Sushant Dinesh (University of Illinois at Urbana Champaign), Grant Garrett-Grossman (University of Illinois at Urbana Champaign), Christopher W. Fletcher (University of Illinois at Urbana Champaign)

Recent attacks have demonstrated that modern microarchitectures are fraught with microarchitectural side channels. Constant-time (CT) programming is a software development methodology where programs are carefully written to avoid these channels. In a nutshell, the idea is to only pass secret data to safe instructions, i.e., those whose execution creates operand-independent hardware resource usage.

Yet, current CT programming practices have significant security and performance issues. CT code is written and compiled once, but may execute on multiple different microarchitectures. Yet, what instructions are safe vs. unsafe is fundamentally a microarchitecture-specific issue. A new microarchitectural optimization (or vulnerability) may change the set of safe instructions and break CT guarantees.

In this work, we develop SynthCT to address the above issues. Given a specification of safe/unsafe instructions, SynthCT automatically synthesizes translations for all unsafe instructions in the ISA using only instructions from the safe set. The synthesized translations can be used as a part of a late-stage compiler pass to generate hardened binaries for a specific microarchitecture. This closes the security hole as the specification, and hence the safe translations, can target each microarchitecture individually. This also allows CT code to reclaim some performance, e.g., use more complex/higher-performing instructions, when they are deemed safe for a specific microarchitecture.

Using the techniques we develop in SynthCT, we are able to synthesize translations for a majority of the x86 64 ISA. Specifically, SynthCT is able to generate safe translations for 75% of the ISA using only the remaining 25% of the ISA. Interestingly, the majority of the instructions that SynthCT was unable to generate translations for are instructions that experts believe are safe instructions on today’s x86 64 microarchitectures.

View More Papers

Demo #9: Dynamic Time Warping as a Tool for...

Mars Rayno (Colorado State University) and Jeremy Daily (Colorado State University)

Read More

Packet-Level Open-World App Fingerprinting on Wireless Traffic

Jianfeng Li (The Hong Kong Polytechnic University), Shuohan Wu (The Hong Kong Polytechnic University), Hao Zhou (The Hong Kong Polytechnic University), Xiapu Luo (The Hong Kong Polytechnic University), Ting Wang (Penn State), Yangyang Liu (The Hong Kong Polytechnic University), Xiaobo Ma (Xi'an Jiaotong University)

Read More

DeepSight: Mitigating Backdoor Attacks in Federated Learning Through Deep...

Phillip Rieger (Technical University of Darmstadt), Thien Duc Nguyen (Technical University of Darmstadt), Markus Miettinen (Technical University of Darmstadt), Ahmad-Reza Sadeghi (Technical University of Darmstadt)

Read More

Get a Model! Model Hijacking Attack Against Machine Learning...

Ahmed Salem (CISPA Helmholtz Center for Information Security), Michael Backes (CISPA Helmholtz Center for Information Security), Yang Zhang (CISPA Helmholtz Center for Information Security)

Read More