On 15-07-14 19:34:42, Jeff Haran wrote:
> 
> The last time I looked into this (and this was quite a while ago so take this 
> with much salt) the lo device was not a "regular network device". The stack 
> sets up an instance of lo in every network name space created, more or less 
> automatically via the registration of its init routine in a struct 
> pernet_operations. So lo doesn't initialize like most other network devices. 
> You might have issues there. I know I did last time I tried to "borrow" 
> loopback.c to do something similar for some prototyping I was doing.

Realized that while looking at this code. But thankd for pointing this.


> 
> If you are willing to monkey with kernel code, it might be easier to just 
> twiddle the link sense routine in driver for the "other NIC" you were 
> planning to use to always report the link as up regardless of what the PHY 
> says.
>
This too is good option. 

Found anther option that looks like should solve the problem I am after. veth
device - as you suggested earlier. So, as I understand veth device creates a
sender and receiver. So, if I send on veth0, I will receive on veth1. So, the
steps that I would need is (using iproute2 utility):

ip link add type veth name veth0 peer name veth1
ip link set dev veth0 up
ip link set dev veth1 up
and then add the address to what ever I require.

I quickly tested and seems that I am able to play on veth0 and receive on
veth1.

One last query on this, does this interface receive any packet by default of
any protocol. If so, how do I make sure that no packets other than one
injected by me are seen on this interface. Sorry, if this sounds too stupid,
but I am noob in this area.

>
>
> 

-- 
 Thanks,
-aka


_______________________________________________
Kernelnewbies mailing list
[email protected]
http://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies

Reply via email to