On Fri, 23 Mar 2001, Andreas Dilger wrote:
> Now I'm wondering, why not have the filesystems export this via /proc?
> Something like /proc/fs/<mountpoint>, creating empty directories if
Aiieee... And you are going to play with that tree whenever we rename
a directory on the path to some mountpoint? Thanks, but I'll pass...
Oh, and here's an instant self-LART: mount something on /proc/fs/bin ;-)
> needed to fill out the tree. Saves the hassle of where to mount all
> of the meta filesystems, and having to do the mounts in the first place
> (i.e. how do you know a filesystem supports meta-mounts)?
If you want to do fs-specific stuff you presumably know what fs you are
dealing with, right? ;-)
Anyway, procfs has a nasty problem regardless of the handling of mount
tree: removal of active entires. You really don't want to go there.
Notice that normally it is solved by having these entries per-driver,
so they stay alive as long as driver is not unloaded. Not an option
here, and yes, procfs sucks putrefied rat intensines through the straw
for stuff like /proc/sys/net/ipv4/neigh/* too.
Cheers,
Al
-
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to [EMAIL PROTECTED]
