[c-nsp] allow self ping

[Phil Mayers]

On Tue, 2007-08-28 at 09:33 +0200, Gert Doering wrote:
> Hi,
> 
> On Fri, Aug 24, 2007 at 02:14:56PM -0500, Zhao, Wenmei (Sarah) wrote:
> > I have a MultiLinkPPP session up. Everything is working,
> > traffic is flowing and I am able to ping the remote side of the link, 
> 
> If you have anti-spoofing filters (or uRPF) configured, this is intentional.
> 
> Reason: on a self-ping, the router sends out the packet via the link
> in question (you can use that to test the link), and when the packet comes
> *back* from the other end, it fails the anti-spoofing test.

Interesting. I'd always assumed that such a packet didn't actually
physically egress anything, and was never entirely certain how the
"allow self ping" did what it does (it plainly does it - you need it to
ping yourself)

So if I have:

6500 [g8/1] --- l2switch --- (lots of hosts)

int g8/1
 switchport mode access
 switchport access vlan 10
int vl10
 ip address 10.1.1.1 255.255.255.0
 ip verify unicast source reachable-via rx

...and I do "ping ip 10.1.1.1 sourve vl10", what are the source/dest
ethernet MACs of the packet leaving g8/1 in order to make it come back
to the router? Or does it not actually leave the gig port, but gets
looped back inside the chassis somehow?

Just curious.