[Nets-seminars] Fwd: Nets seminar: Petr Marchenko, 4 PM TODAY (7th Oct), MPEB 1.20

[Brad Karp]

Final reminder: Petr Marchenko will give a talk at 4 PM TODAY in MPEB  
1.20 on his recent systems security work.

It has come to my attention that many people in the Nets group aren't  
on the "nets-sse-seminars" mailing list, which is meant to receive  
announcements of all Nets seminars *and* all SSE seminars. I've sent  
this email to the "nets" and "nets-seminars" lists, in an attempt to  
make sure everyone in the Nets group receives it.

See you there,
-Brad, bkarp at cs.ucl.ac.uk

Begin forwarded message:

> From: Brad Karp <B.Karp at cs.ucl.ac.uk>
> Date: October 7, 2009 1:05:36 AM BST
> To: nets-sse-seminars at cs.ucl.ac.uk
> Cc: Brad Karp <B.Karp at cs.ucl.ac.uk>
> Subject: Nets seminar: Petr Marchenko, 4 PM TODAY (7th Oct), MPEB 1.20
>
> We kick off the fall series of Nets seminars today, the 7th of  
> October at 4 PM in MPEB 1.20.
>
> Petr Marchenko will give a talk on his recent work on design rules  
> for building exploit-resistant implementations of crypto protocols  
> for networked applications.
>
> (Apologies for the short notice--in the crush of start-of-term, we  
> had no reply to the room booking request until after 5 PM on Tuesday.)
>
> Title and abstract follow.
>
> See you all there,
> -Brad, bkarp at cs.ucl.ac.uk
>
> ---
>
> Speaker: Petr Marchenko
>
> Time and location: 4 PM, 7th October, MPEB 1.20
>
> Title:
>
> Extending End-to-End Protection of Sensitive Data in Network  
> Protocols to Vulnerable Software
>
> Abstract:
>
> Today's widely used networked applications routinely process users'  
> sensitive data, such as credit card numbers and social security  
> numbers. They protect such sensitive data against disclosure,  
> corruption, and replay within the network using cryptographic  
> protocols. To provide true end-to-end security, however,  
> applications must be robust against exploit-based attacks on such  
> protocols' *implementations* at the end systems. Recent advances  
> such as privilege separation and Distributed Information Flow  
> Control, while useful building blocks for limiting the harm exploits  
> of networked applications can cause, do not inherently protect  
> cryptographic protocol implementations from these attacks. We  
> demonstrate that neither technique as used to date protects against  
> either of two general attacks: an *oracle attack* and *active man in  
> the middle attack*. These attacks target a cryptographic protocol's  
> specific design and implementation. In this work, we show that  
> mechanisms within a cryptographic protocol that defend against in- 
> network attacks imply mechanisms that should be deployed within the  
> protocol's end-system implementation, to defend against analogous  
> exploit-based attacks on the protocol's implementation. We present  
> simple, general rules for privilege-separating cryptographic  
> protocol implementations that defend against these attacks. To  
> demonstrate these rules' practical use, even for pre-existing code  
> bases, we apply them to OpenSSH and OpenSSL, yielding  
> implementations of these applications' cryptographic protocols that  
> are the first to be robust against oracle and active man in the  
> middle attacks.
>