Abstract:Context: Zero-day vulnerabilities are highly destructive and sudden. However, traditional static and dynamic testing methods cannot efficiently detect them. Objective: In this paper, a static fuzzy mutation method for program code is studied. This method can improve the efficiency of mutation sample generation according to the vulnerability evolution law, thus promoting the development of zero-day vulnerability detection methods based on deep learning techniques. Method: A static fuzzy mutation method based on the Abstract Syntax Tree (AST) is proposed. Under the guidance of software vulnerability evolution law, potential evolution paths that threaten program security are detected, and mutation samples containing vulnerabilities are generated at the syntax tree level based on the paths. To verify the effectiveness of static fuzzy mutation based on ASTs, this paper starts with Concurrent Use After Free (CUAF) homologous vulnerability. It uses multi-threaded programs to perform vulnerability feature statement insertion processing to infer the optimal mutation operator execution sequence corresponding to CUAF vulnerabilities triggered by data competition. The Linux kernel code is used to verify whether it can effectively reduce the number of invalid mutation samples. Results: In this paper, we filter the code fragments in the Linux kernel public code containing CUAF vulnerability fix commits and perform static fuzzy mutation on the fix versions of the vulnerabilities to reproduce the vulnerabilities of this type triggered by these code fragments on the timeline. We compare the process with the execution of the random mutation operator in traditional detection methods horizontally and improve the efficiency by 42.4% on average. Conclusion: The static fuzzy mutation based on the AST is effective in stages. When this method is explored in more vulnerability-type evolution laws, it is expected to promote the development of the zero-day vulnerability active detection technology framework.

Vulnerability Mining Method Based On Genetic Algorithm And Model Constraint

Test Case Generation for Vulnerability Detection Using Genetic Algorithm

CVDF DYNAMIC-A Dynamic Fuzzy Testing Sample Generation Framework Based on BI-LSTM and Genetic Algorithm

A Fuzzing technology with a data sample combination

Application of Genetic Algorithm on Multi-dimensional Fuzzing Technology

Software Vulnerability Detection Using Genetic Algorithm And Dynamic Taint Analysis

A Novel Method for Fuzzing Test Cases Generating Based on Genetic Algorithm

Software Vulnerability Detection Based on Code Coverage and Test Cost.

ASVG: Automated Software Vulnerability Sample Generation Technology Based on Source Code

Web Application Vulnerability Fuzzing Based on Improved Genetic Algorithm

Vulnerability Mining System Based on Program Process Tracing Technology

A Novel Seed Generation Approach for Vulnerability Mining Based on Generative Adversarial Networks and Attention Mechanisms

Fuzz Testing Based On Virtualization Technology

Vulnerability Analysis for X86 Executables Using Genetic Algorithm and Fuzzing

Research on Fuzzy Test Technology Based on Generative Adversarial Networks

An Improved Fuzzy Test Of Industrial Control System

V-Fuzz: Vulnerability Prediction-Assisted Evolutionary Fuzzing for Binary Programs

An Abstract Syntax Tree based static fuzzing mutation for vulnerability evolution analysis

V-Fuzz: Vulnerability-Oriented Evolutionary Fuzzing

A Mining Approach to Obtain the Software Vulnerability Characteristics