On Mon, 27 Mar 2006, Brian Eaton wrote:
Properly implemented and configured MAC can prevent the leakage of
confidential (i.e. sensitive personal) information to (unauthorized) web
sites.
You lost me here. How would you design a MAC policy that lets firefox
remember my password for a web site, but doesn't let arbitrary code
running via a buffer overflow get at that same password?
Let's call the object where the password is stored P, the subject
representing the site where P is used X, and the subject representing an
arbitrary evil site Y.
The (partial) mandatory policy is as follows:
1. X has "need to know" for P and is allowed to read it.
2. Y (or any other site) is not allowed to read P.
As soon as the browser process reads P, its (potential) ability to send
any data to Y is lost forever because information flow from P to Y is
prohibited by the mandatory policy. This is a dynamic variant of the
*-property of the Bell-LaPadula security model.
The design of a usable browser on the top of such a security mechanism
would be somewhat tricky because we would need a dedicated process (or
several processes) for every site and it would be almost impossible to
follow external links leading out of password protected sites because a
subverted process might use such a link to leak confidential data.
The solution of the latter problem might be to scrap the process sending
data containing the password to the server as soon as it sends the request
and replace it with a fresh "untainted" process handling (interpreting and
displaying) the server's response (of course, we'd have to trust the
server to never send the password back to the client, e.g to be 100% free
of XSS bugs...).
--Pavel Kankovsky aka Peak [ Boycott Microsoft--http://www.vcnet.com/bms ]
"Resistance is futile. Open your source code and prepare for assimilation."
_______________________________________________
Full-Disclosure - We believe in it.
Charter: http://lists.grok.org.uk/full-disclosure-charter.html
Hosted and sponsored by Secunia - http://secunia.com/
