Hardening Approach for the Scheduler’s Kernel Data Structures

Abstract:
Guaranteeing the correct system behaviour in safety-critical computer-based systems is a challenging issue due to transient faults. This is particularly true considering the effects of faults in memory elements containing data structures belonging to a Unix-based Operating System. Elements involved in processes synchronisation management have been demonstrated to be critical, thus the need for redundancy mechanisms. A solution is proposed in order to overcome this vulnerability, by implementing an information multiplication technique in the core data structures used by the scheduler currently running in Linux systems, aiming to detect and correct every possible fault affecting these elements. The theoretical analysis is presented, aiming for a fault injection campaign to be conducted to evaluate the effectiveness of this approach.