This is an archived copy of the Xen.org mailing list, which we have preserved to ensure that existing links to archives are not broken. The live archive, which contains the latest emails, can be found at http://lists.xen.org/
Home Products Support Community News


Re: [Xen-devel] HVM networking in Xen 3.4-rc3

On Wed, 2009-04-29 at 17:04 +0100, Ian Jackson wrote:
> Gary Grebus writes ("[Xen-devel] HVM networking in Xen 3.4-rc3"):
> > I'm trying to understand how networking is intended to work for HVM
> > domains in Xen 3.4-rc3.  My domain config file contains:
> > 
> > vif = [ 'type=ioemu, mac=00:16:3e:5b:f8:ed, bridge=eth2' ] 
> This looks plausible.  Although I'm not 100% sure that the spaces are
> permitted.  None of my config files have them ...

Spaces don't make any difference.  I've tried various things in the

> > and the bridge is correctly configured on eth2.  But the xm create
> > command gets hung waiting for something to happen with the vif, and
> > eventually fails with:
> What does your /var/log/xen/xen-hotplug.log say ?

It's empty, but that's because nothing is ever running the hotplug

> > Is qemu supposed to be involved in the hotplug mechanism, or is there
> > something else I'm missing?
> It's all very complicated (much more so than it needs to be).  Most of
> the messing about is done by xend and by the hotplug script run by
> xend.
> > If modify qemu-ifup to write a value for the xenstore key, then the
> > domain finishes booting and the network works.
> The xenstore key is supposed to be done by the xend script, not
> qemu-ifup.  And anyway the xenstore key is not relevant to the
> operation of qemu.  I think xend is hanging up on itself.

Which is the xend script?  The hotplug scripts in /etc/xen/scripts seem
to be driven by the "online" and "offline" events
in /etc/udev/xen-backend.rules, but I can't see anything that generates
those events when qemu is involved.  Maybe there is some kernel piece
that I am missing?  I'm using a (openSUSE 11.1) Xen kernel.

I'm not sure how this would work for PV-on-HVM drivers either, since
xend is blocking before it boots the guest.


Xen-devel mailing list

<Prev in Thread] Current Thread [Next in Thread>