Managing concurrent testing of data race with ComRaDe

Abstract

As a result of the increasing number of concurrent programs, the researchers put forward a number of tools with different implementation strategies to detect data race. However, confirming data races from the collection of true and false positives reported by race detectors is extremely the time-consuming process during the evaluation period. In this paper, we presented ComRaDe, a management platform for concurrent testing of data race with three main functions: manage and filter data races, run evaluation programs to select race detectors, generate detection report automatically. We integrated and compared three different race detectors on ComRaDe in terms of race detection capability. The results demonstrated the potential of ComRaDe on effectively identifying the advantages and limitations of different race detectors, and in further helping researchers to select and improve the capability of detectors for its convenience.