Accurate Static Data Race Detection for C

Accurate Static Data Race Detection for C

연구 분야: Strategies

학회: International Symposium on Formal Methods

Data races are a particular kind of subtle, unintended program behaviour arising from thread interference in shared-memory concurrency. In this paper, we propose an automated technique for static detection of data races in multi-threaded C programs with POSIX threads. The key element of our technique is a reduction to reachability. Our prototype implementation combines such reduction with context-bounded analysis. The approach proves competitive against state-of-the-art tools, finding new issues in the implementation of well-known lock-free data structures, and shows a considerably superior accuracy of analysis in the presence of complex shared-memory access patterns.