--KLbH3019drDfQ/1p
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Mon, Mar 20, 2006 at 08:37:15PM -0500, Jed Davis wrote:
> Geert Hendrickx <ghen@telenet.be> writes:
>=20
> > How much is Xen backward/forward compatible?  e.g. once there is support
> > for NetBSD as dom0 on Xen 3, can I run a NetBSD 3.0 domU (i.e. for Xen =
2)
> > with it? =20
>=20
> My understanding is that it's not compatible at all, as important
> interfaces (domain control interface vs. XenBus, introduction of grant
> tables) have been completely changed.

Likewise, that's my understanding: it basically represents ports to
two different "machines".

There is theoretical caveat, mostly as an aside.

Xen3 also includes a shim that runs within the current framework as a
domU, and provides additional emulation on top of that interface to
look like a bare-metal machine, for native operating systems to run
in, like windows.  This uses bits of qemu for all of the things not in
the base hypervisor, like video bios and/boot cpu startup, and relies
on the VT extensions in newer processors as well.  I'm looking forward
to this.

The point, however, is that it may also be feasible/possible to write
a similar shim for back compatibility, that runs as a xen3 domU and
presents a xen2 hypervisor interface to older xen kernels, translating
as necessary. Thus the core hypervisor doesn't need to know about
maintaining compatibility. I recall discussion of this possibility at
one point, with the conclusion that if someone wants to write it they
can go ahead.  I don't know (and rather doubt) that anyone ever has.

--
Dan.
--KLbH3019drDfQ/1p
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2 (NetBSD)

iD8DBQFEH2TWEAVxvV4N66cRArTuAJ9M/EQM+3x+eHmsfjb6kdP9Dc2AoQCfVMcY
i7dLR3Hg8IP3qwKjzvJKiYs=
=qw8k
-----END PGP SIGNATURE-----

--KLbH3019drDfQ/1p--