A comprehensive software-based self-test and self-repair method for statically scheduled superscalar processors
Mario Schölzel, Tobias Koal, Sebastian Muller, Stefan Scharoba, Stephanie Roder, H.T. Vierhaus · 2016
The integration of a diagnostic software-based self-test and a software-based self-repair method into a single statically scheduled superscalar processor is presented. The self-test method is used as start-up test in-the-field in order to detect and localize permanent faults in the processor. The determined fault state is handed over to a software-based self-repair program. This program adapts the user application, which should be executed on the core, to the current fault state. Thereby both methods are able to cope with multiple permanent faults in the core. For this purpose the self-test programs as well as the self-repair program can be adapted to the current fault state, too. The paper also describes the diagnostic capability of the self-test programs and the achieved fault coverage. The hardware and program memory overhead for the self-test and self-repair programs is reported, too.