> What I was thinking about here is more or less a two-liner:
> 
> load the module
> mount -a -t procfs
> 
> To unload, umount -a -t procfs and a moduload.
> 
> I was thinking of ``this script mounts procfs, and if needed loads the
> lkm needed for it'' as one functional unit.

of course, if you're going to provide things like this, why not just
make the process relatively transparent e.g. by changing mount_foo to:
	(1) try to mount the file system of type foo,
	(2) if that failed (with an error code that indicates
	    no such file system exists in the kernel):
		(a) see if it's loaded,
		(b) try to load it if not (if so, fail mount operation),
		(c) if load succeeds, try to mount it again.

With a standard location for loadable modules, etc., (2) could very
easily be a library routine.


Unless i'm mistaken, FreeBSD did something not _too_ dissimilar to
this, though i've not looked at their code in a Long Time.


chris