Reproduce “Empirical Security Analysis of Software-based Fault Isolation through Controlled Fault Injection” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Nils Bars, Lukas Bernhard, Moritz Schloegel, Thorsten Holz
SbxBrk, a fuzzer developed by researchers from CISPA, MPI-SP, and Arizona State, empirically tests Software-based Fault Isolation (SFI) by injecting faults at the boundary between trusted and untrusted code within a single process. It successfully uncovered 19 previously unknown security vulnerabilities, including heap out-of-bounds writes and double-fetch issues, in Google's widely used V8 heap sandbox.
Key insights
- Rigorous testing of SFI requires a specialized approach that models an attacker already inside the sandbox, as traditional fuzzing misses critical vulnerabilities at the isolation boundary.
- Compiler optimizations and specific code generation choices can dynamically alter the "fault domain boundary," influencing whether a vulnerability manifests.
- Logic bugs, such as double-fetches or mishandling of corrupted enum variants across the trust boundary, are a prevalent and critical class of SFI bypasses, often not addressed by memory-safe languages alone.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.