--dc+cDN39EJAMEtIO
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Mon, Sep 04, 2006 at 10:50:22PM +0200, Pavel Cahyna wrote:
> On Mon, Sep 04, 2006 at 10:06:51PM +0200, Martin Husemann wrote:
> > On Mon, Sep 04, 2006 at 08:49:13PM +0100, David Laight wrote:
> > > There is nothing (to my knowledge) to ensure that the 'wedge numbers'
> > > stay anything like consistent across reboots.
> > > So add a 'partition' to an MBR disk, and when you reboot all the numb=
ers
> > > will change.
> >=20
> > Then don't boot from MBR disks ;-)
> >=20
> > We need to be able to mount-by-guid (if using GPT), or some equivalent
> > for other on-disk partitioning schemes that store identifiers.
>=20
> How will it fix filehandles, permissions for nodes in /dev and anything
> else that uses device numbers?

Well, with file handles you always have had the problem that the fsid may=
=20
change between boots. Now that file handles are opaque, things are a bit=20
harder, but something can still be done about it.

As for the other questions, please see the threads in tech-kern over the=20
past few years.

The up-shot is that we need a way to rename and re-minor-number devices.=20
Combine that with a daemon that keeps state in userland, and we force=20
wedges to come back with the same minor # as before.

Take care,

Bill

--dc+cDN39EJAMEtIO
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.3 (NetBSD)

iD8DBQFE/bRcWz+3JHUci9cRAiXHAKCDqV66OOtpxhSH3jyWMCqvVYKaXQCeIPz7
I1yCuVcIeWmQhG7D8rvX9og=
=JJJK
-----END PGP SIGNATURE-----

--dc+cDN39EJAMEtIO--