A Final Return for OpenBSD Anti-Return-Oriented Programming Mitigations

Return-Oriented Programming (ROP) continues to be a serious attack taking advantage of flaws in memory unsafe languages, particularly buffer overflows, to launch arbitrary code execution attacks by chaining together pieces of already existing code in loaded binaries and shared libraries, called gadgets. With the continued reliance on x86_64 CPUs in cloud and personal servers, mitigations that can meaningfully reduce the success of ROP attacks without significant overhead continue to be attractive. We propose the porting of one such software-based anti-ROP mitigation proposed by OpenBSD: compile-time instruction rewriting to avoid opportunities for ROP exploitation. We bring this mitigation, originally developed for the custom OpenBSD implementation of the LLVM compiler suite, to GCC by way of a standalone utility that sits in between the compiler and the assembler and rewrites potential gadget instructions before assembly into object code. Our utility provides a minimal reduction in gadgets with some penalties in binary sizes and performance impacts. We compare our GCC-ported standalone utility to the original OpenBSD LLVM mitigation and discovered that our standalone utility is weaker compared to the original LLVM-based mitigation. However, due to the overall weak reduction in gadgets for both the LLVM-based and GCC-based implementations, we conclude that seemingly obvious mitigations may prove to be anything but, and caution providing security improvements without significant testing and evaluation.