Method and system for improving coverage of fuzzy test

Abstract

The invention provides a method and system for improving the coverage rate of a fuzzy test. The method comprises the steps: S1, preprocessing a target program serving as a test case to obtain a call graph of a function in the target program; S2, determining anchor nodes for fuzzy test in the target program based on the call graph, the anchor nodes comprising a first anchor node based on a complexnetwork and a second anchor node based on a neural network; S3, according to the anchor nodes, determining the distances from the seed nodes to the anchor nodes through an instrumentation tool; and S4, calculating the energy of the seed nodes based on the distance, and selecting the seed nodes of which the energy is higher than a threshold value as a basis for evaluating the mutation test case.

Description

technical field [0001] The invention relates to the field of software testing, in particular to a method and system for improving the coverage rate of fuzz testing. Background technique [0002] Fuzzing testing (fuzzing testing) is a software testing method that provides invalid, unexpected or random input data to the program. Once the program has errors such as crashes or assertion failures, existing defects will be found. Fuzzing test is not only applicable to source software, but also to binary software. It has a wide range of applications. It dynamically executes the target software through a large number of test cases and analyzes the dynamic execution process. The false positive rate is lower than that of static analysis techniques. In addition, the fuzzing implementation principle is simple, without a lot of theoretical derivation and formula calculation, avoiding the path explosion problem in the symbol execution process, with a high degree of automation and a small ...

AI Technical Summary

Problems solved by technology

[0005] Although AFL records the execution path of the test case and increases the number of test cases that cover the new execution path to increase its coverage of the code, this method can improve the effect of Fuzzing to a certain extent, but there is still a certain blindness In terms of the generation of seed files, the analysis granularity of the program is still relatively coarse, and the pertinence is poor

The improvement of AFL is also to improve the code coverage of fuzz testing. The main improvement of these methods is to increase the coverage branch of the forward execution path from the program entry to the current execution point to improve the coverage. This method is fine-grained and subject to program The execution context has a large impact, and since it starts from the current execution point, the contribution to improving coverage is limited

Images