Safety features of SoCs: How can they be re-used?

Summary form only given: Safety is a concept which nowadays is pervading a large spectrum of applications fields. It was pioneered by military and aerospace and is now exploited in a wide spectrum of systems to prevent any form of injury to people to properties and to the environment. In other words, the purpose of safety is to prevent that any type of error generate unacceptable risk conditions. This talk will explore two specific aspects linked to safety. First, an analysis of error sources (limited to hardware and environmentally-dependant errors) will be taken. Secondly, a discussion on the detection and the diagnosis techniques of error root causes will be proposed, to follow-up the question asked by the keynote title. How and when is it possible to exploit differently the error detection and diagnosis features present in SoCs? Could they be re-used during off-line operations? Can we log and access the data they generate on-line? Is this a good chance to learn better about error root causes and activating better (=in a tailored manner) prevention mechanisms? Which are the key enablers to succeed?