Re: [gentoo-security] crypto problem mount loop0

[Venkat Manakkal]

I've used cryptoloop with 2.6.0-test9-mm5 and util-linux 2.12. I've been
successfully using this combination with "losetup -e twofish /dev/loop/#
/path/img-file". I do know for sure that no other version of util-linux
worked out. Due to the many images I am using I am stuck with the kernel
and util-linux until I get the time to do all the steps outlined below.

On another machine I'm using 2.6.1-gentoo and loop-aes ontop of the base
util-linux 2.12. I cannot get the second machine to read the twofish
crypt images - it fails with a complaint that the encrytion module does
not accept the key length directive - I don't remember the exact
message.

In other words you MUST go back to the exact version of kernel and
util-linux. Its complaint about unknown file system is because losetup
will succeed with *any* password - only that you get random data on the
other side of the loop unless the encryption is correctly decoded.

So *always* make encrypted backups of all encrypted images before you
update kernel or baselayout (i.e. util-linux). You can use aespipe for
this from the loop-aes package. So your steps would be - go back to old
kernel and old util-linux, mount the cryptfile, cd to the directory,
"tar czpf - .| aespipe -e AES256 -C 128 -T> /path/to/backup.tgz.aes".
Update (or re-boot to new setup), then create a *new* crypt image,
losetup, mount then restore your backup by "cat
/path/to/backup.tgz.aes|aespipe -d|tar xvzpf -". It does not hurt to
make periodic bakups since power failures can be worse on encrypted file
systems.

Hope this helps.

Best regards,

---Venkat.

-----------------------------------------------------------------------
Venkat Manakkal	                          secure 64 bit workstations
President, rayServers.com                 laptops and servers...
venkat@××××××××××.com                     "Installed Secure"
[phone]                           http://www.rayservers.com/
-----------------------------------------------------------------------

On another machine I've got 
On Sun, 22 Feb 2004 15:36:57 +0100
"i.t " <gentoo@×××××××××××.org> wrote:

> hi,
>
> for months I've been successfully working with cryptoloop on a
> cryptfile with a gento 2.4.22-gss kernel;
> some days ago I've updated the baselayout for the system and lateron I
>
> couldn't start losetup anymore.
> I've recompiled the kernel,
> installed gentoo kernel 2.6.3,
> and for comparison on a fedora partition loop-AES-v2.0f.
> rm -f -r test-file[1234] test-dir1
> *** Test results ok ***
>
> So far I get the following results with gentoo kernel2.6.3 and patched
> fedora (2.4.22):
>
> losetup -e AES /dev/loop0 crfile
> Passwort:
> [ok]
> mount /dev/loop0 /mnt/loop/
> mount: you must give the filesystem (Sie muessen den Dateisystemtyp
> angeben)
>
> I'm pretty sure the filesystem was ext3 for the crfile (ok - meanwhile
> I know ext2 is better),
> but when I try:
>
> mount -t ext3 /dev/loop0 /mnt/loop/
> mount:
> (Falscher Dateisystemtyp, ungÃ_Å_ltige Optionen, der
>        Ã_»SuperblockÃ_« von /dev/loop0 ist beschÃ_â_¬digt oder es
>        sind zu viele Dateisysteme eingehÃ_â_¬ngt)
> wrong filesystem, invalid options, or superblock is damaged...
>
> this goes for every filesystem type what is supported.
> With gentoo 2.4.22-gss #4 SMP I get these results:
>
> losetup -e aes /dev/loop0 DorAES
> Password:
> ioctl: LOOP_SET_STATUS: Invalid argument
>
> I'm a bit desperated after all the time and work. What may have
> happenend? Has anybody any idea? Can there be a change in the offset
> size with losetup? Any help really appreciated
>
> thanks
> --
>  . ___
>  |  |
>  |  |
>
>
> --
> gentoo-security@g.o mailing list
>





--
gentoo-security@g.o mailing list