Operating System Services for Recovering Errant Applications

Conference contribution, 2008

Operating systems often manage critical infrastructures where failures can have serious consequences. This raises great concerns about their robustness. From the user perspective, it is the service delivered by host applications that needs to be dependable. Operating systems should therefore provide comprehensive error detection and recovery services to those applications, so that the system as a whole can be dependable and secure. This paper addresses the recovery flow that takes place after an application error is detected. The goal is to combine existing techniques into a set of operating system services that support application recovery both from software and hardware errors. We describe a prototype system where these services are currently being implemented and outline how we intend to experimentally evaluate them.