In the digital society, cloud computing is driving transformation with virtualization at its core. However, security threats, especially to virtualization engines such as Microsoft Hyper-V, are on the rise. Fuzzy testing is a key method to address these challenges, but traditional frameworks face difficulties in structured validation, limited attack surface, and manual analysis. This article proposes a coverage oriented automation method and state aware input generation for fuzzy testing of virtualization engines. By modifying KVM, implementing hypercall interfaces, and managing memory, we have built an automation framework that improves the effectiveness of test cases. Although the performance is good, limitations include single core CPU applicability and manual proxy construction. The future work aims to address these issues, enhance virtualization security, and ensure cloud stability.