On enhancing the debug architecture of a system-on-chip (SoC) to detect software attacks

Abstract

The prevalent use of systems-on-chip (SoCs) makes them prime targets for software attacks. Proposed security countermeasures monitor software execution in real-time, but are impractical, and require impractical changes to the internal logic of intellectual property (IP) cores. We leverage the software observability provided by the readily available SoC debug architecture to detect attacks without modifying IP cores. We add hardware components to configure the debug architecture for security monitoring, to store a golden software execution model, and to notify a trusted kernel process when an attack is detected. Our evaluations show that the additions do not impact runtime software execution, and incur 9% area and power overheads on a low-cost processor core.