Starting from commit e36f620428(block: split bios to max possible length),
block core starts to split bio in the middle of bvec.
Unfortunately loop dio/aio doesn't consider this situation, and
always treat 'iter.iov_offset' as zero. Then filesystem corruption
is observed.
This patch figures out the offset of the base bvevc via
'bio->bi_iter.bi_bvec_done' and fixes the issue by passing the offset
to iov iterator.
Fixes: e36f6204288088f (block: split bios to max possible length)
Cc: Keith Busch <[email protected]>
Cc: Al Viro <[email protected]>
Cc: [email protected] (4.5)
Signed-off-by: Ming Lei <[email protected]>
---
drivers/block/loop.c | 6 ++++++
1 file changed, 6 insertions(+)
diff --git a/drivers/block/loop.c b/drivers/block/loop.c
index 7e5e27a..1fa8cc2 100644
--- a/drivers/block/loop.c
+++ b/drivers/block/loop.c
@@ -488,6 +488,12 @@ static int lo_rw_aio(struct loop_device *lo, struct loop_cmd *cmd,
bvec = __bvec_iter_bvec(bio->bi_io_vec, bio->bi_iter);
iov_iter_bvec(&iter, ITER_BVEC | rw, bvec,
bio_segments(bio), blk_rq_bytes(cmd->rq));
+ /*
+ * This bio may be started from the middle of the 'bvec'
+ * because of bio splitting, so offset from the bvec must
+ * be passed to iov iterator
+ */
+ iter.iov_offset = bio->bi_iter.bi_bvec_done;
cmd->iocb.ki_pos = pos;
cmd->iocb.ki_filp = file;
--
1.9.1
-------- Forwarded Message --------
Subject: Re: [PATCH] block: loop: fix filesystem corruption in case of aio/dio
Date: Fri, 15 Apr 2016 21:16:42 +0800
From: Ming Lei <[email protected]>
To: Holger Hoffstätte <[email protected]>
On Fri, Apr 15, 2016 at 8:51 PM, Holger Hoffstätte
<[email protected]> wrote:
> (off-list since I'm likely missing something..)
>
> Hello Ming Lei,
>
> quick question -
>
> On 04/15/16 12:51, Ming Lei wrote:
>> Starting from commit e36f620428(block: split bios to max possible length),
>> block core starts to split bio in the middle of bvec.
>>
>> Unfortunately loop dio/aio doesn't consider this situation, and
>> always treat 'iter.iov_offset' as zero. Then filesystem corruption
>> is observed.
>>
>> This patch figures out the offset of the base bvevc via
>> 'bio->bi_iter.bi_bvec_done' and fixes the issue by passing the offset
>> to iov iterator.
>>
>> Fixes: e36f6204288088f (block: split bios to max possible length)
>> Cc: Keith Busch <[email protected]>
>> Cc: Al Viro <[email protected]>
>> Cc: [email protected] (4.5)
> ^^^^^
>
> The previous patch went into 4.4.x-stable, so should this one not go
> into 4.4.x as well? It applies cleanly and does not seem to hurt.
> Apologies in advance if this is a bad idea. :)
You are absolutely right, this patch should have been marked as stable 4.4+,
thanks for point it out!
Jens, please let me know if you need me to resend the patch for fixing
the stable tag.
Thanks,
Ming
On 04/15/2016 04:51 AM, Ming Lei wrote:
> Starting from commit e36f620428(block: split bios to max possible length),
> block core starts to split bio in the middle of bvec.
>
> Unfortunately loop dio/aio doesn't consider this situation, and
> always treat 'iter.iov_offset' as zero. Then filesystem corruption
> is observed.
>
> This patch figures out the offset of the base bvevc via
> 'bio->bi_iter.bi_bvec_done' and fixes the issue by passing the offset
> to iov iterator.
>
> Fixes: e36f6204288088f (block: split bios to max possible length)
> Cc: Keith Busch <[email protected]>
> Cc: Al Viro <[email protected]>
> Cc: [email protected] (4.5)
> Signed-off-by: Ming Lei <[email protected]>
Thanks Ming, queued up.
--
Jens Axboe