not-quite-so-broken TLS: lessons in re-engineering a security protocol specification and implementation

Add to your list(s) Download to your calendar using vCal

• Hannes Mehnert (University of Cambridge)
• Thursday 23 June 2016, 11:00-12:00
• Sloman Lounge, School of Computer Science.

If you have a question about this talk, please contact Andreea Radu.

Lots of TLS implementations have a history of security vulnerabilities (goto fail, Heartbleed, ...). The root causes of most existing TLS implementations are ambiguous specification, choice of inherently unsafe programming language, complex APIs, and legacy code bases.

We re-engineered a TLS stack from scratch which avoids these root causes by using a more modern typed programming language with memory management and a modular and declarative programming style, where effectful operations (input/output, but also mutation of memory) are explicitly annotated. The result is an artifact which can be used both as specification, testing other implementations, and as implementation. Its performance is in the same ballpark as OpenSSL, while its source code is nearly two orders of magnitude smaller. Useful features are implemented, while we avoided to implement legacy (SSL version 2 and 3, export ciphers, ...).

Our implementation is used in the wild, one target is MirageOS, a library operating system (running OCaml directly as virtual machine, without any libc).

Project website: https://nqsb.io

This talk is part of the Computer Security Seminars series.

This talk is included in these lists:

• Computer Science Departmental Series
• Computer Science Distinguished Seminars
• Computer Security Seminars
• Sloman Lounge, School of Computer Science
• computer sience

Note that ex-directory lists are not shown.