Safety supervision layer

This work covers a generic approach to fault detection for operating systems in fail-safe environments. A safety supervision layer between the application layer and the operating system interface is discussed. It is an attempt to detect operating system and hardware faults in an end-to-end way. Standard POSIX system calls are wrapped by procedures that provide fault detection features. Furthermore, potentials of an additional watchdog module on top of the operating system interface are analyzed. Applications that use the Safety Supervision Layer are notified of detected faults and deal with them by providing specific handlers to bring the fail-safe system to its safe state. The goal of the presented layer is to encapsulate the operating system and hardware layers a safety-critical application resides on, in order to detect faults produced by those and bring the system to a safe state. Advantages of such an attempt are portability, lower time-to-market, higher cost efficiency in building fail-safe systems and - most important - reduced error detection latency compared to usual periodic supervision approaches.