K-Miner: Uncovering Memory Corruption in Linux

Operating system kernels are appealing attack targets: compromising the kernel usually allows attackers to bypass all deployed security mechanisms and take control over the entire system. Commodity kernels, like Linux, are written in low-level programming languages that offer only limited type and memory-safety guarantees, enabling adversaries to launch sophisticated run-time attacks against the kernel by exploiting memory-corruption vulnerabilities. Many defenses have been proposed to protect operating systems at run time, such as control-flow integrity (CFI). However, the goal of these run-time monitors is to prevent exploitation as a symptom of memory corruption, rather than eliminating the underlying root cause, i.e., bugs in the kernel code. While finding bugs can be automated, e.g., using static analysis, all existing approaches are limited to local, intra-procedural checks, and face severe scalability challenges due to the large kernel code base. Consequently, there currently exist no tools for conducting global static analysis of operating system kernels. In this paper, we present K-Miner, a new framework to efficiently analyze large, commodity operating system kernels like Linux. Our novel approach exploits the highly standardized interface structure of the kernel code to enable scalable pointer analysis and conduct global, context-sensitive analysis. Through our inter-procedural analysis we show that K-Miner systematically and reliably uncovers several different classes of memory-corruption vulnerabilities, such as dangling pointers, user-after-free, double-free, and double-lock vulnerabilities. We thoroughly evaluate our extensible analysis framework, which leverages the popular and widely used LLVM compiler suite, for the current Linux kernel and demonstrate its effectiveness by reporting several memory-corruption vulnerabilities.