Understanding the Propagation of Hard Errors to Software and its Implications for Resilient System Design

With continued CMOS scaling, future shipped hardware will be increasingly vulnerable to in-the-field faults. To be broadly deployable, the hardware reliability solution must incur low overheads, precluding use of expensive redundancy. We explore a co-designed hardware-software solution that treats most hardware faults as software bugs and leverages common mechanisms for hardware and software reliability, thereby amortizing some of the overhead. Fundamental to such a solution is a characterization of how hardware faults in different microarchitectural structures of a modern processor propagate through the application and OS. This paper aims to provide such a characterization, identify low-cost detection methods to intercept fault propagation, and to provide guidelines for a complete co-designed reliability solution. We focus on hard faults because they are increasingly important and have different system implications than the much studied transients. We achieve our goals through fault injection experiments with a microarchitecture level full system timing simulator