Detection effect evaluation method and device for data race detection tool

Abstract

The embodiment of the invention provides a detection effect evaluation method and device for a data race detection tool. The method comprises the steps of operating a data race detection tool to be evaluated, detecting testing programs separately, and obtaining actual data race information output by the data race detection tool; comparing actual data race information with real data race information, and obtaining a detection effect evaluation effect of the data race detection tool. By comprising the actual data race information with the real data race information, the detection effect evaluation result of the data race detection tool is obtained. Due to the fact that the actual data race information output by adopting the data race detection tool is compared with the real data race information as the standard, the uniform standards can be adopted for evaluating the detection effects of different data race detection tools, high objectivity is achieved, and testers can be helped to select a proper data race detection tool.

Description

technical field [0001] Embodiments of the present invention relate to the field of software analysis, and more specifically, relate to a detection effect evaluation method and device of a data competition detection tool. Background technique [0002] With the improvement of computer hardware's parallel computing capability, more and more software uses multi-threaded implementation to improve computing efficiency. With the improvement of computing efficiency, concurrency defects may also be inadvertently introduced. Concurrency defects can be divided into the following three categories: data race defects, atomicity violation defects and deadlock defects. Among them, the data race defect means that different threads in the same program perform two write operations or one write and one read operation on a shared memory location in an improperly synchronized manner. Data race defects may lead to inconsistent variable values, resulting in program crashes, and serious property l...

AI Technical Summary

Problems solved by technology

Data competition defects may lead to inconsistent variable values, resulting in program crashes, and serious property damage

[0003] In related technologies, there are many kinds of data race detection tools. Since various data race detection tools use different data race detection algorithms, the detection effects of various data race detection tools are not the same.

Images