Ttyjfyk

I have some difficulty to picture a diagram for the meaning of loopback. https://en.wikipedia.org/wiki/Loopback#Virtual_loopback_interface

Any traffic that a computer program sends to a loopback IP address is
simply and immediately passed back up the network software stack as if
it had been received from another device.

What does the action "loop back" mean?

What are the source and destination in the loopback scenario?

How is that different from a regular scenario?

Could you use some commands to show the loopback scenario and the regular one?

For example:

• When I type ssh localhost, it works just like ssh with any other IP address.


• When I type http://localhost:631 in browser's address bar and hit return, it works just like typing any other URL.

"loopback" is supposed to mean "a message or signal ends up (or loops) back to where it started" (https://askubuntu.com/questions/247625/what-is-the-loopback-device-and-how-do-i-use-it). But in the two examples, I don't see that meaning happening, but just the same thing as non-loop back IP addresses.

Thanks.

You are confusing the device driver layers and the IP layers of the stack.

At the IP layer, 127.0.0.1 is just another IP address and treated the same.

At the driver layer, packets sent via the loopback interface are "simply and immediately passed back up the network software stack" as opposed to being sent to a network card.

This concept is not OS specific; various OSes use the same concept.

 | --- > ---
lo | |
 | --- < ---/

 | --- > --- |
veth0 | | veth1
 | --- < --- |

A veth pair is a pair of ethernet devices with a cable between them. Implemented as a virtual device. They are designed to communicate with containers: one end of the veth pair can be moved into a container.

You can think of lo like a veth pair, except there is only one end.

In reality the implementation details are not the same. This is at least clear when you look at the strange fact that ping 127.0.0.2 works, but you cannot see the address 127.0.0.2 in ip -4 addr. But there is no need to care unless you are exploiting that "feature" of the legacy version of Internet Protocol, or you are developing the kernel.

A more advanced way to think of lo is that it does not need to do anything, just like the dummy interface. (You can play with ip link add type dummy).

When you ping the IP address of your computer's ethernet or wifi interface, it works without sending anything over that interface. It is possible to verify this with tcpdump or the packet counters in ip -s link. The same is true for dummy, and the same is true for lo.

But if you haven't observed this, you can use the less advanced explanation. It does not really mislead you.

There is some difference in how lo and dummy are used, but that is the magic 127.0.0.2 stuff that is not important to the question.