Re: [Xen-devel] Netback & NetFront Driver

[Mats Petersson <mats@xxxxxxxxxxxxxxxxx>]

At 06:31 09/07/2007, TEJ wrote:
> On 7/8/07, Mats Petersson <mats@xxxxxxxxxxxxxxxxx> wrote:
> > At 16:12 08/07/2007, TEJ wrote:
> > >On 7/8/07, TEJ <tejparkash.d@xxxxxxxxx> wrote:
> > >>On 7/8/07, Mats Petersson <mats@xxxxxxxxxxxxxxxxx> wrote:
> > >> > At 08:03 2007-07-08, you wrote:
> > >> > >hi
> > >> > >
> > >> > >i am new to the xen development list. I am just curious about the
> > >> > >networking architecture of the XEN.
> > >> > >
> > >> > >there are two separate things i have read.
> > >> > >
> > >> > >first is the Network driver architecture is a spilt driver (Netfront
> > >> > >and NetBack)...
> > >> > >Second is while domu is coming up it creates the virtual interfaces
> > >> > >like (vif<id>.0 <--> (v)eth0).
> > >> > >
> > >> > >I am unable to relate these two concept. As far as my understanding
> > >> > >say that, for all the (v)eth0, eth1, eth2 interface in domu is managed
> > >> > >by Netfront Driver. So IP layer in domu talks to Netfront Driver and
> > >> > >uses one of the veth and then pass packet through that channel i.e.
> > >> > >corresponding vif through hypervisor.
> > >> >
> > >> > That is correct - the packet is sent by netfront (acting on behalf of
> > >> > "eth%d" in DomU) to netback.
> > >> >
> > >> >
> > >> > >All the vif<id>.x interfaces in the dom0 will be managed by NetBack
> > >> > >Driver. Since all the vif's are grouped with bridge which actaully
> > >> > >forward the packet to the actual driver of correspinding vif
> > >> > >interface.
> > >> >
> > >> > The netback receives the packet from the frontend driver, and passes
> > >> > it to the listener for "vif%d.%d". This ends up in the bridge (or
> > >> > route) component, which forwards the packet to the correct hardware
> > >> > port. This part is no different from the normal bridge/route
> > >> > behaviour (taking a network packet on one device and sending it out
> > >> > on the other device, e.g. linking eth0 to eth1 - of course, when
> > >> > doing this, usually there are rules about which packets go where,
> > >> > e.g. for routing, you only send packets intended for the other
> > >> > network across).
> > >
> > >just i am curious to know that why it there is like IP forwarding and
> > >MAC forwarding. because as i understand, MAC forwarding is sufficient
> > >to forward the packet. and it is performance hit if the IP forwarding
> > >is implemented in the DOMU.
> > >
> > >Can you give a link to the page saying this - or some other
> > >reference to what you're on about, since I don't quite know what you
> > >mean (and I may not be able to answer the question, but with a link
> > >I may understand it better and determine if I actually do know how
> > >it works or not).
> hi Mat sorry for this crap... the document which i am referring is
>
> http://ebtables.sourceforge.net/br_fw_ia/br_fw_ia.html
>
> in this there is one paragraph written as
>
> When the bridge code and netfilter is enabled in the kernel, the
> iptables chains are also attached onto the hooks of the bridging code.
> However, this does not mean that they are no longer attached onto
> their standard IP code hooks. For IP packets that get into contact
> with the bridging code, the bridge-nf code will decide in which place
> in the network code the iptables chains will be traversed. Obviously,
> it is guaranteed that no chain is traversed twice by the same packet.
> All packets that do not come into contact with the bridge code
> traverse the iptables chains in the standard way.
> it made me bit confused about the bridging part.
>
> thanks for ur reply
I must confess that this didn't clarify much to me with regards to 
your previous question.
Does this mean that you no longer have a question about bridging and 
how it works in Xen? If so, happy to help. If not, can you please 
rephrase the question so that we can figure out an answer (I 
personally isn't an expert on IP routing/bridging - I know enough to 
understand that I DON'T know nearly enough about it [is that a 
sensible sentence?], but that's about it - I just use it in Xen with 
it's standard default setup and let someone else worry about the more 
complex configurations).
--
Mats


> TEJ
> > --
> > Mats
> >
> >
> >
> > >> >
> > >
> > >>
> > >>thanks for your confirmation Mat,
> > >>
> > >> > --
> > >> > Mats
> > >> >
> > >> >
> > >> > >I just want somebody confirmation on that or somebody want to add
> > >> > >something over this will be helpful to me
> > >> > >
> > >> > >i have attached one doc if some is interested can through some
> > >> light on that..
> > >> > >
> > >> > >i am sorry if it is totally a crap.
> > >> > >
> > >> > >thanks
> > >> > >TEJ
> > >> > >
> > >> > >
> > >> > >Content-Type: application/msword; name=networking.doc
> > >> > >X-Attachment-Id: f_f3v6ewmc
> > >> > >Content-Disposition: attachment; filename="networking.doc"
> > >> > >
> > >> > >_______________________________________________
> > >> > >Xen-devel mailing list
> > >> > >Xen-devel@xxxxxxxxxxxxxxxxxxx
> > >> > >http://lists.xensource.com/xen-devel
> > >> >
> > >> >
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel