|COOKIES: By using this website you agree that we can place Google Analytics Cookies on your device for performance monitoring.|
Wedge: Splitting Applications into Reduced-Privilege Compartments
If you have a question about this talk, please contact Eiko Yoneki.
Most applications today run as single processes, allowing successful attackers to access all of the process’s memory and sensitive data. We intend to reverse this situation by splitting applications into multiple compartments that hold no privileges by default, and allowing programmers to explicitly grant privileges and memory permissions, therefore controlling the damage of potential exploits. Our system Wedge is composed of two synergistic parts: the sthread OS primitives that allow programmers to create default-deny compartments with explicitly set privileges, and Crowbar, a tool that run-time analyzes existing applications to help identify potential sthreads along with their required memory and file descriptor permissions, allowing a simpler migration of existing code to sthreads. We applied sthreads to SSL -enabled Apache protecting the privacy of user data even against a powerful attacker can both exploit large part of the server and also act as a man-in-the-middle in the network; all at a 20—40% performance cost. Finally we describe a userland implementation of sthreads that does not sacrifice performance thanks to the careful (ab)use of UNIX AP Is.
Bio: Andrea Bittau is a PhD student at UCL working on operating system support for application security, supervised by Mark Handley and Brad Karp. His past projects include the fragmentation attack for 802.11 WEP networks, where an attacker can spoof and eavesdrop data without needing the WEP key, and developing the first open source Bluetooth sniffer, based on GNU radio.
This talk is part of the Computer Laboratory Systems Research Group Seminar series.
This talk is included in these lists:
Note that ex-directory lists are not shown.
Other listsCamtessential Group CRASSH events CUUEG talks
Other talksConvex and non-convex worlds in machine learning Branching Brownian Motion in a Bounded Domain Members' Slides Transport Scotland [POSTPONED; DATE TBC] Development & Plasticity of the Brain Connecting Biology to Mathematics by way of Symbolic Computing