Security Assurance of Heterogeneous Software Systems

Modern software systems increasingly use specialized hardware such as Graphics Processing Units (GPUs) and other types of hardware accelerators to run more efficiently. The use of this specialized hardware opens up new cyber-attack vectors and creates opportunities for hiding malware from existing tools for malware analysis and detection. We propose to design and build an automated analysis tool that detects both unintentional vulnerabilities and explicitly introduced malicious code in software written for such heterogeneous hardware systems. The tool will combine both static and dynamic analysis components, which will work together to enable effective security analysis of heterogeneous software. The tool will operate at the machine-code level and will leverage and improve the extensive existing GrammaTech infrastructure for machine-code analysis.