Finding and exploiting faults in hardware and software

Murdock, Kit (2023). Finding and exploiting faults in hardware and software. University of Birmingham. Ph.D.

Preview

Murdock2023PhD.pdf
Text
Available under License All rights reserved.
Download (6MB) | Preview

Abstract

Computers are constantly being enhanced to improve their speed, size, security, and energy consumption. Dynamic Voltage and Frequency Scaling (DVFS) improves energy efficiency by enabling a processor to upscale its power as needed, thus using little energy when idle. And, more recently, hardware-based trusted execution environments such as Software Guard Extensions (SGX) have been created with the promise of securely executing sensitive processes—thus protecting the data and running computations from a root adversary.

In the first part of this thesis, we show how the attempt to make computers more efficient by dynamically responding to their energy needs has created a new attack surface. Specifically, we are able to retrieve keys from both an AES and a RSA cryptographic process running inside an SGX enclave by lowering the operating voltage. We further investigate the undervolting effect and are able to improve the attack to create an out-of-bounds under/overflow.

Meanwhile, fault injection attacks (such as our software undervolting one) represent a major threat to Internet-of-Things and embedded devices. As of today, evaluating to what extent a device is susceptible to fault injection is a mostly manual process, requiring significant expert knowledge and often expensive, complex lab equipment. In addition, even if a fault can be induced, it is often unclear which effect caused the incorrect output. In the second part of this thesis, we address this difficulty by designing and building a performant, exhaustive fault injection tool. We compare our software with three others and demonstrate it out-performs on features and speed.

Type of Work:

Thesis (Doctorates > Ph.D.)

Award Type:

Doctorates > Ph.D.

Supervisor(s):