SecureDrive(IDEA), Realdeal and plaintext attack

[Jean-Francois Avon]

Cc:  cypherpunks at toad.com
	galactus at stack.urc.tue.nl
	Iolo Davidson<iolo at mist.demon.co.uk>

On 18 Jul 96 at 4:19, Edgar Swank wrote:

> JFA wrote:
> > Question:
> >
> > Since realdeal overwrite everything with 0s, and that theses
> > zeroed sectors are encrypted later with IDEA, will that give an
> > attacker an edge?  The attacker will likely know that there are
> > large disk areas that contains 0s.
> >
> > Any comments?


> Yes.  Each sector encrypted by SecureDrive also incorporates a "salt"
> value derived from the sector address and (usually random) volume
> serial.  So encrypted zeroed sectors will be different from each other
> and (without the IDEA key) cannot be distinguished from sectors
> containing data.

IDEA is reputed to be resistant against known plaintext attacks.
But I did not read about wether or not it is resistant to 
several-plaintexts (?choosen plaintext) attack.

If the sectors were not salted, each zeroed sectors would translate in 
an identical way on the encrypted disk.  So, there would be only one 
cyphertext-plaintext pair repeated over many empty sectors.

If you salt the encryptor, there are many different cyphertexts 
corresponding to one single plaintext.

Can the salt be figured out by an attacker?

If yes, would the many-cyphertext to single-ultimate-plaintext could give an 
edge to an attacker?

In that case, it would be effectively better to not wipe a drive with 
zeroes.

The problem is, realdeal cannot be turned selectively for only one 
drive while not wiping the other one.

Thanks for your reply.

JFA
 DePompadour, Societe d'Importation Ltee; Limoges porcelain, silverware and crystal
 JFA Technologies, R&D consultants: physicists, technologists and engineers.

 PGP keys at: http://w3.citenet.net/users/jf_avon
 ID# C58ADD0D : 529645E8205A8A5E F87CC86FAEFEF891