|
2011-11-01
, 17:24
|
|
Posts: 361 |
Thanked: 219 times |
Joined on Sep 2010
|
#42
|
Hi, all!
Sorry for being silent so long. I am the maintainer of the maemo-security-certman package which should be updated to fix this problem.
...
Cheers, JuM
Apropos Rizzo's and Duong's new SSL/TLS exploit, may I recommend a summary by the Tor project:
https://blog.torproject.org/blog/tor...ast-ssl-attack
The Tor blog is an excellent source of security related news IMHO. TL;DR the attack is not really revolutionary but a very serious matter anyhow.
Still, it doesn't exactly mean that SSL /TLS <= 1.0 is totally broken. It means that by some considerable effort an attacker in a MITM position can decrypt such data that is added in every packet sent to a SSL/TLS site at a known position, i.e. a session cookie. And only if the cipher uses CBC (SSL 3.0 supports some 30 different ciphers, both block and stream ciphers, and only block ciphers can use CBC). For instance Google's servers use stream cipher RC4 which is not vulnerable to this particular attack, although it has its own set of problems. Many other sites too.
While these kinds of news are always disturbing, I personally think that what we are seeing is evolution in progress. A CA turns out to be untrustworthy => the CA gets wiped away. DigiNotar just filed for bankruptcy. A vulnerability is found in a crypto protocol => the protocol is fixed and while waiting for that, an alternative protocol is being used.
But then again, I'm an incurable optimist. Even thinking Maemo could still have some future ahead of her.
JuM
Last edited by juhanima; 2011-09-24 at 20:26.