Efficient and Safe Control Flow Recovery Using a Restricted Intermediate Language

摘要: Approaches for the automatic analysis of security policies on source code level cannot trivially be applied to binaries. This is due lacking high-level semantics low-level object code, and fundamental problem that control-flow recovery from binaries difficult. We present a novel approach recover both safe efficient. The key idea our use information contained in mechanisms approximate targets computed branches. To achieve this, we first define restricted control transition intermediate language (RCTIL), which restricts number possible each branch finite given targets. Based this language, demonstrate how model flow can recovered without data-flow analyses. Our evaluation shows makes solution more efficient than existing solutions.