Solidity compiler defect detection method based on cross-component fuzz testing
By using cross-component fuzzing, test programs are generated using historical defect data and large language models, and then executed using the Ethereum Virtual Machine. This solves the problem that existing tools struggle to trigger and identify logic defects in the Solidity compiler, achieving efficient defect detection.
CN122152691APending Publication Date: 2026-06-05SUN YAT SEN UNIV
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SUN YAT SEN UNIV
- Filing Date
- 2026-02-05
- Publication Date
- 2026-06-05
Smart Images

Figure CN122152691A_ABST
Abstract
The application discloses a Solidity compiler defect detection method based on cross-component fuzz testing, obtains a seed program through data cleaning on historical defect data and test programs; data with the same defect component is screened from the historical defect data of the seed program, and defect features are extracted to obtain program generation prompts, and new test programs are generated according to the program generation prompts; bytecode and error information are generated by executing the new test programs through the compiler, the bytecode is executed through the Ethereum virtual machine to output execution logs, consistency comparison is performed according to the execution logs and the error information to determine logical defects, and a defect report is generated according to the logical defects; the program triggering the defect is used to guide program variation, and the program triggering the defect and the defect description are used to trigger the key features of the defect, so that the program triggering the compiler defect can be efficiently generated, and the problem that the generated test program cannot trigger the compiler defect is solved.
Need to check novelty before this filing date? Find Prior Art