SSL is broken on FreeBSD

[Chad Perrin]

On Fri, Apr 01, 2011 at 07:45:11PM +0100, István wrote:
>
> cool, i decided I need everything what I have on windows or on J random
> operating system with firefox. I install the corresponding package
> which is broken and therefore, so I can't verify if somebody i doing a
> MITM while I am shopping on Amazon. Massive win!

You do not seem entirely clear on how the TLS/SSL trust model actually
works.  Having a list of trusted CAs does not guarantee that nobody is
executing a man in the middle attack.  All it does is let the CA tell you
whether or not it wants you to believe there is a man in the middle
attack going on.  The accuracy of such a belief depends entirely on the
motives of the CA in question.

In point of fact, the SSL/TLS PKI (the infrastructure for out-of-band
authentication of certificates via CAs) is basically just a vacant lot
scam, where someone assumes the color of authority by simple act of
declaration in order to charge people for something it is not actually in
the scammer's power to provide or deny -- in this case, verifiable proof
of authenticity.  A discussion of the problem in broad terms is here:

    The TLS/SSL Certifying Authority system is a scam
    http://blogs.techrepublic.com.com/security/?p=2550

A much more verifiable system would be something based on distributed
agreement, such as the Monkeysphere or Perspectives projects provide for
certificate verification.  A simplified discussion of the comparative
characteristics of Perspectives and the CA-based PKI is here:

    Perspectives: better than CAs?
    http://blogs.techrepublic.com.com/security/?p=571


> 
> I understand you do not care about usability. It is viable attitude in
> many opensource communities including FreeBSD.

Part of usability is ensuring that what you think you are accomplishing
actually bears some resemblance to what you are accomplishing.  If you
think that just trusting whoever some compiled list of CAs tells you to
trust to have your best interests at heart is accomplishing something
approaching a guarantee of security, the *actual* usability of the system
is near nil -- though the *apparent* usability of it might be very high
in the estimation of those who have not taken the time to consider the
implications of taking such an approach to cryptographic trust.


> 
> Thank you anyway. I am going to copy that file from Linux ;)

That may give you a heightened *feeling* of security, but is not likely
to greatly increase your *actual* security.  In fact, it may reduce it,
depending on the types of sites you visit and what data you send to them.
By default, software like Firefox will at least warn you when you are
connecting to a site using an encryption certificate you have not already
told it to trust one way or another.  If you just uncritically add all
the CAs in the world to a trusted list, all you are doing is turning off
those warnings.

-- 
Chad Perrin [ original content licensed OWL: http://owl.apotheon.org ]
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 196 bytes
Desc: not available
Url : http://lists.freebsd.org/pipermail/freebsd-security/attachments/20110401/fe9f331a/attachment.pgp