip_src->bip_vcnt;
461 bip->bip_iter = bip_src->bip_iter;
463 return 0;
465 EXPORT_SYMBOL(bio_integrity_clone);
467 int bioset_integrity_create(struct bio_set *bs, int pool_size)
469 if (bs->bio_integrity_pool)
470 return 0;
472 bs->bio_integrity_pool = mempool_create_slab_pool(pool_size, bip_slab);
473 if (!bs->bio_integrity_pool)
474 return -1;
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;
482 return 0;
484 EXPORT_SYMBOL(bioset_integrity_create);
486 void bioset_integrity_free(struct bio_set *bs)
488 if (bs->bio_integrity_pool)
489 mempool_destroy(bs->bio_integrity_pool);
491 if (bs->bvec_integrity_pool)
492 mempool_destroy(bs->bvec_integrity_pool);
494 EXPORT_SYMBOL(bioset_integrity_free);
496 void __init bio_integrity_init(void)
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.
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");
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);