We use UEFI and commodity PC manufacturer "firmware" as a use case for vulnerability discovery and exploit development powered by analytics. BIOS, UEFI, and embedded firmware are recent focus areas for vulnerability analysis and exploit research. There are great offensive-security presentations and research on ring < 0 rootkits, failed implementations of trusted computing concepts, and hardware-assisted exploitation.
This talk complements existing firmware research by applying data-science to UEFI code analysis. This does not attack the UEFI platform or secure boot implementations; it does consider UEFI applications, drivers, and associated environments as attack surface. Analytics of code-usage, features, pervasiveness, update frequency, and vulnerabilities will help determine viability of homogeneous exploit development for seemingly-heterogeneous environments. The talk will review data-science approaches to vulnerability discovery in UEFI code, demonstrate the scalability of UEFI exploitation, and explore the potential for persistence as well as similar fun exercises.
Video demo: https://www.youtube.com/watch?v=6yI-C1aBCiU