Re: Distinguishability of encrypted partition

[Florian Reitmeir <fr@xxxxxxxxx>]

Hi,

On Mon, 19 Jun 2006, Phil H wrote:
> Thanks for the replies - only just saw them since my yahoo bulk folder is so full of junk.

> I suppose I was thinking of watermark-type attacks, showing there actually is a filesystem in that randomness (my understanding is that v3.x loop-aes should be immune to these?), or some type of mathematical investigation designed to show the partition has not been recently overwritten by shred or somesuch but has a suspicious form of "randomness" (if such an investigation exists that is). 

i'cant help it, but i think the discussion about possible watermark attempts
on are disk useless.

Watermark attempts has to assume the cipher/modes you use... 

Normally there is enough evidence on a PC/Laptop that there is Crypto..
somewhere is a kernel with initrd, and normally such a pc asks for a password
after boot.

Not to mention the rest of the computer hardware, like the SMART-disk log
which counts nicely how mant houres the drive was used, how many errors
happend, and maybe makes some read/write stats.

Much more interessting would it be to work on support for PKCS#11 USB tokens
to get rid of the user-password. Only a minority boot from an crypto disk, so a
trojan or "other" software can easily sniff your password (with X11 this is
really very simple).

> So my initial assumption was probably correct - it's the extraneous factors (fstab,  having encryption software, etc) that probably remain the practical indicators.

> Christian Kujau <evil@xxxxxxxxxx> wrote: On Wed, 14 Jun 2006, Florian Reitmeir wrote:
> > "The" evils have much simpler ways to "crack" your security, a common
> > way (rumors) is, that
> >
> > - "they" grab all your computer staff
> > - see its encrypted
> 
> s/see/assume/  ...as they can't be sure and probably won't hire a 
> cryptoexpert to prove this, methinks.
> 
> > - return the computer
> > - ... with an keylogger, small on the mainboard/keyboard/usb-bus/...
> > - then, come about 2 weeks/months later again
> 
> there we go again: triple-aes-1024 won't help if the cryptosystem is lame

-- 
Florian Reitmeir

-
Linux-crypto:  cryptography in and on the Linux system
Archive:       http://mail.nl.linux.org/linux-crypto/