|
Posted by Matthew Fanto on December 18, 2006, 10:51 am
If you were Registered and logged in, you could reply and use other advanced thread options
Sebastian Gottschalk wrote:
> The demand for cryptographic software being open source does not just come
> from trust implications but more from implementation correctness. It's so
> easy to make an entire cryptographic system void with a little
> implementation error, and many companies have proven this. Open source
> gives you and many auditers the opportunity to check the cryptographic core
> and the general quality of the implementation.
But the argument was about backdoors, so I restricted my comment to
backdoors. I think it was Ritchie who showed backdoors by modifying the
compiler. Thus even though the applications code has been reviewed, the
compiler can still insert malicious things.
I'm all in favor of open source software. I was just attempting to show
the fallacy in assuming because it's open source, it's safe.
>
> > Any software you are going to find probably uses AES and TripleDES. DES
> > was a US design, and both AES and DES are certified by NIST.
>
> I presume you still meant 3DES. The classical DES has got its certfication
> revoked some months ago. :-)
Yes, there should be a 3 in front of that DES.
> > DriveCrypt talks about some 1344-bit strength, which in itself
> > nonsense. None of the algorithms listed on their site supports that
> > size key,
>
> Triple-Blowfish (3*448 = 1344)
Good catch. I didn't entertain the idea of 3-BF, and thus didn't think
about 3*448 = 1344. I was thinking of a few different cascade (though I
really didn't think too hard) options and none of them adding to 1344,
and decided the 1344 probably came from some homebrew cipher we see on
sci.crypt so often.
> > and there is no reason for that large of a key anyway.
>
> And neither any good reason why 3-BF should even achieve such a security.
> Paying careful attention to the Meet-in-the-middle attack and using hash
> chains to linearly transfer the memory tradeoff back into a time tradeoff,
> the effective security is always limited to 2*keysize, thus it's only 896
> bits worth.
I agree, but I also think a key size of 896-bits is overkill.
-Matt
| 
XML site map