Abstract
Software Guard eXtensions (SGX) is a hardware-based technology that introduces unobservable portions of memory, called enclaves, that physically screens software components from system tampering. Enclaves can be used to run arbitrary programs (including malicious code), but their actual impact on digital forensics and incident response remains unknown. In our work, we propose a methodical study of what information can be retrieved from an SGX machine and how to use this information to infer the enclaves interfaces and structure layout.
We tested our techniques over a dataset of 45 SGX applications and we showed the practicality of our techniques in a real-product use-case and on two malware-enclaves.
•Discussion of the limitation of current forensic analysis in SGX-machines.•Study of which information can be extracted from a machine running SGX enclaves.•New techniques for SGX enclaves detection, identification, and analysis.