2016/11/21 by Sangho Lee, Lee, Sangho, Ming-Wei Shih +9 · 1 voice · 18 citations
Computer Science · Materials Science · #Advanced Malware Detection Techniques #Diamond and Carbon-based Materials Research #Physical Unclonable Functions (PUFs) and Hardware Security #Security and Verification in Computing #cs.CR
paper · pdf · doi:10.48550/arxiv.1611.06952
openalex publication_date 2016/11/21 · openalex created_date 2019/06/27 · openalex updated_date 2026/07/28
In this paper, we explore a new, yet critical, side-channel attack against\nIntel Software Guard Extension (SGX), called a branch shadowing attack, which\ncan reveal fine-grained control flows (i.e., each branch) of an enclave program\nrunning on real SGX hardware. The root cause of this attack is that Intel SGX\ndoes not clear the branch history when switching from enclave mode to\nnon-enclave mode, leaving the fine-grained traces to the outside world through\na branch-prediction side channel. However, exploiting the channel is not so\nstraightforward in practice because 1) measuring branch\nprediction/misprediction penalties based on timing is too inaccurate to\ndistinguish fine-grained control-flow changes and 2) it requires sophisticated\ncontrol over the enclave execution to force its execution to the interesting\ncode blocks. To overcome these challenges, we developed two novel exploitation\ntechniques: 1) Intel PT- and LBR-based history-inferring techniques and 2)\nAPIC-based technique to control the execution of enclave programs in a\nfine-grained manner. As a result, we could demonstrate our attack by breaking\nrecent security constructs, including ORAM schemes, Sanctum, SGX-Shield, and\nT-SGX. Not limiting our work to the attack itself, we thoroughly studied the\nfeasibility of hardware-based solutions (e.g., branch history clearing) and\nalso proposed a software-based countermeasure, called Zigzagger, to mitigate\nthe branch shadowing attack in practice.\n