>From: Matthias Drochner <drochner@zelux6.zel.kfa-juelich.de>
>
>> Isn't this a bug in mount? To quote the man page (from my infamous i386
>> snapshot :-)
>
>It's a misfeature of the /etc/rc script, /usr is necessary to get
>LKMs loaded, so it can be mounted before it's time, and at the
>"standard" place again.
>It would be nice if "mount" would allow to combine the "a" flag
>with a filesystem specification however.

Why should the "-a" flag even be necessary to prevent the filesystem
from being mounted again?  On an ffs filesystem that's already mounted
I get:

vivax# mount
/dev/sd0a on / type ffs (local)
/dev/sd0e on /usr type ffs (local)
tardis:/usr on /import/tardis/usr type nfs
procfs on /proc type procfs (local)
kernfs on /kern type kernfs (local)
mfs:106 on /tmp type mfs (asynchronous, local)
vivax# mount /usr
ffs: /dev/sd0e on /usr: Device busy

So why can't we just make NFS mounts behave the same way, document it,
and once and for all prevent rc (or anything else) from ever making
the mistake of mounting something over itself?

I am given to undertand that OpenBSD has already fixed this problem,
in this way.

Cheers,
entropy

--
entropy -- it's not just a good idea, it's the second law.