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
,
47 struct bio_integrity_payload
*bip
;
53 bip
= mempool_alloc(bs
->bio_integrity_pool
, gfp_mask
);
54 if (unlikely(bip
== NULL
)) {
55 printk(KERN_ERR
"%s: could not alloc bip\n", __func__
);
59 memset(bip
, 0, sizeof(*bip
));
61 iv
= bvec_alloc_bs(gfp_mask
, nr_vecs
, &idx
, bs
);
62 if (unlikely(iv
== NULL
)) {
63 printk(KERN_ERR
"%s: could not alloc bip_vec\n", __func__
);
64 mempool_free(bip
, bs
->bio_integrity_pool
);
71 bio
->bi_integrity
= bip
;
75 EXPORT_SYMBOL(bio_integrity_alloc_bioset
);
78 * bio_integrity_alloc - Allocate integrity payload and attach it to bio
79 * @bio: bio to attach integrity metadata to
80 * @gfp_mask: Memory allocation mask
81 * @nr_vecs: Number of integrity metadata scatter-gather elements
83 * Description: This function prepares a bio for attaching integrity
84 * metadata. nr_vecs specifies the maximum number of pages containing
85 * integrity metadata that can be attached.
87 struct bio_integrity_payload
*bio_integrity_alloc(struct bio
*bio
,
91 return bio_integrity_alloc_bioset(bio
, gfp_mask
, nr_vecs
, fs_bio_set
);
93 EXPORT_SYMBOL(bio_integrity_alloc
);
96 * bio_integrity_free - Free bio integrity payload
97 * @bio: bio containing bip to be freed
98 * @bs: bio_set this bio was allocated from
100 * Description: Used to free the integrity portion of a bio. Usually
101 * called from bio_free().
103 void bio_integrity_free(struct bio
*bio
, struct bio_set
*bs
)
105 struct bio_integrity_payload
*bip
= bio
->bi_integrity
;
109 /* A cloned bio doesn't own the integrity metadata */
110 if (!bio_flagged(bio
, BIO_CLONED
) && bip
->bip_buf
!= NULL
)
113 mempool_free(bip
->bip_vec
, bs
->bvec_pools
[bip
->bip_pool
]);
114 mempool_free(bip
, bs
->bio_integrity_pool
);
116 bio
->bi_integrity
= NULL
;
118 EXPORT_SYMBOL(bio_integrity_free
);
121 * bio_integrity_add_page - Attach integrity metadata
122 * @bio: bio to update
123 * @page: page containing integrity metadata
124 * @len: number of bytes of integrity metadata in page
125 * @offset: start offset within page
127 * Description: Attach a page containing integrity metadata to bio.
129 int bio_integrity_add_page(struct bio
*bio
, struct page
*page
,
130 unsigned int len
, unsigned int offset
)
132 struct bio_integrity_payload
*bip
= bio
->bi_integrity
;
135 if (bip
->bip_vcnt
>= bvec_nr_vecs(bip
->bip_pool
)) {
136 printk(KERN_ERR
"%s: bip_vec full\n", __func__
);
140 iv
= bip_vec_idx(bip
, bip
->bip_vcnt
);
142 BUG_ON(iv
->bv_page
!= NULL
);
146 iv
->bv_offset
= offset
;
151 EXPORT_SYMBOL(bio_integrity_add_page
);
153 static struct blk_integrity
*bdev_get_integrity(struct block_device
*bdev
)
155 return bdev
->bd_disk
->integrity
;
158 static int bdev_integrity_enabled(struct block_device
*bdev
, int rw
)
160 struct blk_integrity
*bi
= bdev_get_integrity(bdev
);
165 if (rw
== READ
&& bi
->verify_fn
!= NULL
&&
166 (bi
->flags
& INTEGRITY_FLAG_READ
))
169 if (rw
== WRITE
&& bi
->generate_fn
!= NULL
&&
170 (bi
->flags
& INTEGRITY_FLAG_WRITE
))
177 * bio_integrity_enabled - Check whether integrity can be passed
180 * Description: Determines whether bio_integrity_prep() can be called
181 * on this bio or not. bio data direction and target device must be
182 * set prior to calling. The functions honors the write_generate and
183 * read_verify flags in sysfs.
185 int bio_integrity_enabled(struct bio
*bio
)
187 /* Already protected? */
188 if (bio_integrity(bio
))
191 return bdev_integrity_enabled(bio
->bi_bdev
, bio_data_dir(bio
));
193 EXPORT_SYMBOL(bio_integrity_enabled
);
196 * bio_integrity_hw_sectors - Convert 512b sectors to hardware ditto
197 * @bi: blk_integrity profile for device
198 * @sectors: Number of 512 sectors to convert
200 * Description: The block layer calculates everything in 512 byte
201 * sectors but integrity metadata is done in terms of the hardware
202 * sector size of the storage device. Convert the block layer sectors
203 * to physical sectors.
205 static inline unsigned int bio_integrity_hw_sectors(struct blk_integrity
*bi
,
206 unsigned int sectors
)
208 /* At this point there are only 512b or 4096b DIF/EPP devices */
209 if (bi
->sector_size
== 4096)
210 return sectors
>>= 3;
216 * bio_integrity_tag_size - Retrieve integrity tag space
217 * @bio: bio to inspect
219 * Description: Returns the maximum number of tag bytes that can be
220 * attached to this bio. Filesystems can use this to determine how
221 * much metadata to attach to an I/O.
223 unsigned int bio_integrity_tag_size(struct bio
*bio
)
225 struct blk_integrity
*bi
= bdev_get_integrity(bio
->bi_bdev
);
227 BUG_ON(bio
->bi_size
== 0);
229 return bi
->tag_size
* (bio
->bi_size
/ bi
->sector_size
);
231 EXPORT_SYMBOL(bio_integrity_tag_size
);
233 int bio_integrity_tag(struct bio
*bio
, void *tag_buf
, unsigned int len
, int set
)
235 struct bio_integrity_payload
*bip
= bio
->bi_integrity
;
236 struct blk_integrity
*bi
= bdev_get_integrity(bio
->bi_bdev
);
237 unsigned int nr_sectors
;
239 BUG_ON(bip
->bip_buf
== NULL
);
241 if (bi
->tag_size
== 0)
244 nr_sectors
= bio_integrity_hw_sectors(bi
,
245 DIV_ROUND_UP(len
, bi
->tag_size
));
247 if (nr_sectors
* bi
->tuple_size
> bip
->bip_size
) {
248 printk(KERN_ERR
"%s: tag too big for bio: %u > %u\n",
249 __func__
, nr_sectors
* bi
->tuple_size
, bip
->bip_size
);
254 bi
->set_tag_fn(bip
->bip_buf
, tag_buf
, nr_sectors
);
256 bi
->get_tag_fn(bip
->bip_buf
, tag_buf
, nr_sectors
);
262 * bio_integrity_set_tag - Attach a tag buffer to a bio
263 * @bio: bio to attach buffer to
264 * @tag_buf: Pointer to a buffer containing tag data
265 * @len: Length of the included buffer
267 * Description: Use this function to tag a bio by leveraging the extra
268 * space provided by devices formatted with integrity protection. The
269 * size of the integrity buffer must be <= to the size reported by
270 * bio_integrity_tag_size().
272 int bio_integrity_set_tag(struct bio
*bio
, void *tag_buf
, unsigned int len
)
274 BUG_ON(bio_data_dir(bio
) != WRITE
);
276 return bio_integrity_tag(bio
, tag_buf
, len
, 1);
278 EXPORT_SYMBOL(bio_integrity_set_tag
);
281 * bio_integrity_get_tag - Retrieve a tag buffer from a bio
282 * @bio: bio to retrieve buffer from
283 * @tag_buf: Pointer to a buffer for the tag data
284 * @len: Length of the target buffer
286 * Description: Use this function to retrieve the tag buffer from a
287 * completed I/O. The size of the integrity buffer must be <= to the
288 * size reported by bio_integrity_tag_size().
290 int bio_integrity_get_tag(struct bio
*bio
, void *tag_buf
, unsigned int len
)
292 BUG_ON(bio_data_dir(bio
) != READ
);
294 return bio_integrity_tag(bio
, tag_buf
, len
, 0);
296 EXPORT_SYMBOL(bio_integrity_get_tag
);
299 * bio_integrity_generate - Generate integrity metadata for a bio
300 * @bio: bio to generate integrity metadata for
302 * Description: Generates integrity metadata for a bio by calling the
303 * block device's generation callback function. The bio must have a
304 * bip attached with enough room to accommodate the generated
305 * integrity metadata.
307 static void bio_integrity_generate(struct bio
*bio
)
309 struct blk_integrity
*bi
= bdev_get_integrity(bio
->bi_bdev
);
310 struct blk_integrity_exchg bix
;
312 sector_t sector
= bio
->bi_sector
;
313 unsigned int i
, sectors
, total
;
314 void *prot_buf
= bio
->bi_integrity
->bip_buf
;
317 bix
.disk_name
= bio
->bi_bdev
->bd_disk
->disk_name
;
318 bix
.sector_size
= bi
->sector_size
;
320 bio_for_each_segment(bv
, bio
, i
) {
321 void *kaddr
= kmap_atomic(bv
->bv_page
, KM_USER0
);
322 bix
.data_buf
= kaddr
+ bv
->bv_offset
;
323 bix
.data_size
= bv
->bv_len
;
324 bix
.prot_buf
= prot_buf
;
327 bi
->generate_fn(&bix
);
329 sectors
= bv
->bv_len
/ bi
->sector_size
;
331 prot_buf
+= sectors
* bi
->tuple_size
;
332 total
+= sectors
* bi
->tuple_size
;
333 BUG_ON(total
> bio
->bi_integrity
->bip_size
);
335 kunmap_atomic(kaddr
, KM_USER0
);
339 static inline unsigned short blk_integrity_tuple_size(struct blk_integrity
*bi
)
342 return bi
->tuple_size
;
348 * bio_integrity_prep - Prepare bio for integrity I/O
349 * @bio: bio to prepare
351 * Description: Allocates a buffer for integrity metadata, maps the
352 * pages and attaches them to a bio. The bio must have data
353 * direction, target device and start sector set priot to calling. In
354 * the WRITE case, integrity metadata will be generated using the
355 * block device's integrity function. In the READ case, the buffer
356 * will be prepared for DMA and a suitable end_io handler set up.
358 int bio_integrity_prep(struct bio
*bio
)
360 struct bio_integrity_payload
*bip
;
361 struct blk_integrity
*bi
;
362 struct request_queue
*q
;
364 unsigned long start
, end
;
365 unsigned int len
, nr_pages
;
366 unsigned int bytes
, offset
, i
;
367 unsigned int sectors
;
369 bi
= bdev_get_integrity(bio
->bi_bdev
);
370 q
= bdev_get_queue(bio
->bi_bdev
);
372 BUG_ON(bio_integrity(bio
));
374 sectors
= bio_integrity_hw_sectors(bi
, bio_sectors(bio
));
376 /* Allocate kernel buffer for protection data */
377 len
= sectors
* blk_integrity_tuple_size(bi
);
378 buf
= kmalloc(len
, GFP_NOIO
| __GFP_NOFAIL
| q
->bounce_gfp
);
379 if (unlikely(buf
== NULL
)) {
380 printk(KERN_ERR
"could not allocate integrity buffer\n");
384 end
= (((unsigned long) buf
) + len
+ PAGE_SIZE
- 1) >> PAGE_SHIFT
;
385 start
= ((unsigned long) buf
) >> PAGE_SHIFT
;
386 nr_pages
= end
- start
;
388 /* Allocate bio integrity payload and integrity vectors */
389 bip
= bio_integrity_alloc(bio
, GFP_NOIO
, nr_pages
);
390 if (unlikely(bip
== NULL
)) {
391 printk(KERN_ERR
"could not allocate data integrity bioset\n");
398 bip
->bip_sector
= bio
->bi_sector
;
401 offset
= offset_in_page(buf
);
402 for (i
= 0 ; i
< nr_pages
; i
++) {
404 bytes
= PAGE_SIZE
- offset
;
412 ret
= bio_integrity_add_page(bio
, virt_to_page(buf
),
426 /* Install custom I/O completion handler if read verify is enabled */
427 if (bio_data_dir(bio
) == READ
) {
428 bip
->bip_end_io
= bio
->bi_end_io
;
429 bio
->bi_end_io
= bio_integrity_endio
;
432 /* Auto-generate integrity metadata if this is a write */
433 if (bio_data_dir(bio
) == WRITE
)
434 bio_integrity_generate(bio
);
438 EXPORT_SYMBOL(bio_integrity_prep
);
441 * bio_integrity_verify - Verify integrity metadata for a bio
442 * @bio: bio to verify
444 * Description: This function is called to verify the integrity of a
445 * bio. The data in the bio io_vec is compared to the integrity
446 * metadata returned by the HBA.
448 static int bio_integrity_verify(struct bio
*bio
)
450 struct blk_integrity
*bi
= bdev_get_integrity(bio
->bi_bdev
);
451 struct blk_integrity_exchg bix
;
453 sector_t sector
= bio
->bi_integrity
->bip_sector
;
454 unsigned int i
, sectors
, total
, ret
;
455 void *prot_buf
= bio
->bi_integrity
->bip_buf
;
458 bix
.disk_name
= bio
->bi_bdev
->bd_disk
->disk_name
;
459 bix
.sector_size
= bi
->sector_size
;
461 bio_for_each_segment(bv
, bio
, i
) {
462 void *kaddr
= kmap_atomic(bv
->bv_page
, KM_USER0
);
463 bix
.data_buf
= kaddr
+ bv
->bv_offset
;
464 bix
.data_size
= bv
->bv_len
;
465 bix
.prot_buf
= prot_buf
;
468 ret
= bi
->verify_fn(&bix
);
471 kunmap_atomic(kaddr
, KM_USER0
);
475 sectors
= bv
->bv_len
/ bi
->sector_size
;
477 prot_buf
+= sectors
* bi
->tuple_size
;
478 total
+= sectors
* bi
->tuple_size
;
479 BUG_ON(total
> bio
->bi_integrity
->bip_size
);
481 kunmap_atomic(kaddr
, KM_USER0
);
488 * bio_integrity_verify_fn - Integrity I/O completion worker
489 * @work: Work struct stored in bio to be verified
491 * Description: This workqueue function is called to complete a READ
492 * request. The function verifies the transferred integrity metadata
493 * and then calls the original bio end_io function.
495 static void bio_integrity_verify_fn(struct work_struct
*work
)
497 struct bio_integrity_payload
*bip
=
498 container_of(work
, struct bio_integrity_payload
, bip_work
);
499 struct bio
*bio
= bip
->bip_bio
;
500 int error
= bip
->bip_error
;
502 if (bio_integrity_verify(bio
)) {
503 clear_bit(BIO_UPTODATE
, &bio
->bi_flags
);
507 /* Restore original bio completion handler */
508 bio
->bi_end_io
= bip
->bip_end_io
;
511 bio
->bi_end_io(bio
, error
);
515 * bio_integrity_endio - Integrity I/O completion function
516 * @bio: Protected bio
517 * @error: Pointer to errno
519 * Description: Completion for integrity I/O
521 * Normally I/O completion is done in interrupt context. However,
522 * verifying I/O integrity is a time-consuming task which must be run
523 * in process context. This function postpones completion
526 void bio_integrity_endio(struct bio
*bio
, int error
)
528 struct bio_integrity_payload
*bip
= bio
->bi_integrity
;
530 BUG_ON(bip
->bip_bio
!= bio
);
532 bip
->bip_error
= error
;
533 INIT_WORK(&bip
->bip_work
, bio_integrity_verify_fn
);
534 queue_work(kintegrityd_wq
, &bip
->bip_work
);
536 EXPORT_SYMBOL(bio_integrity_endio
);
539 * bio_integrity_mark_head - Advance bip_vec skip bytes
540 * @bip: Integrity vector to advance
541 * @skip: Number of bytes to advance it
543 void bio_integrity_mark_head(struct bio_integrity_payload
*bip
,
549 bip_for_each_vec(iv
, bip
, i
) {
553 } else if (skip
>= iv
->bv_len
) {
555 } else { /* skip < iv->bv_len) */
556 iv
->bv_offset
+= skip
;
565 * bio_integrity_mark_tail - Truncate bip_vec to be len bytes long
566 * @bip: Integrity vector to truncate
567 * @len: New length of integrity vector
569 void bio_integrity_mark_tail(struct bio_integrity_payload
*bip
,
575 bip_for_each_vec(iv
, bip
, i
) {
579 } else if (len
>= iv
->bv_len
) {
581 } else { /* len < iv->bv_len) */
589 * bio_integrity_advance - Advance integrity vector
590 * @bio: bio whose integrity vector to update
591 * @bytes_done: number of data bytes that have been completed
593 * Description: This function calculates how many integrity bytes the
594 * number of completed data bytes correspond to and advances the
595 * integrity vector accordingly.
597 void bio_integrity_advance(struct bio
*bio
, unsigned int bytes_done
)
599 struct bio_integrity_payload
*bip
= bio
->bi_integrity
;
600 struct blk_integrity
*bi
= bdev_get_integrity(bio
->bi_bdev
);
601 unsigned int nr_sectors
;
606 nr_sectors
= bio_integrity_hw_sectors(bi
, bytes_done
>> 9);
607 bio_integrity_mark_head(bip
, nr_sectors
* bi
->tuple_size
);
609 EXPORT_SYMBOL(bio_integrity_advance
);
612 * bio_integrity_trim - Trim integrity vector
613 * @bio: bio whose integrity vector to update
614 * @offset: offset to first data sector
615 * @sectors: number of data sectors
617 * Description: Used to trim the integrity vector in a cloned bio.
618 * The ivec will be advanced corresponding to 'offset' data sectors
619 * and the length will be truncated corresponding to 'len' data
622 void bio_integrity_trim(struct bio
*bio
, unsigned int offset
,
623 unsigned int sectors
)
625 struct bio_integrity_payload
*bip
= bio
->bi_integrity
;
626 struct blk_integrity
*bi
= bdev_get_integrity(bio
->bi_bdev
);
627 unsigned int nr_sectors
;
631 BUG_ON(!bio_flagged(bio
, BIO_CLONED
));
633 nr_sectors
= bio_integrity_hw_sectors(bi
, sectors
);
634 bip
->bip_sector
= bip
->bip_sector
+ offset
;
635 bio_integrity_mark_head(bip
, offset
* bi
->tuple_size
);
636 bio_integrity_mark_tail(bip
, sectors
* bi
->tuple_size
);
638 EXPORT_SYMBOL(bio_integrity_trim
);
641 * bio_integrity_split - Split integrity metadata
642 * @bio: Protected bio
643 * @bp: Resulting bio_pair
646 * Description: Splits an integrity page into a bio_pair.
648 void bio_integrity_split(struct bio
*bio
, struct bio_pair
*bp
, int sectors
)
650 struct blk_integrity
*bi
;
651 struct bio_integrity_payload
*bip
= bio
->bi_integrity
;
652 unsigned int nr_sectors
;
654 if (bio_integrity(bio
) == 0)
657 bi
= bdev_get_integrity(bio
->bi_bdev
);
659 BUG_ON(bip
->bip_vcnt
!= 1);
661 nr_sectors
= bio_integrity_hw_sectors(bi
, sectors
);
663 bp
->bio1
.bi_integrity
= &bp
->bip1
;
664 bp
->bio2
.bi_integrity
= &bp
->bip2
;
666 bp
->iv1
= bip
->bip_vec
[0];
667 bp
->iv2
= bip
->bip_vec
[0];
669 bp
->bip1
.bip_vec
= &bp
->iv1
;
670 bp
->bip2
.bip_vec
= &bp
->iv2
;
672 bp
->iv1
.bv_len
= sectors
* bi
->tuple_size
;
673 bp
->iv2
.bv_offset
+= sectors
* bi
->tuple_size
;
674 bp
->iv2
.bv_len
-= sectors
* bi
->tuple_size
;
676 bp
->bip1
.bip_sector
= bio
->bi_integrity
->bip_sector
;
677 bp
->bip2
.bip_sector
= bio
->bi_integrity
->bip_sector
+ nr_sectors
;
679 bp
->bip1
.bip_vcnt
= bp
->bip2
.bip_vcnt
= 1;
680 bp
->bip1
.bip_idx
= bp
->bip2
.bip_idx
= 0;
682 EXPORT_SYMBOL(bio_integrity_split
);
685 * bio_integrity_clone - Callback for cloning bios with integrity metadata
687 * @bio_src: Original bio
688 * @bs: bio_set to allocate bip from
690 * Description: Called to allocate a bip when cloning a bio
692 int bio_integrity_clone(struct bio
*bio
, struct bio
*bio_src
,
695 struct bio_integrity_payload
*bip_src
= bio_src
->bi_integrity
;
696 struct bio_integrity_payload
*bip
;
698 BUG_ON(bip_src
== NULL
);
700 bip
= bio_integrity_alloc_bioset(bio
, GFP_NOIO
, bip_src
->bip_vcnt
, bs
);
705 memcpy(bip
->bip_vec
, bip_src
->bip_vec
,
706 bip_src
->bip_vcnt
* sizeof(struct bio_vec
));
708 bip
->bip_sector
= bip_src
->bip_sector
;
709 bip
->bip_vcnt
= bip_src
->bip_vcnt
;
710 bip
->bip_idx
= bip_src
->bip_idx
;
714 EXPORT_SYMBOL(bio_integrity_clone
);
716 int bioset_integrity_create(struct bio_set
*bs
, int pool_size
)
718 bs
->bio_integrity_pool
= mempool_create_slab_pool(pool_size
,
720 if (!bs
->bio_integrity_pool
)
725 EXPORT_SYMBOL(bioset_integrity_create
);
727 void bioset_integrity_free(struct bio_set
*bs
)
729 if (bs
->bio_integrity_pool
)
730 mempool_destroy(bs
->bio_integrity_pool
);
732 EXPORT_SYMBOL(bioset_integrity_free
);
734 void __init
bio_integrity_init_slab(void)
736 bio_integrity_slab
= KMEM_CACHE(bio_integrity_payload
,
737 SLAB_HWCACHE_ALIGN
|SLAB_PANIC
);
740 static int __init
integrity_init(void)
742 kintegrityd_wq
= create_workqueue("kintegrityd");
745 panic("Failed to create kintegrityd\n");
749 subsys_initcall(integrity_init
);