session-key proposal

[Noll Janos <johnzero@johnzero.hu>]

Hi!

I have the following idea about implementing session keys in loop-AES,
and I'd like to hear what you think of it.

Goals:

 1. The user must be able to change the password of the encrypted device
 2. More than one user has to be able to access (unlock) the
    device (w/password)
 3. Users (passwords) can be added or removed as time goes on

Definitions:

 1. Every losetup'd virtual device would have it's own SESSIONKEY
    (random, generated)
 2. Every device would have a random SEED (also generated)
 3. Every device could have one or more PASSWORDs, which can "unlock" it.
    The user only has to know one of the PASSWORDs.

Implementation thoughts:

 1. On the device, the first 32(?) kbytes of space would be "reserved",
    and used by the encryption system.
    This equals to an "offset+=32768", easy to implement.
 2. For each PASSWORD, there would be an encoded datablock, which
    would contain the SESSIONKEY and the SEED, encrypted.
 3. Functionality gets more complex.
    You need:
    - initialization/creation phase
    - unlocking (with a password) function
    - add/modify/remove/list keys

Possible weakness(es):

 1. If someone knows one PASSWORD, he might possibly be able to get
    the SESSIONKEY, too. This way, password "revokation" can be nulled.

    If the SESSIONKEY is held in kernel-space, the raw device cannot be
    read and the user has no root access, this is not a problem.

 2. The "encoded datablock", which contains the keys might be easier to
    crack, by brute force, if it's very small.


| Noll Janos <johnzero@johnzero.hu> | http://www.johnzero.hu |
| "Expect the unexpected!"    |   ICQ# 4547866   |  Be free! |
-
Linux-crypto:  cryptography in and on the Linux system
Archive:       http://mail.nl.linux.org/linux-crypto/