Automated CFI Policy Assessment with Reckon

摘要: Protecting programs against control-flow hijacking attacks recently has become an arms race between defenders and attackers. While certain defenses, e.g., \textit{Control Flow Integrity} (CFI), restrict the targets of indirect transfers through static dynamic analysis, attackers could search program for available gadgets that fall into legitimate target sets to bypass defenses. There are several tools helping both in developing exploits analysts strengthening their Yet, these fail adequately (1) model deployed (2) compare them a head-to-head way, (3) use semantic information help craft attack countermeasures. Control Integrity (CFI) proved be one promising defenses control flow hijacks tons efforts have been made improve CFI various ways past decade. However, there is lack systematic assessment existing In this paper, we present Reckon, source code analysis tool assessing state-of-the-art by first precisely modeling then evaluating unified framework. Reckon helps determine level security offered different find usable even after were applied, thus providing important step towards successful stronger We used assess eight on real-world such as Google's Chrome Apache Httpd. provides precise measurements residual surfaces, accordingly ranks policies each other. It also successfully paves way construct reuse eliminate remaining surface, disclosing calltargets under most restrictive