Dongyoon Lee and Changhee Jung, assistant professors in computer science, have been selected for a Google Faculty Research Award. The project is entitled “TxRace: Efficient Data Race Detection Using Commodity Hardware Transactional Memory”. Read the full abstract below or click here:
Detecting data races in multithreaded programs is crucial for ensuring their correct execution, but the high runtime overhead prevents the wide use of dynamic data race detectors. We propose TxRace, a new software data race detector that leverages commodity hardware transactional memory (HTM) to speed up data race detection. TxRace instruments a multithreaded program to transform synchronization-free regions into transactions, and exploits the conﬂict detection mechanism of HTM for lightweight data race detection at runtime. However, the limitations of the current commodity HTMs expose several challenges in using them for data race detection: (1) frequent transaction abort due to non-conﬂict reasons (e.g., unknown or capacity), (2) lack of ability to pinpoint racy instructions, and (3) cache line granularity of conﬂict detection leading to false positives. To overcome such challenges, this proposal seeks to build an efﬁcient, practical tool that performs lightweight HTM-based data race detection at ﬁrst, and occasionally switches to slow yet precise data race detection only for small fraction of execution intervals in which potential races are reported.