Towards a test automation framework for data race testing

E. A. Gerlits, V. S. Mutilin

Ivannikov Institute for System Programming of the RAS

Abstract: In this paper, we examine shared memory concurrent programs and errors occurring in them, specifically data races. We design a test automation framework to develop data race revealing testing scenarios by analogy to test automation frameworks for sequential programs. The main problem complicating development of testing scenarios is the nondeterministic nature of multithreaded program executions. To provide repeatable testing scenarios we define a notion of synchronization points in the source code of the computer program where a test regulates execution of parallel threads with synchronization actions. Tests being developed with our test automation framework can be executed automatically and can be used for regression testing.

Keywords: data race; race condition; data race reproducing; data race testing; multithreaded program; shared memory concurrent program; testing framework; test automation framework; thread synchronization

DOI: 10.15514/ISPRAS-2025-37(1)-6