Re: [Xen-users] Mtu/802.1Q limits on vif ??

[Yann Dupont <Yann.Dupont@xxxxxxxxxxxxxx>]

Yann Dupont a écrit :

>Nuno Silva a écrit :
>
>  
>
>>Yann Dupont wrote:
>>
>>[..]
>>
>>    
>>
>>>The idea is to get the 802.1Q port on the Router6 host (via the bridge),
>>>and then use virtual interfaces (eth1.xxx where xxx are the vlan ID)
>>>inside the XenU.
>>>
>>>Well, it works, only if I use mtu 1496 for the vifs.
>>>I know the physical interfaces (eepro1000) can cope with vlan natively.
>>>Don't know for the bridge code but i'm not aware of limitations there.
>>>      
>>>
>>This happens on real hardware too. The usual workaround is to set the
>>eth0 to MTU 1504. Like this you can set the vlans to 1500.
>>
>>    
>>
>Well, I Tried lots of things ( raise the MTU on physicall interface
>(dom0) , on the bridge (dom0), on the vifs (dom0/domU), etc etc)
>neither of this works (various errors).
>
>Don't know where the limitation is : bridge or xen vif.
>
>I'll now go for PCI virtualisation, as I really don't have usage of THIS
>physical interface on dom0.
>I bet this will work.
>
>Thanks for your answer,
>
>  
>
Well. this is working as expected. Using the native interface in this
domain gives me the full MTU, which proves
that the problems lies in the bridge code, or in the vif code.

Anyway, two minors things :

As noted by someone else on the list,
I had to enable "privileged guest (domain 0)" To have a working kernel.

Just enabling physical device access isn't sufficient. In fact, the
kernel immediately crash.
So it seems there is something broken here.

Tried this with 2.0.6 and 2.0.testing (version of 07 18 2005)

Other (minor) problem : Using a domain0 kernel works fine, But when I
shutdown this domain, and re create it, most of the time physicall
access to the net device (epro1000 if that matters) is not working
anymore (the driver gives an error)

Anyway ,this is great jobs, guys.

-- 
Yann Dupont, Cri de l'université de Nantes
Tel: 02.51.12.53.91 - Fax: 02.51.12.58.60 - Yann.Dupont@xxxxxxxxxxxxxx


_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users