Re: MD5CRK is now LIVE

From: Jean-Luc Cooke (jlcooke_at_engsoc.org)
Date: 03/08/04 

Date: 8 Mar 2004 21:42:23 GMT

Francois Grieu <fgrieu@micronet.fr> wrote:
> Yes. Your latest post acted like an eye-opener to me.
> I mean it. My reasoning on why SSL certificates would be
> out of reach of MD5CRK, or simimar, was incorrect.

> As you show, SSL certificates are vulnerable to certain
> form of message collision attacks. I now have to dive into
> the details of the format of SSL certs to try determine
> if they are vulnerable, or not, to Paul van Oorschot and
> Michael Wiener's attack, or simple extensions thereof.
> Until that's done, and maybe after, I'm no longer sure
> SSL certificates based on MD5 are safe.

> BTW, unless I am badly mistaken in the other direction,
> Paul van Oorschot and Michael Wiener's attack can be
> extended, with only a doubling of the effort, to find
> MD5-colliding messages with entirely chosen and
> different beginning, provided
> - the messages are of the same length
> - one is willing to tolerate about 128 bits of
> randomness in the last 512 bits block where the
> (padded) messages differ.

I would agree, since the 128bit state would be the same into the last
md5-compress function. Back in the day when I tried to "latch on" to
the NEO project, I proposed a contract from Homer Simpson to Bart
Simpson where Homer promised Bart $1, but Bart being a deceptivly
cleaver cryptographer would re-write the contract to some random 128 bit
value (odds are, a very large number).

Two things changed - 1) I saw that NEO wasn't going to be the privider I
wanted 2) explaining the contract and the collision attack and how it's
very different from RC5 attacks was too much for the average user to
take in when looking for a DC project.

> Again, my apologies; and thanks for showing me the
> light.

Glad I could help, you too helped me by forcing me to distill my
explaination. Thank you.

JLC 

--

Relevant Pages

  • Re: Did Mcshame wet start and murder hundreds?
    ... Poe wrote: ... thought your original jab at Bart and Peter was unfair. ... I do not think that McCain is the best choice for president this year, ... it isn't fair when they do that slippery "Muslim" attack on ...
    (alt.true-crime)
  • Re: Somthing else for the spelling police!!!
    ... I suggested you drop an entirely _separate_ issue, ... you're suggesting that I drop the issue of you lying about what I've ... serious harm or injury, and as a result of the attack, the attackee was ... didn't _have_ a contract, which is clearly untrue, ...
    (uk.people.gothic)
  • Re: Why vitzivanu lehadlik ner?
    ... >attack, it's merely an observation. ... this "deconstruction" thing seems to be closely ... >associated with the type of analysis / thought that Jacko is talking about. ... Thus a contract that's been "mesora'd" to a beit din - that ...
    (soc.culture.jewish.moderated)
  • Re: Somthing else for the spelling police!!!
    ... Dave, long words aren't meaningless just because you don't understand them. ... serious harm or injury, and as a result of the attack, the attackee was ... You're lying again. ... have a fixed term contract, but what you said is that you don't have a ...
    (uk.people.gothic)
  • Re: [fw-wiz] Legal Release for Security Work
    ... >Here is a sample PENETRATION TESTING CONTRACT. ... actions the provider is allowed to perform, including what machines can ... Department A within a company hires a penetration tester; the attack ...
    (Firewall-Wizards)