The goal of this study is to gauge how difficult it is to bypass the protections offered by EMET, a popular Microsoft zero-day prevention capability. We initially focused on just the ROP protections, but later expanded the study to include a real world example. We were able to bypass EMET’s protections in example code and with a real world browser exploit. The primary novel elements in our research are: 1.Deep study regarding the ROP protections, using example applications to show how to bypass each of the five ROP checks in a generic manner. 2.Detailed real world example showing how to defeat all relevant protections. Look for a new technique to bypass the stack pivot protection, shellcode complete with an EAF bypass, and more. These bypasses leverage generic limitations, and not easily repaired. The impact of this study shows that technologies that operate on the same plane of execution as potentially malicious code, offer little lasting protection. This is true of EMET and other similar userland protections.
Jared DeMott is a security researcher for Bromium, Inc. He has spoken at security conferences such as DerbyCon, Black Hat, Defcon, ToorCon, Shakacon, DakotaCon, and GRRCon. He is active in the security community by teaching his Application Security course, and has co-authored a book on Fuzzing. Mr. DeMott has been an invited lecturer at prestigious institutions such as the United States Military Academy, and previously worked for the National Security Agency. DeMott holds a PhD from Michigan State University.