Concurrent-program test method based on lock-object splitting strategy and test system thereof

Abstract

The invention discloses a concurrent-program test method based on the lock-object splitting strategy and a test system thereof. The concurrent-program test method includes the steps that relevant codes are analyzed with the static program analysis technology, and shared variables protected without same lock objects are determined; then abnormal access codes of the shared variables are split, the split codes are protected through the same lock objects, and the fine-grained synchronization method is generated; reachability testing (all feasible synchronization sequences are operated) is carried out, interleaved sequences between fine-grained synchronization events on different monitor objects in a raw to-be-test concurrent program are generated and carried out, and as the shared variables are protected through false lock objects, concurrent program errors such as data hazards, atomic against and deadlocking are shown. By means of the concurrent-program test method based on the lock-object splitting strategy and the test system thereof, the capacity of detecting the concurrent program errors can be effectively improved.

Description

Technical field [0001] The invention relates to software analysis and testing technology in software engineering, and specifically relates to a concurrent program testing method based on a lock object split strategy and a testing system thereof. Background technique [0002] The popularity of multi-core technology and the support of high-level programming languages ​​for concurrent facilities have made the development and use of concurrent software increasingly widespread. Concurrent software testing is an important means to ensure the quality of concurrent software. Due to the existence of concurrency and competition, under the same input, different concurrent program operations may produce different results due to the selection of different synchronization sequences. The uncertainty of concurrent program operation makes it very difficult to test, debug and repair concurrent program errors. Concurrent program errors mainly include data competition, atomic violations, deadlocks...

AI Technical Summary

Problems solved by technology

[0006] Although there are many advantages such as being able to dynamically capture program behavior, generating a relatively complete concurrent program synchronization sequence set, and not needing to record any synchronization sequence history records, reachability testing still faces the problem that the generated test case set is not sufficient and cannot effectively detect shared variables. Problems such as concurrent program errors caused by improper protection

For example, in the process of reachability testing, generally only generate competition variants between synchronization methods or synchronization blocks protected by the same lock object, run the corresponding synchronization sequence, and access methods or shared variables not protected by the same lock object Code blocks cannot generate and run interleaved sequences, so they cannot effectively detect concurrent program errors such as data races, atomic violations, and deadlocks caused by using wrong lock objects to protect shared variables

Images