DigHR: precise dynamic detection of hidden races with weak causal relation analysis |
| |
Authors: | Peng Luo Deqing Zou Hai Jin Yajuan Du Long Zheng Jinan Shen |
| |
Affiliation: | 1.Cluster and Grid Computing Lab, Services Computing Technology and System Lab,Huazhong University of Science and Technology,Wuhan,China |
| |
Abstract: | Present advanced precise race detection techniques leverage strict causal relation to locate data races, leading to false negatives and missing many hidden races. In this paper, we present a novel race detection technique, called DigHR, which can detect more hidden races than existing precise techniques. DigHR is based on a weak causal relation, called afterward-confirm (AC) relation, which relaxes restrictions of existing causal relations and is proved to be sound. To represent AC relations between events, we propose a new data structure, called water-flow graph, which enable DigHR to analyze AC relations and detect hidden races at run-time. We have implemented DigHR in C/C++ and evaluated the effectiveness of DigHR by detecting races from real-world benchmarks and applications. The experimental results show that DigHR can precisely detect 21% more races than previous techniques, including FastTrack and CP. |
| |
Keywords: | |
本文献已被 SpringerLink 等数据库收录! |
|