Next: 8. Appendix: Vnode Interface
Up: A File System Component
Previous: 6.5 4.4 BSD's nullfs
The proposed work strives for a radical improvement in the ease and
flexibility with which new file systems can be written and deployed. I
expect the most significant contributions of my thesis to be:
- The first language for the abstract description of file system
- The first method for writing file systems without access to the
sources for the target operating system.
- The first method for writing file systems that are portable across
different operating systems.
- A mechanism to produce either kernel or user-level file systems from
the same higher-level description.
- The performance degradation added by my mechanism would be small.