Micro-virtualization memory tracing to detect and prevent spraying attacks
Stefano Cristalli, Mattia Pagnozzi, Mariano Graziano, Andrea Lanzi, Davide Balzarotti · USENIX Security Symposium · 2016
Spraying is a common payload delivery technique used by attackers to execute arbitrary code in presence of Address Space Layout Randomisation (ASLR). In this paper we present Graffiti, an efficient hypervisorbased memory analysis framework for the detection and prevention of spraying attacks. Compared with previous solutions, our system is the first to offer an efficient, complete, extensible, and OS independent protection against all spraying techniques known to date. We developed a prototype open source framework based on our approach, and we thoroughly evaluated it against all known variations of spraying attacks on two operating systems: Linux and Microsoft Windows. Our tool can be applied out of the box to protect any application, and its overhead can be tuned according to the application behavior and to the desired level of protection.