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
) && !bio_flagged(bio
, BIO_FS_INTEGRITY
)
111 && bip
->bip_buf
!= NULL
)
114 bvec_free_bs(bs
, bip
->bip_vec
, bip
->bip_pool
);
115 mempool_free(bip
, bs
->bio_integrity_pool
);
117 bio
->bi_integrity
= NULL
;
119 EXPORT_SYMBOL(bio_integrity_free
);
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
128 * Description: Attach a page containing integrity metadata to bio.
130 int bio_integrity_add_page(struct bio
*bio
, struct page
*page
,
131 unsigned int len
, unsigned int offset
)
133 struct bio_integrity_payload
*bip
= bio
->bi_integrity
;
136 if (bip
->bip_vcnt
>= bvec_nr_vecs(bip
->bip_pool
)) {
137 printk(KERN_ERR
"%s: bip_vec full\n", __func__
);
141 iv
= bip_vec_idx(bip
, bip
->bip_vcnt
);
143 BUG_ON(iv
->bv_page
!= NULL
);
147 iv
->bv_offset
= offset
;
152 EXPORT_SYMBOL(bio_integrity_add_page
);
154 static int bdev_integrity_enabled(struct block_device
*bdev
, int rw
)
156 struct blk_integrity
*bi
= bdev_get_integrity(bdev
);
161 if (rw
== READ
&& bi
->verify_fn
!= NULL
&&
162 (bi
->flags
& INTEGRITY_FLAG_READ
))
165 if (rw
== WRITE
&& bi
->generate_fn
!= NULL
&&
166 (bi
->flags
& INTEGRITY_FLAG_WRITE
))
173 * bio_integrity_enabled - Check whether integrity can be passed
176 * Description: Determines whether bio_integrity_prep() can be called
177 * on this bio or not. bio data direction and target device must be
178 * set prior to calling. The functions honors the write_generate and
179 * read_verify flags in sysfs.
181 int bio_integrity_enabled(struct bio
*bio
)
183 /* Already protected? */
184 if (bio_integrity(bio
))
187 return bdev_integrity_enabled(bio
->bi_bdev
, bio_data_dir(bio
));
189 EXPORT_SYMBOL(bio_integrity_enabled
);
192 * bio_integrity_hw_sectors - Convert 512b sectors to hardware ditto
193 * @bi: blk_integrity profile for device
194 * @sectors: Number of 512 sectors to convert
196 * Description: The block layer calculates everything in 512 byte
197 * sectors but integrity metadata is done in terms of the hardware
198 * sector size of the storage device. Convert the block layer sectors
199 * to physical sectors.
201 static inline unsigned int bio_integrity_hw_sectors(struct blk_integrity
*bi
,
202 unsigned int sectors
)
204 /* At this point there are only 512b or 4096b DIF/EPP devices */
205 if (bi
->sector_size
== 4096)
206 return sectors
>>= 3;
212 * bio_integrity_tag_size - Retrieve integrity tag space
213 * @bio: bio to inspect
215 * Description: Returns the maximum number of tag bytes that can be
216 * attached to this bio. Filesystems can use this to determine how
217 * much metadata to attach to an I/O.
219 unsigned int bio_integrity_tag_size(struct bio
*bio
)
221 struct blk_integrity
*bi
= bdev_get_integrity(bio
->bi_bdev
);
223 BUG_ON(bio
->bi_size
== 0);
225 return bi
->tag_size
* (bio
->bi_size
/ bi
->sector_size
);
227 EXPORT_SYMBOL(bio_integrity_tag_size
);
229 int bio_integrity_tag(struct bio
*bio
, void *tag_buf
, unsigned int len
, int set
)
231 struct bio_integrity_payload
*bip
= bio
->bi_integrity
;
232 struct blk_integrity
*bi
= bdev_get_integrity(bio
->bi_bdev
);
233 unsigned int nr_sectors
;
235 BUG_ON(bip
->bip_buf
== NULL
);
237 if (bi
->tag_size
== 0)
240 nr_sectors
= bio_integrity_hw_sectors(bi
,
241 DIV_ROUND_UP(len
, bi
->tag_size
));
243 if (nr_sectors
* bi
->tuple_size
> bip
->bip_size
) {
244 printk(KERN_ERR
"%s: tag too big for bio: %u > %u\n",
245 __func__
, nr_sectors
* bi
->tuple_size
, bip
->bip_size
);
250 bi
->set_tag_fn(bip
->bip_buf
, tag_buf
, nr_sectors
);
252 bi
->get_tag_fn(bip
->bip_buf
, tag_buf
, nr_sectors
);
258 * bio_integrity_set_tag - Attach a tag buffer to a bio
259 * @bio: bio to attach buffer to
260 * @tag_buf: Pointer to a buffer containing tag data
261 * @len: Length of the included buffer
263 * Description: Use this function to tag a bio by leveraging the extra
264 * space provided by devices formatted with integrity protection. The
265 * size of the integrity buffer must be <= to the size reported by
266 * bio_integrity_tag_size().
268 int bio_integrity_set_tag(struct bio
*bio
, void *tag_buf
, unsigned int len
)
270 BUG_ON(bio_data_dir(bio
) != WRITE
);
272 return bio_integrity_tag(bio
, tag_buf
, len
, 1);
274 EXPORT_SYMBOL(bio_integrity_set_tag
);
277 * bio_integrity_get_tag - Retrieve a tag buffer from a bio
278 * @bio: bio to retrieve buffer from
279 * @tag_buf: Pointer to a buffer for the tag data
280 * @len: Length of the target buffer
282 * Description: Use this function to retrieve the tag buffer from a
283 * completed I/O. The size of the integrity buffer must be <= to the
284 * size reported by bio_integrity_tag_size().
286 int bio_integrity_get_tag(struct bio
*bio
, void *tag_buf
, unsigned int len
)
288 BUG_ON(bio_data_dir(bio
) != READ
);
290 return bio_integrity_tag(bio
, tag_buf
, len
, 0);
292 EXPORT_SYMBOL(bio_integrity_get_tag
);
295 * bio_integrity_generate - Generate integrity metadata for a bio
296 * @bio: bio to generate integrity metadata for
298 * Description: Generates integrity metadata for a bio by calling the
299 * block device's generation callback function. The bio must have a
300 * bip attached with enough room to accommodate the generated
301 * integrity metadata.
303 static void bio_integrity_generate(struct bio
*bio
)
305 struct blk_integrity
*bi
= bdev_get_integrity(bio
->bi_bdev
);
306 struct blk_integrity_exchg bix
;
308 sector_t sector
= bio
->bi_sector
;
309 unsigned int i
, sectors
, total
;
310 void *prot_buf
= bio
->bi_integrity
->bip_buf
;
313 bix
.disk_name
= bio
->bi_bdev
->bd_disk
->disk_name
;
314 bix
.sector_size
= bi
->sector_size
;
316 bio_for_each_segment(bv
, bio
, i
) {
317 void *kaddr
= kmap_atomic(bv
->bv_page
, KM_USER0
);
318 bix
.data_buf
= kaddr
+ bv
->bv_offset
;
319 bix
.data_size
= bv
->bv_len
;
320 bix
.prot_buf
= prot_buf
;
323 bi
->generate_fn(&bix
);
325 sectors
= bv
->bv_len
/ bi
->sector_size
;
327 prot_buf
+= sectors
* bi
->tuple_size
;
328 total
+= sectors
* bi
->tuple_size
;
329 BUG_ON(total
> bio
->bi_integrity
->bip_size
);
331 kunmap_atomic(kaddr
, KM_USER0
);
335 static inline unsigned short blk_integrity_tuple_size(struct blk_integrity
*bi
)
338 return bi
->tuple_size
;
344 * bio_integrity_prep - Prepare bio for integrity I/O
345 * @bio: bio to prepare
347 * Description: Allocates a buffer for integrity metadata, maps the
348 * pages and attaches them to a bio. The bio must have data
349 * direction, target device and start sector set priot to calling. In
350 * the WRITE case, integrity metadata will be generated using the
351 * block device's integrity function. In the READ case, the buffer
352 * will be prepared for DMA and a suitable end_io handler set up.
354 int bio_integrity_prep(struct bio
*bio
)
356 struct bio_integrity_payload
*bip
;
357 struct blk_integrity
*bi
;
358 struct request_queue
*q
;
360 unsigned long start
, end
;
361 unsigned int len
, nr_pages
;
362 unsigned int bytes
, offset
, i
;
363 unsigned int sectors
;
365 bi
= bdev_get_integrity(bio
->bi_bdev
);
366 q
= bdev_get_queue(bio
->bi_bdev
);
368 BUG_ON(bio_integrity(bio
));
370 sectors
= bio_integrity_hw_sectors(bi
, bio_sectors(bio
));
372 /* Allocate kernel buffer for protection data */
373 len
= sectors
* blk_integrity_tuple_size(bi
);
374 buf
= kmalloc(len
, GFP_NOIO
| __GFP_NOFAIL
| q
->bounce_gfp
);
375 if (unlikely(buf
== NULL
)) {
376 printk(KERN_ERR
"could not allocate integrity buffer\n");
380 end
= (((unsigned long) buf
) + len
+ PAGE_SIZE
- 1) >> PAGE_SHIFT
;
381 start
= ((unsigned long) buf
) >> PAGE_SHIFT
;
382 nr_pages
= end
- start
;
384 /* Allocate bio integrity payload and integrity vectors */
385 bip
= bio_integrity_alloc(bio
, GFP_NOIO
, nr_pages
);
386 if (unlikely(bip
== NULL
)) {
387 printk(KERN_ERR
"could not allocate data integrity bioset\n");
394 bip
->bip_sector
= bio
->bi_sector
;
397 offset
= offset_in_page(buf
);
398 for (i
= 0 ; i
< nr_pages
; i
++) {
400 bytes
= PAGE_SIZE
- offset
;
408 ret
= bio_integrity_add_page(bio
, virt_to_page(buf
),
422 /* Install custom I/O completion handler if read verify is enabled */
423 if (bio_data_dir(bio
) == READ
) {
424 bip
->bip_end_io
= bio
->bi_end_io
;
425 bio
->bi_end_io
= bio_integrity_endio
;
428 /* Auto-generate integrity metadata if this is a write */
429 if (bio_data_dir(bio
) == WRITE
)
430 bio_integrity_generate(bio
);
434 EXPORT_SYMBOL(bio_integrity_prep
);
437 * bio_integrity_verify - Verify integrity metadata for a bio
438 * @bio: bio to verify
440 * Description: This function is called to verify the integrity of a
441 * bio. The data in the bio io_vec is compared to the integrity
442 * metadata returned by the HBA.
444 static int bio_integrity_verify(struct bio
*bio
)
446 struct blk_integrity
*bi
= bdev_get_integrity(bio
->bi_bdev
);
447 struct blk_integrity_exchg bix
;
449 sector_t sector
= bio
->bi_integrity
->bip_sector
;
450 unsigned int i
, sectors
, total
, ret
;
451 void *prot_buf
= bio
->bi_integrity
->bip_buf
;
454 bix
.disk_name
= bio
->bi_bdev
->bd_disk
->disk_name
;
455 bix
.sector_size
= bi
->sector_size
;
457 bio_for_each_segment(bv
, bio
, i
) {
458 void *kaddr
= kmap_atomic(bv
->bv_page
, KM_USER0
);
459 bix
.data_buf
= kaddr
+ bv
->bv_offset
;
460 bix
.data_size
= bv
->bv_len
;
461 bix
.prot_buf
= prot_buf
;
464 ret
= bi
->verify_fn(&bix
);
467 kunmap_atomic(kaddr
, KM_USER0
);
471 sectors
= bv
->bv_len
/ bi
->sector_size
;
473 prot_buf
+= sectors
* bi
->tuple_size
;
474 total
+= sectors
* bi
->tuple_size
;
475 BUG_ON(total
> bio
->bi_integrity
->bip_size
);
477 kunmap_atomic(kaddr
, KM_USER0
);
484 * bio_integrity_verify_fn - Integrity I/O completion worker
485 * @work: Work struct stored in bio to be verified
487 * Description: This workqueue function is called to complete a READ
488 * request. The function verifies the transferred integrity metadata
489 * and then calls the original bio end_io function.
491 static void bio_integrity_verify_fn(struct work_struct
*work
)
493 struct bio_integrity_payload
*bip
=
494 container_of(work
, struct bio_integrity_payload
, bip_work
);
495 struct bio
*bio
= bip
->bip_bio
;
496 int error
= bip
->bip_error
;
498 if (bio_integrity_verify(bio
)) {
499 clear_bit(BIO_UPTODATE
, &bio
->bi_flags
);
503 /* Restore original bio completion handler */
504 bio
->bi_end_io
= bip
->bip_end_io
;
507 bio
->bi_end_io(bio
, error
);
511 * bio_integrity_endio - Integrity I/O completion function
512 * @bio: Protected bio
513 * @error: Pointer to errno
515 * Description: Completion for integrity I/O
517 * Normally I/O completion is done in interrupt context. However,
518 * verifying I/O integrity is a time-consuming task which must be run
519 * in process context. This function postpones completion
522 void bio_integrity_endio(struct bio
*bio
, int error
)
524 struct bio_integrity_payload
*bip
= bio
->bi_integrity
;
526 BUG_ON(bip
->bip_bio
!= bio
);
528 bip
->bip_error
= error
;
529 INIT_WORK(&bip
->bip_work
, bio_integrity_verify_fn
);
530 queue_work(kintegrityd_wq
, &bip
->bip_work
);
532 EXPORT_SYMBOL(bio_integrity_endio
);
535 * bio_integrity_mark_head - Advance bip_vec skip bytes
536 * @bip: Integrity vector to advance
537 * @skip: Number of bytes to advance it
539 void bio_integrity_mark_head(struct bio_integrity_payload
*bip
,
545 bip_for_each_vec(iv
, bip
, i
) {
549 } else if (skip
>= iv
->bv_len
) {
551 } else { /* skip < iv->bv_len) */
552 iv
->bv_offset
+= skip
;
561 * bio_integrity_mark_tail - Truncate bip_vec to be len bytes long
562 * @bip: Integrity vector to truncate
563 * @len: New length of integrity vector
565 void bio_integrity_mark_tail(struct bio_integrity_payload
*bip
,
571 bip_for_each_vec(iv
, bip
, i
) {
575 } else if (len
>= iv
->bv_len
) {
577 } else { /* len < iv->bv_len) */
585 * bio_integrity_advance - Advance integrity vector
586 * @bio: bio whose integrity vector to update
587 * @bytes_done: number of data bytes that have been completed
589 * Description: This function calculates how many integrity bytes the
590 * number of completed data bytes correspond to and advances the
591 * integrity vector accordingly.
593 void bio_integrity_advance(struct bio
*bio
, unsigned int bytes_done
)
595 struct bio_integrity_payload
*bip
= bio
->bi_integrity
;
596 struct blk_integrity
*bi
= bdev_get_integrity(bio
->bi_bdev
);
597 unsigned int nr_sectors
;
602 nr_sectors
= bio_integrity_hw_sectors(bi
, bytes_done
>> 9);
603 bio_integrity_mark_head(bip
, nr_sectors
* bi
->tuple_size
);
605 EXPORT_SYMBOL(bio_integrity_advance
);
608 * bio_integrity_trim - Trim integrity vector
609 * @bio: bio whose integrity vector to update
610 * @offset: offset to first data sector
611 * @sectors: number of data sectors
613 * Description: Used to trim the integrity vector in a cloned bio.
614 * The ivec will be advanced corresponding to 'offset' data sectors
615 * and the length will be truncated corresponding to 'len' data
618 void bio_integrity_trim(struct bio
*bio
, unsigned int offset
,
619 unsigned int sectors
)
621 struct bio_integrity_payload
*bip
= bio
->bi_integrity
;
622 struct blk_integrity
*bi
= bdev_get_integrity(bio
->bi_bdev
);
623 unsigned int nr_sectors
;
627 BUG_ON(!bio_flagged(bio
, BIO_CLONED
));
629 nr_sectors
= bio_integrity_hw_sectors(bi
, sectors
);
630 bip
->bip_sector
= bip
->bip_sector
+ offset
;
631 bio_integrity_mark_head(bip
, offset
* bi
->tuple_size
);
632 bio_integrity_mark_tail(bip
, sectors
* bi
->tuple_size
);
634 EXPORT_SYMBOL(bio_integrity_trim
);
637 * bio_integrity_split - Split integrity metadata
638 * @bio: Protected bio
639 * @bp: Resulting bio_pair
642 * Description: Splits an integrity page into a bio_pair.
644 void bio_integrity_split(struct bio
*bio
, struct bio_pair
*bp
, int sectors
)
646 struct blk_integrity
*bi
;
647 struct bio_integrity_payload
*bip
= bio
->bi_integrity
;
648 unsigned int nr_sectors
;
650 if (bio_integrity(bio
) == 0)
653 bi
= bdev_get_integrity(bio
->bi_bdev
);
655 BUG_ON(bip
->bip_vcnt
!= 1);
657 nr_sectors
= bio_integrity_hw_sectors(bi
, sectors
);
659 bp
->bio1
.bi_integrity
= &bp
->bip1
;
660 bp
->bio2
.bi_integrity
= &bp
->bip2
;
662 bp
->iv1
= bip
->bip_vec
[0];
663 bp
->iv2
= bip
->bip_vec
[0];
665 bp
->bip1
.bip_vec
= &bp
->iv1
;
666 bp
->bip2
.bip_vec
= &bp
->iv2
;
668 bp
->iv1
.bv_len
= sectors
* bi
->tuple_size
;
669 bp
->iv2
.bv_offset
+= sectors
* bi
->tuple_size
;
670 bp
->iv2
.bv_len
-= sectors
* bi
->tuple_size
;
672 bp
->bip1
.bip_sector
= bio
->bi_integrity
->bip_sector
;
673 bp
->bip2
.bip_sector
= bio
->bi_integrity
->bip_sector
+ nr_sectors
;
675 bp
->bip1
.bip_vcnt
= bp
->bip2
.bip_vcnt
= 1;
676 bp
->bip1
.bip_idx
= bp
->bip2
.bip_idx
= 0;
678 EXPORT_SYMBOL(bio_integrity_split
);
681 * bio_integrity_clone - Callback for cloning bios with integrity metadata
683 * @bio_src: Original bio
684 * @bs: bio_set to allocate bip from
686 * Description: Called to allocate a bip when cloning a bio
688 int bio_integrity_clone(struct bio
*bio
, struct bio
*bio_src
,
691 struct bio_integrity_payload
*bip_src
= bio_src
->bi_integrity
;
692 struct bio_integrity_payload
*bip
;
694 BUG_ON(bip_src
== NULL
);
696 bip
= bio_integrity_alloc_bioset(bio
, GFP_NOIO
, bip_src
->bip_vcnt
, bs
);
701 memcpy(bip
->bip_vec
, bip_src
->bip_vec
,
702 bip_src
->bip_vcnt
* sizeof(struct bio_vec
));
704 bip
->bip_sector
= bip_src
->bip_sector
;
705 bip
->bip_vcnt
= bip_src
->bip_vcnt
;
706 bip
->bip_idx
= bip_src
->bip_idx
;
710 EXPORT_SYMBOL(bio_integrity_clone
);
712 int bioset_integrity_create(struct bio_set
*bs
, int pool_size
)
714 bs
->bio_integrity_pool
= mempool_create_slab_pool(pool_size
,
716 if (!bs
->bio_integrity_pool
)
721 EXPORT_SYMBOL(bioset_integrity_create
);
723 void bioset_integrity_free(struct bio_set
*bs
)
725 if (bs
->bio_integrity_pool
)
726 mempool_destroy(bs
->bio_integrity_pool
);
728 EXPORT_SYMBOL(bioset_integrity_free
);
730 void __init
bio_integrity_init_slab(void)
732 bio_integrity_slab
= KMEM_CACHE(bio_integrity_payload
,
733 SLAB_HWCACHE_ALIGN
|SLAB_PANIC
);
736 static int __init
integrity_init(void)
738 kintegrityd_wq
= create_workqueue("kintegrityd");
741 panic("Failed to create kintegrityd\n");
745 subsys_initcall(integrity_init
);