Optimising OpenGBF for Detecting Concurrency Issues: Strategic Improvements for Enhanced Fuzzing

Abstract

Although concurrency testing is crucial for improving software security and stability, it frequently encounters considerable difficulties, as investigating every potential non-deterministic interleaving that the concurrent execution may have is necessary to evaluate concurrent systems and avoid issues such as Deadlock. OpenGBF —a completely new, open-source grey-box fuzzer with concurrency awareness that tests various thread schedules by injecting random delays. This research endeavours to enhance the thread scheduling technique suggested by OpenGBF in order to increase issue identification rate while reducing performance overhead. Four different approaches were evaluated and put to the test using various benchmarks. From these tests, it became clear that some of the proposed methods were promising while the case is not the same for the rest, it also highlighted the necessity of further refinement for greater improvements.