2 * bio-integrity.c - bio data integrity extensions
4 * Copyright (C) 2007, 2008 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,
23 #include <linux/blkdev.h>
24 #include <linux/mempool.h>
25 #include <linux/bio.h>
26 #include <linux/workqueue.h>
28 static struct kmem_cache
*bio_integrity_slab __read_mostly
;
29 static struct workqueue_struct
*kintegrityd_wq
;
32 * bio_integrity_alloc_bioset - Allocate integrity payload and attach it to bio
33 * @bio: bio to attach integrity metadata to
34 * @gfp_mask: Memory allocation mask
35 * @nr_vecs: Number of integrity metadata scatter-gather elements
36 * @bs: bio_set to allocate from
38 * Description: This function prepares a bio for attaching integrity
39 * metadata. nr_vecs specifies the maximum number of pages containing
40 * integrity metadata that can be attached.
42 struct bio_integrity_payload
*bio_integrity_alloc_bioset(struct bio
*bio
, gfp_t gfp_mask
, unsigned int nr_vecs
, struct bio_set
*bs
)
44 struct bio_integrity_payload
*bip
;
50 bip
= mempool_alloc(bs
->bio_integrity_pool
, gfp_mask
);
51 if (unlikely(bip
== NULL
)) {
52 printk(KERN_ERR
"%s: could not alloc bip\n", __func__
);
56 memset(bip
, 0, sizeof(*bip
));
58 iv
= bvec_alloc_bs(gfp_mask
, nr_vecs
, &idx
, bs
);
59 if (unlikely(iv
== NULL
)) {
60 printk(KERN_ERR
"%s: could not alloc bip_vec\n", __func__
);
61 mempool_free(bip
, bs
->bio_integrity_pool
);
68 bio
->bi_integrity
= bip
;
72 EXPORT_SYMBOL(bio_integrity_alloc_bioset
);
75 * bio_integrity_alloc - Allocate integrity payload and attach it to bio
76 * @bio: bio to attach integrity metadata to
77 * @gfp_mask: Memory allocation mask
78 * @nr_vecs: Number of integrity metadata scatter-gather elements
80 * Description: This function prepares a bio for attaching integrity
81 * metadata. nr_vecs specifies the maximum number of pages containing
82 * integrity metadata that can be attached.
84 struct bio_integrity_payload
*bio_integrity_alloc(struct bio
*bio
, gfp_t gfp_mask
, unsigned int nr_vecs
)
86 return bio_integrity_alloc_bioset(bio
, gfp_mask
, nr_vecs
, fs_bio_set
);
88 EXPORT_SYMBOL(bio_integrity_alloc
);
91 * bio_integrity_free - Free bio integrity payload
92 * @bio: bio containing bip to be freed
93 * @bs: bio_set this bio was allocated from
95 * Description: Used to free the integrity portion of a bio. Usually
96 * called from bio_free().
98 void bio_integrity_free(struct bio
*bio
, struct bio_set
*bs
)
100 struct bio_integrity_payload
*bip
= bio
->bi_integrity
;
104 /* A cloned bio doesn't own the integrity metadata */
105 if (!bio_flagged(bio
, BIO_CLONED
) && bip
->bip_buf
!= NULL
)
108 mempool_free(bip
->bip_vec
, bs
->bvec_pools
[bip
->bip_pool
]);
109 mempool_free(bip
, bs
->bio_integrity_pool
);
111 bio
->bi_integrity
= NULL
;
113 EXPORT_SYMBOL(bio_integrity_free
);
116 * bio_integrity_add_page - Attach integrity metadata
117 * @bio: bio to update
118 * @page: page containing integrity metadata
119 * @len: number of bytes of integrity metadata in page
120 * @offset: start offset within page
122 * Description: Attach a page containing integrity metadata to bio.
124 int bio_integrity_add_page(struct bio
*bio
, struct page
*page
,
125 unsigned int len
, unsigned int offset
)
127 struct bio_integrity_payload
*bip
= bio
->bi_integrity
;
130 if (bip
->bip_vcnt
>= bvec_nr_vecs(bip
->bip_pool
)) {
131 printk(KERN_ERR
"%s: bip_vec full\n", __func__
);
135 iv
= bip_vec_idx(bip
, bip
->bip_vcnt
);
137 BUG_ON(iv
->bv_page
!= NULL
);
141 iv
->bv_offset
= offset
;
146 EXPORT_SYMBOL(bio_integrity_add_page
);
149 * bio_integrity_enabled - Check whether integrity can be passed
152 * Description: Determines whether bio_integrity_prep() can be called
153 * on this bio or not. bio data direction and target device must be
154 * set prior to calling. The functions honors the write_generate and
155 * read_verify flags in sysfs.
157 int bio_integrity_enabled(struct bio
*bio
)
159 /* Already protected? */
160 if (bio_integrity(bio
))
163 return bdev_integrity_enabled(bio
->bi_bdev
, bio_data_dir(bio
));
165 EXPORT_SYMBOL(bio_integrity_enabled
);
168 * bio_integrity_hw_sectors - Convert 512b sectors to hardware ditto
169 * @bi: blk_integrity profile for device
170 * @sectors: Number of 512 sectors to convert
172 * Description: The block layer calculates everything in 512 byte
173 * sectors but integrity metadata is done in terms of the hardware
174 * sector size of the storage device. Convert the block layer sectors
175 * to physical sectors.
177 static inline unsigned int bio_integrity_hw_sectors(struct blk_integrity
*bi
, unsigned int sectors
)
179 /* At this point there are only 512b or 4096b DIF/EPP devices */
180 if (bi
->sector_size
== 4096)
181 return sectors
>>= 3;
187 * bio_integrity_tag_size - Retrieve integrity tag space
188 * @bio: bio to inspect
190 * Description: Returns the maximum number of tag bytes that can be
191 * attached to this bio. Filesystems can use this to determine how
192 * much metadata to attach to an I/O.
194 unsigned int bio_integrity_tag_size(struct bio
*bio
)
196 struct blk_integrity
*bi
= bdev_get_integrity(bio
->bi_bdev
);
198 BUG_ON(bio
->bi_size
== 0);
200 return bi
->tag_size
* (bio
->bi_size
/ bi
->sector_size
);
202 EXPORT_SYMBOL(bio_integrity_tag_size
);
204 int bio_integrity_tag(struct bio
*bio
, void *tag_buf
, unsigned int len
, int set
)
206 struct bio_integrity_payload
*bip
= bio
->bi_integrity
;
207 struct blk_integrity
*bi
= bdev_get_integrity(bio
->bi_bdev
);
208 unsigned int nr_sectors
;
210 BUG_ON(bip
->bip_buf
== NULL
);
212 if (bi
->tag_size
== 0)
215 nr_sectors
= bio_integrity_hw_sectors(bi
, DIV_ROUND_UP(len
, bi
->tag_size
));
217 if (nr_sectors
* bi
->tuple_size
> bip
->bip_size
) {
218 printk(KERN_ERR
"%s: tag too big for bio: %u > %u\n",
219 __func__
, nr_sectors
* bi
->tuple_size
, bip
->bip_size
);
224 bi
->set_tag_fn(bip
->bip_buf
, tag_buf
, nr_sectors
);
226 bi
->get_tag_fn(bip
->bip_buf
, tag_buf
, nr_sectors
);
232 * bio_integrity_set_tag - Attach a tag buffer to a bio
233 * @bio: bio to attach buffer to
234 * @tag_buf: Pointer to a buffer containing tag data
235 * @len: Length of the included buffer
237 * Description: Use this function to tag a bio by leveraging the extra
238 * space provided by devices formatted with integrity protection. The
239 * size of the integrity buffer must be <= to the size reported by
240 * bio_integrity_tag_size().
242 int bio_integrity_set_tag(struct bio
*bio
, void *tag_buf
, unsigned int len
)
244 BUG_ON(bio_data_dir(bio
) != WRITE
);
246 return bio_integrity_tag(bio
, tag_buf
, len
, 1);
248 EXPORT_SYMBOL(bio_integrity_set_tag
);
251 * bio_integrity_get_tag - Retrieve a tag buffer from a bio
252 * @bio: bio to retrieve buffer from
253 * @tag_buf: Pointer to a buffer for the tag data
254 * @len: Length of the target buffer
256 * Description: Use this function to retrieve the tag buffer from a
257 * completed I/O. The size of the integrity buffer must be <= to the
258 * size reported by bio_integrity_tag_size().
260 int bio_integrity_get_tag(struct bio
*bio
, void *tag_buf
, unsigned int len
)
262 BUG_ON(bio_data_dir(bio
) != READ
);
264 return bio_integrity_tag(bio
, tag_buf
, len
, 0);
266 EXPORT_SYMBOL(bio_integrity_get_tag
);
269 * bio_integrity_generate - Generate integrity metadata for a bio
270 * @bio: bio to generate integrity metadata for
272 * Description: Generates integrity metadata for a bio by calling the
273 * block device's generation callback function. The bio must have a
274 * bip attached with enough room to accommodate the generated
275 * integrity metadata.
277 static void bio_integrity_generate(struct bio
*bio
)
279 struct blk_integrity
*bi
= bdev_get_integrity(bio
->bi_bdev
);
280 struct blk_integrity_exchg bix
;
282 sector_t sector
= bio
->bi_sector
;
283 unsigned int i
, sectors
, total
;
284 void *prot_buf
= bio
->bi_integrity
->bip_buf
;
287 bix
.disk_name
= bio
->bi_bdev
->bd_disk
->disk_name
;
288 bix
.sector_size
= bi
->sector_size
;
290 bio_for_each_segment(bv
, bio
, i
) {
291 void *kaddr
= kmap_atomic(bv
->bv_page
, KM_USER0
);
292 bix
.data_buf
= kaddr
+ bv
->bv_offset
;
293 bix
.data_size
= bv
->bv_len
;
294 bix
.prot_buf
= prot_buf
;
297 bi
->generate_fn(&bix
);
299 sectors
= bv
->bv_len
/ bi
->sector_size
;
301 prot_buf
+= sectors
* bi
->tuple_size
;
302 total
+= sectors
* bi
->tuple_size
;
303 BUG_ON(total
> bio
->bi_integrity
->bip_size
);
305 kunmap_atomic(kaddr
, KM_USER0
);
310 * bio_integrity_prep - Prepare bio for integrity I/O
311 * @bio: bio to prepare
313 * Description: Allocates a buffer for integrity metadata, maps the
314 * pages and attaches them to a bio. The bio must have data
315 * direction, target device and start sector set priot to calling. In
316 * the WRITE case, integrity metadata will be generated using the
317 * block device's integrity function. In the READ case, the buffer
318 * will be prepared for DMA and a suitable end_io handler set up.
320 int bio_integrity_prep(struct bio
*bio
)
322 struct bio_integrity_payload
*bip
;
323 struct blk_integrity
*bi
;
324 struct request_queue
*q
;
326 unsigned long start
, end
;
327 unsigned int len
, nr_pages
;
328 unsigned int bytes
, offset
, i
;
329 unsigned int sectors
;
331 bi
= bdev_get_integrity(bio
->bi_bdev
);
332 q
= bdev_get_queue(bio
->bi_bdev
);
334 BUG_ON(bio_integrity(bio
));
336 sectors
= bio_integrity_hw_sectors(bi
, bio_sectors(bio
));
338 /* Allocate kernel buffer for protection data */
339 len
= sectors
* blk_integrity_tuple_size(bi
);
340 buf
= kmalloc(len
, GFP_NOIO
| __GFP_NOFAIL
| q
->bounce_gfp
);
341 if (unlikely(buf
== NULL
)) {
342 printk(KERN_ERR
"could not allocate integrity buffer\n");
346 end
= (((unsigned long) buf
) + len
+ PAGE_SIZE
- 1) >> PAGE_SHIFT
;
347 start
= ((unsigned long) buf
) >> PAGE_SHIFT
;
348 nr_pages
= end
- start
;
350 /* Allocate bio integrity payload and integrity vectors */
351 bip
= bio_integrity_alloc(bio
, GFP_NOIO
, nr_pages
);
352 if (unlikely(bip
== NULL
)) {
353 printk(KERN_ERR
"could not allocate data integrity bioset\n");
360 bip
->bip_sector
= bio
->bi_sector
;
363 offset
= offset_in_page(buf
);
364 for (i
= 0 ; i
< nr_pages
; i
++) {
366 bytes
= PAGE_SIZE
- offset
;
374 ret
= bio_integrity_add_page(bio
, virt_to_page(buf
),
388 /* Install custom I/O completion handler if read verify is enabled */
389 if (bio_data_dir(bio
) == READ
) {
390 bip
->bip_end_io
= bio
->bi_end_io
;
391 bio
->bi_end_io
= bio_integrity_endio
;
394 /* Auto-generate integrity metadata if this is a write */
395 if (bio_data_dir(bio
) == WRITE
)
396 bio_integrity_generate(bio
);
400 EXPORT_SYMBOL(bio_integrity_prep
);
403 * bio_integrity_verify - Verify integrity metadata for a bio
404 * @bio: bio to verify
406 * Description: This function is called to verify the integrity of a
407 * bio. The data in the bio io_vec is compared to the integrity
408 * metadata returned by the HBA.
410 static int bio_integrity_verify(struct bio
*bio
)
412 struct blk_integrity
*bi
= bdev_get_integrity(bio
->bi_bdev
);
413 struct blk_integrity_exchg bix
;
415 sector_t sector
= bio
->bi_integrity
->bip_sector
;
416 unsigned int i
, sectors
, total
, ret
;
417 void *prot_buf
= bio
->bi_integrity
->bip_buf
;
420 bix
.disk_name
= bio
->bi_bdev
->bd_disk
->disk_name
;
421 bix
.sector_size
= bi
->sector_size
;
423 bio_for_each_segment(bv
, bio
, i
) {
424 void *kaddr
= kmap_atomic(bv
->bv_page
, KM_USER0
);
425 bix
.data_buf
= kaddr
+ bv
->bv_offset
;
426 bix
.data_size
= bv
->bv_len
;
427 bix
.prot_buf
= prot_buf
;
430 ret
= bi
->verify_fn(&bix
);
433 kunmap_atomic(kaddr
, KM_USER0
);
437 sectors
= bv
->bv_len
/ bi
->sector_size
;
439 prot_buf
+= sectors
* bi
->tuple_size
;
440 total
+= sectors
* bi
->tuple_size
;
441 BUG_ON(total
> bio
->bi_integrity
->bip_size
);
443 kunmap_atomic(kaddr
, KM_USER0
);
450 * bio_integrity_verify_fn - Integrity I/O completion worker
451 * @work: Work struct stored in bio to be verified
453 * Description: This workqueue function is called to complete a READ
454 * request. The function verifies the transferred integrity metadata
455 * and then calls the original bio end_io function.
457 static void bio_integrity_verify_fn(struct work_struct
*work
)
459 struct bio_integrity_payload
*bip
=
460 container_of(work
, struct bio_integrity_payload
, bip_work
);
461 struct bio
*bio
= bip
->bip_bio
;
462 int error
= bip
->bip_error
;
464 if (bio_integrity_verify(bio
)) {
465 clear_bit(BIO_UPTODATE
, &bio
->bi_flags
);
469 /* Restore original bio completion handler */
470 bio
->bi_end_io
= bip
->bip_end_io
;
473 bio
->bi_end_io(bio
, error
);
477 * bio_integrity_endio - Integrity I/O completion function
478 * @bio: Protected bio
479 * @error: Pointer to errno
481 * Description: Completion for integrity I/O
483 * Normally I/O completion is done in interrupt context. However,
484 * verifying I/O integrity is a time-consuming task which must be run
485 * in process context. This function postpones completion
488 void bio_integrity_endio(struct bio
*bio
, int error
)
490 struct bio_integrity_payload
*bip
= bio
->bi_integrity
;
492 BUG_ON(bip
->bip_bio
!= bio
);
494 bip
->bip_error
= error
;
495 INIT_WORK(&bip
->bip_work
, bio_integrity_verify_fn
);
496 queue_work(kintegrityd_wq
, &bip
->bip_work
);
498 EXPORT_SYMBOL(bio_integrity_endio
);
501 * bio_integrity_mark_head - Advance bip_vec skip bytes
502 * @bip: Integrity vector to advance
503 * @skip: Number of bytes to advance it
505 void bio_integrity_mark_head(struct bio_integrity_payload
*bip
, unsigned int skip
)
510 bip_for_each_vec(iv
, bip
, i
) {
514 } else if (skip
>= iv
->bv_len
) {
516 } else { /* skip < iv->bv_len) */
517 iv
->bv_offset
+= skip
;
526 * bio_integrity_mark_tail - Truncate bip_vec to be len bytes long
527 * @bip: Integrity vector to truncate
528 * @len: New length of integrity vector
530 void bio_integrity_mark_tail(struct bio_integrity_payload
*bip
, unsigned int len
)
535 bip_for_each_vec(iv
, bip
, i
) {
539 } else if (len
>= iv
->bv_len
) {
541 } else { /* len < iv->bv_len) */
549 * bio_integrity_advance - Advance integrity vector
550 * @bio: bio whose integrity vector to update
551 * @bytes_done: number of data bytes that have been completed
553 * Description: This function calculates how many integrity bytes the
554 * number of completed data bytes correspond to and advances the
555 * integrity vector accordingly.
557 void bio_integrity_advance(struct bio
*bio
, unsigned int bytes_done
)
559 struct bio_integrity_payload
*bip
= bio
->bi_integrity
;
560 struct blk_integrity
*bi
= bdev_get_integrity(bio
->bi_bdev
);
561 unsigned int nr_sectors
;
566 nr_sectors
= bio_integrity_hw_sectors(bi
, bytes_done
>> 9);
567 bio_integrity_mark_head(bip
, nr_sectors
* bi
->tuple_size
);
569 EXPORT_SYMBOL(bio_integrity_advance
);
572 * bio_integrity_trim - Trim integrity vector
573 * @bio: bio whose integrity vector to update
574 * @offset: offset to first data sector
575 * @sectors: number of data sectors
577 * Description: Used to trim the integrity vector in a cloned bio.
578 * The ivec will be advanced corresponding to 'offset' data sectors
579 * and the length will be truncated corresponding to 'len' data
582 void bio_integrity_trim(struct bio
*bio
, unsigned int offset
, unsigned int sectors
)
584 struct bio_integrity_payload
*bip
= bio
->bi_integrity
;
585 struct blk_integrity
*bi
= bdev_get_integrity(bio
->bi_bdev
);
586 unsigned int nr_sectors
;
590 BUG_ON(!bio_flagged(bio
, BIO_CLONED
));
592 nr_sectors
= bio_integrity_hw_sectors(bi
, sectors
);
593 bip
->bip_sector
= bip
->bip_sector
+ offset
;
594 bio_integrity_mark_head(bip
, offset
* bi
->tuple_size
);
595 bio_integrity_mark_tail(bip
, sectors
* bi
->tuple_size
);
597 EXPORT_SYMBOL(bio_integrity_trim
);
600 * bio_integrity_split - Split integrity metadata
601 * @bio: Protected bio
602 * @bp: Resulting bio_pair
605 * Description: Splits an integrity page into a bio_pair.
607 void bio_integrity_split(struct bio
*bio
, struct bio_pair
*bp
, int sectors
)
609 struct blk_integrity
*bi
;
610 struct bio_integrity_payload
*bip
= bio
->bi_integrity
;
611 unsigned int nr_sectors
;
613 if (bio_integrity(bio
) == 0)
616 bi
= bdev_get_integrity(bio
->bi_bdev
);
618 BUG_ON(bip
->bip_vcnt
!= 1);
620 nr_sectors
= bio_integrity_hw_sectors(bi
, sectors
);
622 bp
->bio1
.bi_integrity
= &bp
->bip1
;
623 bp
->bio2
.bi_integrity
= &bp
->bip2
;
625 bp
->iv1
= bip
->bip_vec
[0];
626 bp
->iv2
= bip
->bip_vec
[0];
628 bp
->bip1
.bip_vec
= &bp
->iv1
;
629 bp
->bip2
.bip_vec
= &bp
->iv2
;
631 bp
->iv1
.bv_len
= sectors
* bi
->tuple_size
;
632 bp
->iv2
.bv_offset
+= sectors
* bi
->tuple_size
;
633 bp
->iv2
.bv_len
-= sectors
* bi
->tuple_size
;
635 bp
->bip1
.bip_sector
= bio
->bi_integrity
->bip_sector
;
636 bp
->bip2
.bip_sector
= bio
->bi_integrity
->bip_sector
+ nr_sectors
;
638 bp
->bip1
.bip_vcnt
= bp
->bip2
.bip_vcnt
= 1;
639 bp
->bip1
.bip_idx
= bp
->bip2
.bip_idx
= 0;
641 EXPORT_SYMBOL(bio_integrity_split
);
644 * bio_integrity_clone - Callback for cloning bios with integrity metadata
646 * @bio_src: Original bio
647 * @bs: bio_set to allocate bip from
649 * Description: Called to allocate a bip when cloning a bio
651 int bio_integrity_clone(struct bio
*bio
, struct bio
*bio_src
, struct bio_set
*bs
)
653 struct bio_integrity_payload
*bip_src
= bio_src
->bi_integrity
;
654 struct bio_integrity_payload
*bip
;
656 BUG_ON(bip_src
== NULL
);
658 bip
= bio_integrity_alloc_bioset(bio
, GFP_NOIO
, bip_src
->bip_vcnt
, bs
);
663 memcpy(bip
->bip_vec
, bip_src
->bip_vec
,
664 bip_src
->bip_vcnt
* sizeof(struct bio_vec
));
666 bip
->bip_sector
= bip_src
->bip_sector
;
667 bip
->bip_vcnt
= bip_src
->bip_vcnt
;
668 bip
->bip_idx
= bip_src
->bip_idx
;
672 EXPORT_SYMBOL(bio_integrity_clone
);
674 int bioset_integrity_create(struct bio_set
*bs
, int pool_size
)
676 bs
->bio_integrity_pool
= mempool_create_slab_pool(pool_size
,
678 if (!bs
->bio_integrity_pool
)
683 EXPORT_SYMBOL(bioset_integrity_create
);
685 void bioset_integrity_free(struct bio_set
*bs
)
687 if (bs
->bio_integrity_pool
)
688 mempool_destroy(bs
->bio_integrity_pool
);
690 EXPORT_SYMBOL(bioset_integrity_free
);
692 void __init
bio_integrity_init_slab(void)
694 bio_integrity_slab
= KMEM_CACHE(bio_integrity_payload
,
695 SLAB_HWCACHE_ALIGN
|SLAB_PANIC
);
697 EXPORT_SYMBOL(bio_integrity_init_slab
);
699 static int __init
integrity_init(void)
701 kintegrityd_wq
= create_workqueue("kintegrityd");
704 panic("Failed to create kintegrityd\n");
708 subsys_initcall(integrity_init
);