-----BEGIN PGP SIGNED MESSAGE----- [To: Cypherpunks, Lucky Green ## Date: 01/24/96 12:33 pm ## Subject: Re: Lotus Notes]
Date: Tue, 23 Jan 1996 23:24:39 -0800 From: shamrock@netcom.com (Lucky Green) Subject: Re: Lotus Notes
You are assuming that they *want* the hole to be unpatchable. I see no reason why they should. "We tried out best, but these darn hackers found a way to enable full 64 bits. Sorry, but we tried." Perhaps the most intelligent thing to do was to keep the GAK subject to a simple patch.
I'm sorry, I don't think I was very clear in this post. I wasn't concerned with whether Lotus left the escrow feature easy to disable, I wanted to know whether they'd intelligently padded their RSA-encrypted 24-bit key leak. If they thought this through, they did, but if not, then they have essentially left their exportable security level at 40 bits, because of the dictionary attack David and some other people pointed out. This ought to be relatively easy to check from disassembled code, but it can also be checked by simply generating a few thousand messages (maybe six or seven thousand, to be safe), and seeing whether or not we ever get a duplicate LEAF. We expect to, after about 2^12 encryptions, if they're using fixed padding. Of course, RSA key exchange blobs for short keys must always be padded out like this, or be vulnerable to dictionary attacks. P.S. Does anyone know whether or not the RSA key used to partially escrow the session key is a reasonable length (i.e., 1024 bits)? If it's another 512-bit RSA key, then it was born with a bullseye on its chest.
-- Lucky Green <mailto:shamrock@netcom.com> PGP encrypted mail preferred.
Note: Please respond via e-mail as well as or instead of posting, as I get CP-LITE instead of the whole list. --John Kelsey, jmkelsey@delphi.com / kelsey@counterpane.com PGP 2.6 fingerprint = 4FE2 F421 100F BB0A 03D1 FE06 A435 7E36 -----BEGIN PGP SIGNATURE----- Version: 2.6.2 iQCVAwUBMQfoEEHx57Ag8goBAQH7mQQAwHZ5ZH++AbVGER88rtRbgiu+syYNI9AI bwgeUT3gYpf1kqRksg5dLluAabEo+OSorzb5x/WrF1bemkqr3Y+GtEhh8HfSGaZG pmAe1hwSyGLQImqonZ/MxYz17eOK2Win9VBt1o+0jQCceUN8pc/QXRZvEAzjdkS4 lKijlYa/XYE= =Ii7i -----END PGP SIGNATURE-----
[Birthday paradoxing to get test for non-random padding]
simply generating a few thousand messages (maybe six or seven thousand, to be safe), and seeing whether or not we ever get a duplicate LEAF. We expect to, after about 2^12 encryptions, if
If you were to try this, you'ld probably want to try around 12,000 to reach the 95% confidence interval. However, I seriously doubt that this is going to be the case; they're using BSAFE, which does random padding to PCKS1 in just about all it's RSA modes. The only people Lotus could hire to get it that wrong probably have too much tied up in options to be easily head-hunted. Simon
participants (2)
-
JMKELSEY@delphi.com -
Simon Spero