[saag] The real Heartbleed problem

Phillip Hallam-Baker <hallam@gmail.com> Fri, 11 April 2014 12:21 UTC

MIME-Version: 1.0
Date: Fri, 11 Apr 2014 08:21:09 -0400
Message-ID: <CAMm+Lwi0th8yBk8bW8y3z7t4nY+1unHgMhrzP4wx_KgZYPXyQQ@mail.gmail.com>
From: Phillip Hallam-Baker <hallam@gmail.com>
To: "saag@ietf.org" <saag@ietf.org>
Content-Type: text/plain; charset="UTF-8"
Archived-At: http://mailarchive.ietf.org/arch/msg/saag/qjsw3Xlpq2_H9MtVFjgBFyg7eBA
Subject: [saag] The real Heartbleed problem
Precedence: list

Right now we have two billion users being told to 'change their
passwords' and it will do them absolutely no good because virtually
none of the servers they are contacting with their new password will
have been updated.

The fundamental problem here is that we are using disclosure of bearer
tokens as authentication. Passwords and cookies are both terrible ways
to authenticate users.


There are two parts to this problem, one of them is to get a password
validation mechanism into HTML that does not disclose the password to
the other side. The other is how to gradually end use of cookies for
authentication.

The way to do that is to offer an extension to cookies that allows a
server to offer a challenge-response option to the client. If the
client supports this option it will only present a proof of knowledge
of the authentication secret and not the authentication secret itself.

Combined with appropriate defenses against replay attacks, and TLS
channel binding this can be made a very solid approach.


But we are not going to get good security if people are relying on
transport layer encryption to secure passwords.


https://datatracker.ietf.org/doc/draft-hallambaker-httpsession/


-- 
Website: http://hallambaker.com/

[saag] The real Heartbleed problem Phillip Hallam-Baker