IPCONFIG(8)IPCONFIG(8)

NAME

ipconfig, rip, linklocal – Internet configuration and routing

SYNOPSIS

unhandled troff command .in

unhandled troff command .ti

ip/ipconfig [-6DGNOPdnpruX] [-b baud] [-c ctl] [-g gateway] [-h host] [-m mtu] [-o dhcp-opt] [-f dbfile] [-x netmtpt] [ type [ device ] ] [verb] [ local [ mask [ remote [ file-server [ auth ] ] ] ] ]

ip/rip [-bdr] [-x netmtpt]

ip/linklocal [ -t gwipv4 ] mac ...

DESCRIPTION

Ipconfig binds a device interface to a mounted IP stack (default /net) and configures the interface with a local address and optionally a mask, a remote address, a file server and an authentication server address. If no device is specified, the first ether device on the mounted IP stack is used. The addresses can be specified in the command line or obtained via DHCP. If DHCP is requested, it will also obtain the addresses of DNS servers, NTP servers, gateways, a Plan 9 file server, and a Plan 9 authentication server. Information from DHCP and IPv6 router advertisements is written to /net/ndb in the form of an ndb(8) entry unless the P flag has been specified.

Type may be ether, gbe, ppp, pkt, or loopback. The gbe type is equivalent to ether except that it allows jumbo packets (up to ~9KB). The pkt interface passes all IP packets to and from a user program. For ppp the device can be any byte stream device.

The verb (default add) determines the action performed. The usual verbs are:

add

if the device is not bound to the IP stack, bind it. Add the given local address, mask, and remote address to the interface. An interface may have multiple addresses.  

remove

remove the address from the device interface.  

unbind

unbind the device interface and all its addresses from the IP stack.  

The IPv6-specific verbs, which take different arguments, are:

add6 prefix pfx-len onlink auto validlt preflt

sets the named IPv6 parameters; see ip(3) for more detail.  

ra6 [ keyword value ] ...

sets IPv6 router advertisement parameter keyword’s value. See ip(3) for more detail. Setting recvra non-zero also forks a process to receive and process router advertisements. Setting sendra non-zero also enables IP routing on the interface, forks a process to send router advertisements, and if no recvra process is running, forks one.  

The options are:

6

if adding an address (the default action), add the IPv6 link-local address.  

b

the baud rate to use on a serial line when configuring PPP.  

c

write the control string ctl to the ethernet device control file before starting to configure it. May be repeated to specify multiple control writes.  

d

use DHCP to determine any unspecified configuration parameters.  

D

turn on debugging.  

g

the default gateway.  

G

use only generic DHCP and RA options. Without this option, ipconfig adds to requests a Vendor Class option with value plan9_$cputype and also requests vendor specific options 128 and 129 which we interpret as the Plan 9 file server and auth server. Replies to these options contain a list of IP addresses for possible file servers and auth servers.  

h

the hostname to add to DHCP requests. Some DHCP servers, such as the one used by Comcast, will not respond unless a correct hostname is in the request.  

m

the maximum IP packet size to use on this interface.  

n

determine parameters but don’t configure the interface.  

N

look in dbfile (default /lib/ndb/local) for the IP parameters for the specified local IP address or if local is omited and the device is an ethernet then all IP parameters associated with the MAC address. IPv6 addresses are added only if a IPv6 link-local address exists on the interface or the 6 flag has been given to automatically configure one.  

O

addresses specified on the command line override those obtained via DHCP. A command line address of 0 implies no override.  

p

write configuration information to /net/ndb.  

P

do not write configuration information to /net/ndb.  

r

by default, ipconfig exits after trying DHCP for 15 seconds with no answer. This option directs ipconfig instead to fork a background process that keeps trying forever.  

u

disable IPv6 duplicate discovery detection, which removes any existing ARP table entry for one of our IPv6 addresses before adding new ones.  

f

use the ndb database file dbfile.  

x

use the IP stack mounted at netmtpt instead of at /net.  

X

don’t fork a process to keep the DHCP lease alive.  

o

adds dhcpoption to the list of paramters requested of the DHCP server. The result will appear in /net/ndb should this be the first interface. The known options are:  

arptimeout, baddr, bflen, bootfile, clientid, cookie, discovermask, discoverrouter, dns, dom, dumpfile, etherencap, extpath, finger, homeagent, impress, ipaddr, ipforward, ipgw, ipmask, irc, lease, log, lpr, maxdatagram, maxmsg, message, mtu, name, netbiosdds, netbiosns, netbiosscope, netbiostype, ni, nisdomain, nisplus, nisplusdomain, nntp, nonlocal, ntp, overload, params, pathplateau, pathtimeout, policyfilter, pop3, rebindingtime, renewaltime, rl, rootpath, rs, serverid, smtp, st, staticroutes, stdar, subnetslocal, supplymask, swap, sys, tcpka, tcpkag, tcpttl, tftp, time, timeoff, trailerencap, ttl, type, vendorclass, www, xdispmanager, xfont

The options ipmask, ipgw, dns, sys, and ntp are always requested.  

If DHCP is requested, a process is forked off to renew the lease before it runs out. If the lease does run out, this process will remove any configured addresses from the interface.

Rip runs the routing protocol RIP. It listens for RIP packets on connected networks and updates the kernel routing tables. The options are:

b

broadcasts routing information onto the networks.  

n

gathers routing information but doesn’t write to the route table. This is useful with –d to debug a network.  

x

use the IP stack mounted at netmtpt instead of at /net.  

d

turn on (voluminous) debugging.  

Linklocal prints the IPv6 link-local address corresponding to the given mac address. Given -t, linklocal instead prints the 6to4 EUI-64-based IPv6 address corresponding to mac and 6to4 gateway gwipv4.

EXAMPLES

Configure Ethernet 0 as the primary IP interface. Get all addresses via DHCP. Start up a connection server and DNS resolver for this IP stack.

 

% bind -b '#l0' /net
% bind -a '#I0' /net
% ip/ipconfig
% ndb/cs
% ndb/dns -r

Add a second address to the stack.

 

% ip/ipconfig ether /net/ether0 add 12.1.1.2 255.255.255.0

At Bell Labs, our primary IP stack is always to the company’s internal firewall-protected network. The following creates an external IP stack to directly access the outside Internet. Note that the connection server uses a different set of ndb files. This prevents us from confusing inside and outside name/address bindings.

 

% bind -b '#l1' /net.alt
% bind -b '#I1' /net.alt
% ip/ipconfig -x /net.alt -g 204.178.31.1 ether /net.alt/ether1\\
	204.178.31.6 255.255.255.0
% ndb/cs -x /net.alt -f /lib/ndb/external
% ndb/dns -sx /net.alt -f /lib/ndb/external
% aux/listen -d /rc/bin/service.alt /net.alt/tcp

Configure the IPv6 link-local address automatically and listen for router announcements.

 

ip/ipconfig -6 
ip/ipconfig ra6 recvra 1

FILES

/sys/log/ipconfig

SOURCE

/sys/src/cmd/ip/ipconfig
/sys/src/cmd/ip/rip.c
/sys/src/cmd/ip/linklocal.c

SEE ALSO

ether(3), ip(3), loopback(3), ndb(6), 6in4(8), dhcpd(8), ppp(8)
/lib/rfc/rfc2373 for IPv6’s modified EUI-64