In this paper, we investigate an architectural-level mitigation technique based on the coordinated action of multiple checksum codes, to detect and correct errors at run-time.