A novel resource related faults detecting approach

In order to detect resource related faults in operating system, an approach based on path-insensitive analysis is proposed. The approach, which can detect a wider variety of resources issues, is context-sensitive. Test cases to identify errors are automatically generated. The models for the C code and resource related faults are set up. Furthermore a platform for resource faults detection is developed. We evaluate our approach by applying it to Linux 2.6.34 kernel. The results show that most resources related faults are successfully detected and located, with lower rate of false positive and false negative. Test cases generated by the platform greatly improve the efficiency of identifying real defects from the results of static analysis.