Dynamic taint analysis for vulnerability exploits detection

Untrusted Data originating from network input and configuration files, causes many software security problems. Keeping track of the propagation of untrusted data in program runtime is the main idea of dynamic taint analysis for vulnerability exploits detection. In this method data from network user input and configuration files were labeled as taint. In virtue of data flow analysis we design taint propagating algorithm, and define several taint detection policies for security-critical function which used taint data in dangerous ways that could cause vulnerability exploit. A vulnerability exploit detection prototype system was implemented. In contrast to other taint analysis systems, our prototype system has higher accuracy and vulnerability exploits coverage and low workloads.