OSDev.org

I'm currently writing a VFS interface, and I want to manipulate my memory manager (malloc/free/realloc ...) as a file system.
after starting kernel, a page is allowed (by low level memory manager), and considered as a virtual disk that can grow...
the superblock stores global information, and every file stores the base and size of free/used blocks.
on demand the virtual disk can grow (by allowing new pages) to accept more files.

so I want to know what do you think about this ?
do u thing it is efficient ?