Diagnosis and correction of multiple design errors using critical path tracing and mutation analysis

Abstract

Identification of the presence of design errors, i.e. verification is a well-studied field with a range of methods developed. Yet, most of the verification cycle is consumed for debugging, which consists of localization and correction of errors. Current paper presents a method for automated debug of multiple simultaneous design errors for RTL designs. We propose a critical path tracing based error localization method, which performs statistical analysis in order to rank suspected error locations. Then, an error matching approach to correction is applied implementing mutation operations. Experiments carried out in this work analyze localizing multiple erroneous data operations and their mutation-based correction. We compare two metrics of statistical analysis and show their capabilities in localizing multiple errors.