Detecting Data Races in Interrupt-Driven Programs based on Static Analysis and Dynamic Simulation

Interrupt-driven programs are often embedded in safety-critical systems to perform hardware/resource dependent data operation tasks, such as data acquisition, processing, and transformation. The interrupt programs and tasks may happen in parallel which in a result causes indeterminist concurrent problems at runtime. Data race is one of the most popular problems challenging researchers and practitioners. Various static analysis, software testing approaches have been proposed to detect data races in source code, testing, and even production run. However, static analysis may report too many false positives due to the lack of execution information. Dynamic testing may miss some important races since it could not generate adequate test cases to test all possible execution scenarios. In this paper, we propose a hybrid approach to detect data races in interrupt-driven programs based on static analysis and dynamic simulation. We implemented a prototype tool and conducted a controlled experiment to demonstrate the applicability of our approach.