Abstract
Critical systems widely depend on operating systems to perform their mission. Device drivers are a critical and defect-prone part of operating systems. Software defects in device drivers often cause corruption of data that may lead to data losses, that are a significant source of costs for large enterprise systems. This paper describes an ongoing research that aims at mitigating the impact of data corruption due to device driver defects on the availability and the integrity of data. We discuss a methodology for run-time detection and the tolerance of protocol violations in device drivers and then we present a preliminary activity that we are currently performing.