On Tue, May 23, 2000 at 04:49:50PM -0400, Thomas Bushnell, BSG wrote:
> Marcus Brinkmann <[EMAIL PROTECTED]> writes:
>
> > Mmmh. Assume a rename happens while there is an open file handle.
> > What happens if then fsys_getfile is called, which does an lookup_hard on
> > the cache_id in the file handle?
>
> Rename has to change the cache_id in the file handle itself too.
> Certainly everything must be kept consistent.
Ah, okay.
> Because multiple links to a file are not permitted, you have to define
> your own diskfs_dir_rename anyway; the generic one will not be
> sufficient.
Yes. While we are at it, do I have to define my own diskfs_init_dir, too?
It increases nlink for . and .., and checks the number of allowed links for
the parent directory for the latter. (if yes, this should probably be added
to the info doc on diskfs_link_max).
About the inode stuff: Currently I settled on a virtual inode concept which
decouples the inode numbering from the directory location. This is easier on
the reader of the code (no weirdness), simpler to implement and it should be
possible to use a common implementation for all inodeless filesystems there
to come.
Thanks,
Marcus
--
`Rhubarb is no Egyptian god.' Debian http://www.debian.org Check Key server
Marcus Brinkmann GNU http://www.gnu.org for public PGP Key
[EMAIL PROTECTED], [EMAIL PROTECTED] PGP Key ID 36E7CD09
http://homepage.ruhr-uni-bochum.de/Marcus.Brinkmann/ [EMAIL PROTECTED]
