Automated memory corruption detection through analysis of static variables and dynamic memory usage

Various methods for memory fault detection have been developed through continuous study. However, many memory defects remain that are difficult to resolve. Memory corruption is one such defect, and can cause system crashes, making debugging important. However, the locations of the system crash and the actual source of the memory corruption often differ, which makes it difficult to solve these defects using the existing methods. In this paper, we propose a method that detects memory defects in which the location causing the defect is different from the actual location, providing useful information for debugging. This study presents a method for the real-time detection of memory defects in software based on data obtained through static and dynamic analysis. The data we used for memory defect analysis were (1) information of static global variables (data, address, size) derived through the analysis of executable binary files, and (2) dynamic memory usage information obtained by tracking memory-related functions that are called during the real-time execution of the process. We implemented the proposed method as a tool and applied it to applications running on the Linux. The results indicate the defect-detection efficacy of our tool for this application. Our method accurately detects defects with different cause and detected-fault locations, and also requires a very low overhead for fault detection.