Manuel Bouyer wrote:
> On Wed, Dec 19, 2007 at 09:18:55PM -0200, Marco Poli wrote:
>   
>> Hello world,
>>
>>
>> I must confess I have been waiting for the release of the amd64 port for 
>> asome months now... to be precise since I bought this Xeon 5120 in this 
>> PowerEdge 1900 server, around june I guess.
>>
>> Since the release I have built up a testing environment using -daily 
>> from 20071215 (kernel 4.99.42). And the latest  xenkernel3 and xentools3 
>> packages. I did make my own kernel with VT100, Veriexec and CGD support, 
>> based on the config of the XEN3_DOM0 kernel.
>>
>> For this test I have:
>>
>> kern.securelevel=-1
>> kern.veriexec.setric=0
>>
>>
>> So far so good. The problem appears when trying to bring the first DomU 
>> up, with the following config file:
>>
>> kernel = "/netbsd-INSTALL_XEN3_DOMU.gz"
>> memory = 128
>> name = "machine"
>> vif = [ '' ]
>> disk = [ 'file:/mnt/virtual/netbsd-common.img,hda1,w' ]
>>
>>
>> When bringing up xend, in the logs I have:
>>
>> [...]
>>
>> Error: Device 0 (vif) could not be connected. Hotplug scripts not working.
>>     
>
> If you don't want network interface, I think you just have to delete the
> vif = ['']
> line. Otherwise it create a vif with some defaut values that are not
> right for NetBSD.
> If you want a vif, you need to add a bridge= parameter pointing to an
> existing bridge interface.
>
>   
Was the attempt to invoke brctl an indication that the linux variant 
will automatically create a bridge if none exists or was it just a side 
effect of an empty vif value? I'm just curious as to whether this is a 
facility not utilised in NetBSD ... or if it's just the code's way of 
saying the vif value was bad.

Sarton