Over the past few years, numerous real-time and embedded systems have been adopting multi-core architectures for either better performances, or energy efficiency. For the case of real-time applications, where tasks can have critical deadlines, it is desirable to ensure the schedulability of the application statically, taking into account the possible software and hardware failures. While a lot of effort have been made to handle software misbehaviours, resilience to hardware failures has often been overlooked. In this paper, we propose to study the schedulability of multi-core applications. Specifically, we want to check statically whether or not a real-time system will be able to meet the deadlines of its most critical tasks, even when one or more of its cores are offline. In order to achieve this goal, we translate the schedulability problem into a state space exploration, using Data Decision Diagrams to support the computation and analysis of such state space.
展开▼
