Messages in this thread | | | From | "H. Peter Anvin" <> | Subject | Suggested changes to the VFS | Date | Mon, 23 Mar 1998 14:51:52 -0800 (PST) |
| |
I would like to make the following changes against the VFS interface in 2.1, and I would very much appreciate your comments (just exactly what have I overlooked):
put_super:
This function would no longer be responsible for unlocking the superblock and setting the s_dev entry in the superblock to zero. Instead this would be performed by the VFS. This would permit moving invalidate_inodes() until after the superblock has been put, which is required to support virtual filesystems without requiring backing store for its inode information.
open: The individual filesystems would be responsible for allocating a file structure pointer and return it; a canned function would be available that would perform the equivalent of the current code in do_open (and, of course, this would still be performed equivalently for inodes which have NULL in their open fields.) This would allow the filesystems to return something other than a newly created file pointer. This could include a socket, a pipe, or a dup of an existing file pointer.
Incidentally, open is defined as a file_operation, which IMO is bogus, since you can only open() an inode, and the open function is passed an inode anyway. It would appear more logical to me to move open to the inode_operations structure.
Comments, flames?
-hpa
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu
| |