Re: [libvirt] [PATCH] Add support for 'passthru' mode for direct network interfaces

[Laine Stump <laine laine org>]

On 05/11/2011 08:21 AM, Daniel P. Berrange wrote:
> On Tue, May 10, 2011 at 04:21:29PM -0600, Eric Blake wrote:
> > On 05/10/2011 05:59 AM, D. Herrendoerfer wrote:
> > > Hi all,
> > >
> > > starting with kernel 2.6.38 macvtap supports a 'passthru' mode for
> > > attaching virtual functions
> > > of a SRIOV capable network card directly to a VM.
> > > This patch adds the capability to configure such a device.
> > >
> > > Signed-off-by: Dirk Herrendoerfer<d.herrendoerfer at herrendoerfer.name>
> > >
> > > diff --git a/docs/schemas/domain.rng b/docs/schemas/domain.rng
> > > index 7163c6e..e2a66ae 100644
> > > --- a/docs/schemas/domain.rng
> > > +++ b/docs/schemas/domain.rng
> > > @@ -2351,7 +2351,7 @@
> > >     </define>
> > >     <define name="bridgeMode">
> > >       <data type="string">
> > > -<param name="pattern">(vepa|bridge|private)</param>
> > > +<param name="pattern">(vepa|bridge|private|passthru)</param>
> > >       </data>
> > >     </define>
> > >     <define name="addrMAC">
> > Missing corresponding documentation in docs/formatdomain.html.in to
> > describe the new mode.
> >
> > What happens if you try to use this mode but you have an older kernel?
> > Do we need to add any sanity checking to give a saner error message than
> > just blindly trying the string and reporting back whatever message gets
> > returned about an unknown argument?
> >
> > The rest of this patch looks okay to me (with my disclaimer that I've
> > never used an SRIOV card), but I'd rather wait for a v2 than push this
> > without documentation.
> The kernel GIT commit actually has some useful info
>
>
> commit eb06acdc85585f28864261f28659157848762ee4
> Author: Sridhar Samudrala<sri us ibm com>
> Date:   Thu Oct 28 13:10:50 2010 +0000
>
>      macvlan: Introduce 'passthru' mode to takeover the underlying device
>
>      With the current default 'vepa' mode, a KVM guest using virtio with
>      macvtap backend has the following limitations.
>      - cannot change/add a mac address on the guest virtio-net
>      - cannot create a vlan device on the guest virtio-net
>      - cannot enable promiscuous mode on guest virtio-net

I had read that commit notice earlier, and didn't understand the first 
item - which MAC address needs to be changed? Doesn't the guest have its 
own MAC address, unrelated to the MAC address of the underlying device 
on the host?


>      To address these limitations, this patch introduces a new mode called
>      'passthru' when creating a macvlan device which allows takeover of the
>      underlying device and passing it to a guest using virtio with macvtap
>      backend.
>
>      Only one macvlan device is allowed in passthru mode and it inherits
>      the mac address from the underlying device and sets it in promiscuous
>      mode to receive and forward all the packets.
>
> Daniel