基于Xen的声音设备虚拟化的研究-计算机软件与理论专业论文.docxVIP

华中科技大学硕士学位论文 华 中 科 技 大 学 硕 士 学 位 论 文 II II Abstract With the introduce of virtualization technology, there are many achievements about the research of virtualizing CPU and memory , high-quality I/O virtualization is still a big challenge. Currently Xen paravirtualization solution doesn’t provide the support for sound device, so user can’t play audio in Xen paravirtualized environment, which directly impacts the user experience in Xen paravirtualized environment. Based on split driver model, creating a frontend sound device driver in unprivileged guest and creating a backend sound device driver in privileged guest can implement sound device virtualization in Xen paravirtualized environment. The frontend sound device driver builts a virtual sound device based on the ALSA driver architecture, which provides abstract interfaces for upper. The frontend sound device driver intercepts the audio requests via these interfaces and sends them to the backend sound device driver, which receives audio requests and passes them to native sound device driver to operate sound device. For high-efficiency, using the device I/O rings to pass requests and responses, the grant tables to transfer data without copying and event channel for asynchronous notifications. The connection between frontend and backend sound devices is set up based on their XenBus states, which ensures the operating orders of frontend and backend. In Xen paravirtualized environment, frontend and backend devices are registered on the XenBus. To register frontend and backend sound devices, add support for sound device in the configuration file, and populate the frontend and backend sound device to the XenStore to register them. The test results show that user can play fluently audio in Xen paravirtualized environment with the help of our sound device virtualization, and it provides better user audio experience than in Xen full virtualized environment. Key words: I/O virtualization, Split driver model, Paravirtualizati