On Fri, 2 Jun 2000 22:31:21 +0200, Jamie Lokier
<lk@tantalophile.demon.co.uk> wrote:
>I still don't see _any_ reason for the new vfsmnt/dentry stuff. What do
>you gain? The ability to mount devfs in chroot jails. In a chroot
>jail, you don't want the same devfs tree as you find in /dev anyway.
>
>What do you lose? The vfs data structure is even more complicated now,
>with yet another layer of indirection. And the user space view of
>filesystems is more complicated too.
>
>I'm sure it's very clever, but I really don't see the point.
Consider the case where you have several filesystems that you want to
export via NFS to different machines. Say you want each machine to get a
unique view of those filesystems. There is no convienient way to do this
without something similar to the bind functionality -- symlinks are
resolved on the client with NFS.
Here's a simple example:
/export1/
fs1/
fs2/
/export2/
fs2/
fs3/
/export3/
fs1/
fs3/
This exact situation has come up in the past for a product I worked on
(except with a far more complicated set of exported trees). The solution
we came up with was less flexible, less clean, and less elegant than using
bind mounts. (Also note, we had a requirement that the client should only
have to mount one directory from the server to get its entire tree.)
Aaron
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Wed Jun 07 2000 - 21:00:16 EST