18

In Linux, the command

ip address add [...]

has a scope argument. The man page says that the scope is "the scope of the area where this address is valid". Follows the list of legal scopes:

  • global
  • site
  • link
  • host

What does this "area" of "validity" refer to?

21

from http://linux-ip.net/html/tools-ip-address.html :

Scope | Description

global | valid everywhere

site | valid only within this site (IPv6)

link | valid only on this device

host | valid only inside this host (machine)

Scope is normally determined by the ip utility without explicit use on the command line. (...)

The following citations are from the book Understanding Linux network internals by Christian Benvenuti, O'Reilly:

"The scope of a route in Linux is an indicator of the distance to the destination network. The scope of an IP address is an indicator of how far from the local host the address is known, which, to some extent also tells you how far the owner of that address is from the local host (...).

Host: An address has a host scope when it is used only to communicate within the host itself. Outside the host this address is not known and can not be used. An Example is the loopback address, 127.0.0.1

Link: An address has a link scope when it is meaningful and can be used only within a LAN. An example is a subnet's broadcast address.

Global: An address has global scope when it can be used anywhere. This is the default scope for most addresses. (...)"

The main reason to use scopes seems to be that a host with multiple interfaces and addresses has to decide when to use which address. For communication with itself a loopback address (scope host) can be used. With communication elswhere, a different address has to be selected.

  • 2
    Thanks for your answer. I alread had read that. What I really don't understand is what 'valid' means here. In other words: how will the scope argument impact the network device behaviour? – rolaf Sep 7 '09 at 12:46
  • Ip addresses and routes have a scope to indicate to the kernel the distance to other networks/addresses. See books.google.de/… :) – brengo Sep 7 '09 at 12:56
  • I should have added: "scope" is for routing decisions and sanity checks of the routing configuration. It helps the kernel to decide (faster) where the packets should go. Hope that helps? – brengo Sep 7 '09 at 13:01
  • 1
    Sorry brengo, your link is not consultable right now. But isn't the metric argument used to specify the distance? – rolaf Sep 7 '09 at 13:01
  • Oh, sorry, the google-link which works fine at my end :( has an exerpt of exactly the needed topic (Understanding Linux network internals, by Christian Benvenuti, O'Reilly) which is hard to press in 600 comment-characters but I'll edit my answer above – brengo Sep 7 '09 at 13:16
14

To begin to understand the definition of scope you should start by knowing what link-local and site-local addresses are. Once those are cleared up then the others will fall into place. Quoting some snippets from IPv6 docs.

Link

Link-local address are supposed to be used for addressing nodes on a single link. Packets originating from or destined to a link-local address will not be forwarded by a router.

An example of this is the range 169.254/16. Which you may have seen before when devices are unable to obtain a valid address from DHCP.

Site

Site-local address are supposed to be used within a site. Routers will not forward any packet with site-local source or destination address outside the site.

This only applies to IPv6. There is no notion of site-local addresses in IPv4.

Host

A host address is something that will only exist within the host machine itself. For instance 127.0.0.1 is a host address commonly assigned to the loopback interface. The loopback interface has no external connectivity and so it's scope is confined to within that of the host machine.

Global

A global address is what you might currently consider a "normal" address. That is, a unicast address, which is visible on and routable across an external network.

  • 2
    What about the 192.168.0.0/16 block -- isn't that essentially site-local? – SamB May 4 '11 at 1:51
  • 1
    You should never see site being used, as these addresses were deprecated in 2004. – Michael Hampton Jan 6 '17 at 15:52
0

Find these comments in fib_semantics.c

 704  * Every prefix is assigned a "scope" value: "host" is local address,
 705  * "link" is direct route,
 706  * [ ... "site" ... "interior" ... ]
 707  * and "universe" is true gateway route with global meaning.
 708  *
 709  * Every prefix refers to a set of "nexthop"s (gw, oif),
 710  * where gw must have narrower scope. This recursion stops
 711  * when gw has LOCAL scope or if "nexthop" is declared ONLINK,
 712  * which means that gw is forced to be on link.
 ....
 719  * Normally it looks as following.
 720  *
 721  * {universe prefix}  -> (gw, oif) [scope link]
 722  *                |
 723  *                |-> {link prefix} -> (gw, oif) [scope local]
 724  *                                      |
 725  *                                      |-> {local prefix} (terminal node)
 726  */

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.