Automating Seccomp Filter Generation for Linux Applications
This work provides a faster and more broadly applicable method for generating seccomp filters, improving the security posture of Linux applications by reducing their attack surface for developers and system administrators.
This paper introduces Chestnut, an automated system for generating strict syscall filters for Linux applications. Chestnut achieves up to 73x faster filter generation with its compiler-based approach and blocks an average of 302 syscalls (86.5%) via the compiler and 288 syscalls (82.5%) via binary analysis across 18 applications. It also prevents exploitation of over 62% of 175 kernel-targeting CVEs.
Software vulnerabilities in applications undermine the security of applications. By blocking unused functionality, the impact of potential exploits can be reduced. While seccomp provides a solution for filtering syscalls, it requires manual implementation of filter rules for each individual application. Recent work has investigated automated approaches for detecting and installing the necessary filter rules. However, as we show, these approaches make assumptions that are not necessary or require overly time-consuming analysis. In this paper, we propose Chestnut, an automated approach for generating strict syscall filters for Linux userspace applications with lower requirements and limitations. Chestnut comprises two phases, with the first phase consisting of two static components, i.e., a compiler and a binary analyzer, that extract the used syscalls during compilation or in an analysis of the binary. The compiler-based approach of Chestnut is up to factor 73 faster than previous approaches without affecting the accuracy adversely. On the binary analysis level, we demonstrate that the requirement of position-independent binaries of related work is not needed, enlarging the set of applications for which Chestnut is usable. In an optional second phase, Chestnut provides a dynamic refinement tool that allows restricting the set of allowed syscalls further. We demonstrate that Chestnut on average blocks 302 syscalls (86.5%) via the compiler and 288 (82.5%) using the binary-level analysis on a set of 18 widely used applications. We found that Chestnut blocks the dangerous exec syscall in 50% and 77.7% of the tested applications using the compiler- and binary-based approach, respectively. For the tested applications, Chestnut prevents exploitation of more than 62% of the 175 CVEs that target the kernel via syscalls. Finally, we perform a 6 month long-term study of a sandboxed Nginx server.