[dm-crypt] Help needed please with luks volume

[dm-crypt] Help needed please with luks volume

[Bhasker C V]

Hi all

 Sorry for the cross-list posting; I think this will help.

  I have a luks formatted volume and on debian this volume just stopped
working after a  dist-upgrade
 The error reported is as below


$ sudo cryptsetup luksOpen /dev/sdc dsk
LUKS keyslot 6 is invalid.
LUKS keyslot 7 is invalid.
$ cryptsetup --version
cryptsetup 1.4.1
$ uname -a
Linux bcv 3.2.9 #1 SMP Sat Mar 10 20:36:56 GMT 2012 x86_64 GNU/Linux

Whereas when used in an older version (in my office) this disk works
perfectly fine and I am able to view data.
This home system of mine running debian with cryptsetup version as above
has a problem and so I am not able to
take the data out wheras cryptsetup version cryptsetup 1.0.6 (which runs
kernel 2.6.39-rc6 ) works perfectly okay.

I am stuck with not able to see my private data at home.

Can someone help please ? Or is this a bug introduced in the newer
versions ?

thanks

-- 
Bhasker C V

Re: [dm-crypt] Help needed please with luks volume

[Milan Broz]

On 04/01/2012 09:09 PM, Bhasker C V wrote:
> Hi all
> 
>  Sorry for the cross-list posting; I think this will help.
> 
>   I have a luks formatted volume and on debian this volume just stopped
> working after a  dist-upgrade
>  The error reported is as below
> 
> 
> $ sudo cryptsetup luksOpen /dev/sdc dsk
> LUKS keyslot 6 is invalid.
> LUKS keyslot 7 is invalid.
> $ cryptsetup --version
> cryptsetup 1.4.1
> $ uname -a
> Linux bcv 3.2.9 #1 SMP Sat Mar 10 20:36:56 GMT 2012 x86_64 GNU/Linux
> 
> Whereas when used in an older version (in my office) this disk works
> perfectly fine and I am able to view data.

Hi,

see http://code.google.com/p/cryptsetup/wiki/FrequentlyAskedQuestions#4._Troubleshooting
and http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=662768#10

I think it is exactly your problem, your LUKS header is corrupted
by written partition table signature - but it can be fixed.

If you want, send me (in private mail) the first 4k of your LUKS
device (IOW the visible part, no keyslot data) and I can fix that.

IOW output of "dd if=/dev/<your luks device> of=bck.img bs=1k count=4

> Can someone help please ? Or is this a bug introduced in the newer
> versions ?

See my comment on the bug - it is additional safety check, using
corrupted header can cause data corruption later.

(If there are more cases, I'll enable repair code by default but it
seems this problem is very rare. It is not bug in cryptsetup, just
previous version ignored unsafe keyslot offsets.)

Milan