PDOS: Experimenting with Garbage Collection in an Otherwise Conventional OS

点击翻译按钮获取中文摘要

Security holes and crashes at the operating system (OS) levelrepresent a serious infrastructure problem. These OS bugs are oftenthe result of memory management errors, which are very hard fordevelopers to avoid. In the long run, a broad class ofmemory-management errors can be prevented by using a high-level,type-safe language, but whether these languages are suitable forkernel implementation remains an open question. Unfortunately, anygiven experiment to answer the question (i.e., using a particularhigh-level language) involves many differences in implementationcompared to a conventional kernel, making it difficult to drawconclusions about which parts of the high-level approach work andwhich parts do not. This project is an experiment specifically aboutgarbage collection for legacy OS kernels, changing as few otherimplementation issues as possible. Concretely, the research is aboutdeveloping tools to automatically transform the C source code of aconventional kernel so that it is compatible with a variety of precisegarbage-collection strategies. The project's ultimate goal is to checkwhether the OS becomes usefully more reliable as a result of garbagecollection, and to measure the performance costs in both desktop andembedded environments. If the experiment shows that the OS improves ata reasonable cost, then the research will have shown how to increasethe reliability of a major part of today's computer systemsinfrastructure. Tools generated by the project will be made publiclyavailable for use in research, for application to practical software,and for pedagogical purposes.