Why does ping to the same interface fail when loop-back is down?

When I down the loopback interface and try to ping my own IP, it's showing 100% packet loss, even when my eth0 is up. Why is this?

Here's what I did:

root@faisal-desktop# ifconfig lo down
root@faisal-desktop# ping -I eth0 172.16.10.112 -c 2
PING 172.16.10.112 (172.16.10.112) from 172.16.10.112 eth0: 56(84) bytes of data.

--- 172.16.10.112 ping statistics ---
2 packets transmitted, 0 received, 100% packet loss, time 1007ms

root@faisal-desktop# ifconfig
eth0     Link encap:Ethernet  HWaddr 6c:f0:49:f6:82:03  
         inet addr:172.16.10.112  Bcast:172.16.10.255  Mask:255.255.255.0
         inet6 addr: fe80::6ef0:49ff:fef6:8203/64 Scope:Link
                          .
                          .

Check the ifconfig TX packets count

root@faisal-desktop# ifconfig 
eth0  Link encap:Ethernet  HWaddr 6c:f0:49:f6:82:03  
      inet addr:172.16.10.112  Bcast:172.16.10.255  Mask:255.255.255.0
      inet6 addr: fe80::6ef0:49ff:fef6:8203/64 Scope:Link
      UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
      RX packets:25822 errors:0 dropped:0 overruns:0 frame:0
      TX packets:24825 errors:0 dropped:0 overruns:0 carrier:0
      collisions:0 txqueuelen:1000 
      RX bytes:19581301 (19.5 MB)  TX bytes:4013322 (4.0 MB)
      Interrupt:26 Base address:0x6000 

root@faisal-desktop# ping -c 1 -I eth0 172.16.10.112
PING 172.16.10.112 (172.16.10.112) from 172.16.10.112 eth0: 56(84) bytes of data.

--- 172.16.10.112 ping statistics ---
1 packets transmitted, 0 received, 100% packet loss, time 0ms

root@faisal-desktop# ifconfig 
eth0  Link encap:Ethernet  HWaddr 6c:f0:49:f6:82:03  
      inet addr:172.16.10.112  Bcast:172.16.10.255  Mask:255.255.255.0
      inet6 addr: fe80::6ef0:49ff:fef6:8203/64 Scope:Link
      UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
      RX packets:25840 errors:0 dropped:0 overruns:0 frame:0
      TX packets:24845 errors:0 dropped:0 overruns:0 carrier:0
      collisions:0 txqueuelen:1000 
      RX bytes:19588641 (19.5 MB)  TX bytes:4017338 (4.0 MB)
      Interrupt:26 Base address:0x6000

Faisal

Posted 2011-08-12T11:06:14.217

Reputation: 121

Answers

I got the answer
Pinging to same interface uses loopback interface only. Check here
And one doubt, in which function/module does the ip checking and replacing with loopback?

Faisal

Posted 2011-08-12T11:06:14.217

Reputation: 121

It doesn't say "*pinging to the same interface." it says "Anything sent to one of the host's own IP addresses is sent to the loopback interface." It also says "a loopback interface [...] allows a client and server on the same host to communicate with each other using TCP/IP.*" The implication is that without a loopback interface this is not possible. – RedGrittyBrick – 2011-08-16T09:31:49.470

Yea thats it. I have a doubt, do you know in which module / function that comparison happens? – Faisal – 2011-08-16T10:05:33.360

Alright, but i couldn't find it. Can anyone point out the file/function which is responsible for comparing packets(sending out) destination ip with interfaces ip, and changing to loopback !!. – Faisal – 2011-08-17T06:27:09.257

I doubt its in $(KERNEL_SOURCE)/net/ipv4/route.c – Faisal – 2011-08-17T12:38:21.157

Why does ping to the same interface fail when loop-back is down?

Because you don't ping an interface you ping an IP-address.

Your operating system works out which interface is the most appropriate one to use based on the target IP-address. If the target IP-address is your own, the most appropriate interface to use is the loopback interface as it has the lowest overheads.

RedGrittyBrick

Posted 2011-08-12T11:06:14.217

Reputation: 70 632

Agreed but now there is no loopback interface, so it should use other appropriate one. But it is not even calling the driver start_xmit function. – Faisal – 2011-08-16T04:49:55.650

Yep, it's simply verify ("via dev lo"):

# ifconfig eth0
eth0      Link encap:Ethernet  HWaddr 08:00:27:8b:38:0c
          inet addr:10.0.2.15  Bcast:10.0.2.255  Mask:255.255.255.0
          inet6 addr: fe80::a00:27ff:fe8b:380c/64 Scope:Link
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:539 errors:0 dropped:0 overruns:0 frame:0
          TX packets:550 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000
          RX bytes:60478 (59.0 KiB)  TX bytes:38274 (37.3 KiB)

# ip ro get 10.0.2.15
local 10.0.2.15 dev lo src 10.0.2.15
    cache

t3mp

Posted 2011-08-12T11:06:14.217

Reputation: 101