Defending against Return-Oriented Programming

Return-oriented programming (ROP) has become the primary exploitation technique for system compromise in the presence of non-executable page protections. ROP exploits are facilitated mainly by the lack of complete address space randomization coverage or the presence of memory disclosure vulnerabilities, necessitating additional ROP-specific mitigations. Existing defenses against ROP exploits either require source code or symbolic debugging information, or impose a significant runtime overhead, which limits their applicability for the protection of third-party applications. We propose two novel techniques to prevent ROP exploits on third-party applications without requiring their source code or debug symbols, while at the same time incurring a minimal performance overhead. Their effectiveness is based on breaking an invariant of ROP attacks: knowledge of the code layout, and a common characteristic: unrestricted use of indirect branches. When combined, they still retain their applicability and efficiency, while maximizing the protection coverage against ROP. The first technique, in-place code randomization, uses narrow-scope code transformations that can be applied statically, without changing the location of basic blocks, allowing the safe randomization of stripped binaries even with partial disassembly coverage. These transformations effectively eliminate 10%, and probabilistically break 80% of the useful instruction sequences found in a large set of PE files. Since no additional code is inserted, in-place code randomization does not incur any measurable runtime overhead, enabling it to be easily used in tandem with existing exploit mitigations such as address space layout randomization. Our evaluation using publicly available ROP exploits and two ROP code generation toolkits demonstrates that our technique prevents the exploitation of the tested vulnerable Windows 7 applications, including Adobe Reader, as well as the automated construction of alternative ROP payloads that aim to circumvent in-place code randomization using solely any remaining unaffected instruction sequences. The second technique is based on the detection of abnormal control transfers that take place during ROP code execution. This is achieved using hardware features of commodity processors, which incur negligible runtime overhead and allow for completely transparent operation without requiring any modifications to the protected applications. Our implementation for Windows 7, named kBouncer, can be selectively enabled for installed programs in the same fashion as user-friendly mitigation toolkits like Microsoft's EMET. The results of our evaluation demonstrate that kBouncer has low runtime overhead of up to 4%, when stressed with specially crafted workloads that continuously trigger its core detection component, while it has negligible overhead for actual user applications. In our experiments with in-the-wild ROP exploits, kBouncer successfully protected all tested applications, including Internet Explorer, Adobe Flash Player, and Adobe Reader. In addition, we introduce a technique that enables ASLR for executables with stripped relocation information by incrementally adjusting stale absolute addresses at runtime. The technique relies on runtime monitoring of memory accesses and control flow transfers to the original location of a module using page table manipulation. We have implemented a prototype of the proposed technique for Windows 8, which is transparently applicable to third-party stripped binaries. Our results demonstrate that incremental runtime relocation patching is practical, incurs a runtime overhead of up to 83% in most of the cases for initial runs of protected programs, and has a low runtime overhead of 5% on subsequent runs