NAME

ip − network protocols over IP

SYNOPSIS

­bind -a #Ispec /net
 ­/net/ipifc
­/net/ipifc/clone
­/net/ipifc/stats
/net/ipifc/n
/net/ipifc/n/status
/net/ipifc/n/ctl
...
 ­/net/arp
­/net/log
­/net/ndb
­/net/iproute
­/net/ipselftab
 ­/net/esp
­/net/gre
­/net/icmp
­/net/il
­/net/ipmux
­/net/rudp
­/net/tcp
­/net/udp
 ­/net/tcp/clone
­/net/tcp/stats
/net/tcp/n
/net/tcp/n/data
/net/tcp/n/ctl
/net/tcp/n/local
/net/tcp/n/remote
/net/tcp/n/status
/net/tcp/n/listen
...

DESCRIPTION

The IP device provides the interface to Internet protocol stacks.  ­Spec is an integer from 0 to 15 identifying a stack.  Each stack is physically independent of all others: the only information transfer between them is via programs that mount multiple stacks.  Normally a system uses only one stack.  However multiple stacks can be used for debugging new IP networks or implementing firewalls or proxy services. 

All addresses used are 16-byte IPv6 addresses.  Though we currently implement only IPv4, the IPv6 format is intended to prepare the way for an IPv6 implementation.  IPv4 addresses are a subset of the IPv6 addresses and both standard ASCII formats are accepted.  In binary, all v4 addresses start with the 12 bytes:

00 00 00 00 00 00 00 00 00 00 ff ff

Configuring interfaces

Each stack may have multiple interfaces and each interface may have multiple addresses.  The ­/net/ipifc directory contains a ­clone file, a ­stats file, and numbered subdirectories for each physical interface. 

Opening the ­clone file reserves an interface.  The file descriptor returned from the open(2) will point to the control file, ctl, of the newly allocated interface.  Reading ­ctl returns a text string representing the number of the interface.  Writing ­ctl alters aspects of the interface.  The possible ­ctl messages are:

bind ether path
Treat the device mounted at ­path as an Ethernet medium carrying IP and ARP packets and associate it with this interface.  The kernel will dial(2) path!0x800 and path!0x806 and use the two connections for IP and ARP respectively.

­bind pkt
Treat this interface as a packet interface.  Assume a user program will read and write the ­data file to receive and transmit IP packets to the kernel.  This is used by programs such as ppp(8) to mediate IP packet transfer between the kernel and a PPP encoded device.

bind netdev path
Treat this interface as a packet interface. The kernel will open ­path and read and write the resulting file descriptor to receive and transmit IP packets. 

bind loopback
Treat this interface as a local loopback.  Anything written to it will be looped back.

­unbind
Disassociate the physical device from an IP interface.

add local mask remote mtu proxy
Add a local IP address to the interface.  The mask, remote, mtu, and ­proxy arguments are all optional.  The default mask is the class mask for the local address.  The default remote address is ­local ANDed with mask. The default mtu is 1514 for Ethernet and 4096 for packet media. Proxy, if specified, means that this machine should answer ARP requests for the remote address. Ppp(8) does this to make remote machines appear to be connected to the local Ethernet.

remove local mask
Remove a local IP address from an interface.

mtu n Set the maximum transfer unit for this device to n. The mtu is the maximum size of the packet including any medium-specific headers.

reassemble
Reassemble IP fragments before forwarding to this interface

iprouting n
Allow (nismissing or non-zero) or disallow (n is 0) forwarding packets between this interface and others. 

addmulti addr
Treat the multicast ­addr on this interface as a local address. 

remmulti addr
Remove the multicast address ­addr from this interface. 

Reading the interface’s ­status file returns information about the interface, one line for each local address on that interface.  The first line has 9 white-space-separated fields: device, mtu, local address, mask, remote or network address, packets in, packets out, input errors, output errors.  Each subsequent line contains all but the device and mtu.  See ­readipifc in ip(2).

Routing

The file ­iproute controls information about IP routing.  When read, it returns one line per routing entry.  Each line contains six white-space-separated fields: target address, target mask, address of next hop, flags, tag, and interface number.  The entry used for routing an IP packet is the one with the longest mask for which destination address ANDed with target mask equals the target address.  The one character flags are:

­4 IPv4 route

­6 IPv6 route

­i local interface

­b broadcast address

­u local unicast address

­m multicast route

­p point-to-point route

The tag is an arbitrary, up to 4 character, string.  It is normally used to indicate what routing protocol originated the route. 

Writing to ­/net/iproute changes the route table.  The messages are:

­flush
Remove all routes.

tag string
Associate the tag, string, with all subsequent routes added via this file descriptor.

add target mask nexthop
Add the route to the table.  If one already exists with the same target and mask, replace it.

remove target mask
Remove a route with a matching target and mask.

Address resolution

The file ­/net/arp controls information about address resolution.  The kernel automatically updates the ARP information for Ethernet interfaces.  When read, the file returns one line per address containing the type of medium, the status of the entry (OK, WAIT), the IP address, and the medium address.  Writing to ­/net/arp administers the ARP information.  The control messages are:

­flush
Remove all entries.

add type IP-addr Media-addr
Add an entry or replace an existing one for the same IP address.

del IP-addr
Delete an individual entry.

ARP entries do not time out.  The ARP table is a cache with an LRU replacement policy.  The IP stack listens for all ARP requests and, if the requester is in the table, the entry is updated.  Also, whenever a new address is configured onto an Ethernet, an ARP request is sent to help update the table on other systems. 

Currently, the only medium type is ether. 

Debugging and stack information

If any process is holding ­/net/log open, the IP stack queues debugging information to it.  This is intended primarily for debugging the IP stack.  The information provided is implementation-defined; see the source for details.  Generally, what is returned is error messages about bad packets. 

Writing to ­/net/log controls debugging.  The control messages are:

set arglist
­Arglist is a space-separated list of items for which to enable debugging.  The possible items are: ppp, ip, fs, tcp, il, icmp, udb, compress, ilmsg, gre, tcpmsg, udpmsg, ipmsg, and esp. 

clear arglist
­Arglist is a space-separated list of items for which to disable debugging. 

only addr
If ­addr is non-zero, restrict debugging to only those packets whose source or destination is that address. 

The file ­/net/ndb can be read or written by programs.  It is normally used by ipconfig(8) to leave configuration information for other programs such as ­dns and ­cs (see ndb(8)). ­/net/ndb may contain up tp 1024 bytes. 

The file ­/net/ipselftab is a read-only file containing all the IP addresses considered local.  Each line in the file contains three white-space-separated fields: IP address, usage count, and flags.  The usage count is the number of interfaces to which the address applies.  The flags are the same as for routing entries. 

Protocol directories

The ­ip device supports IP as well as several protocols that run over it: TCP, IL, UDP, GRE, ESP, ICMP, and RUDP.  TCP and UDP provide the standard Internet protocols for reliable stream and unreliable datagram communication.  IL provides a reliable datagram service for communication between Plan 9 machines.  GRE is a general encapsulation protocol.  ESP is the encapsulation protocol for IPSEC.  ICMP is IP’s catch-all control protocol used to send low level error messages and to implement ping(8). RUDP is a locally developed reliable datagram protocol based on UDP.

Each protocol is a subdirectory of the IP stack.  The top level directory of each protocol contains a ­clone file, a ­stats file, and subdirectories numbered from zero to the number of connections opened for this protocol. 

Opening the ­clone file reserves a connection.  The file descriptor returned from the open(2) will point to the control file, ctl, of the newly allocated connection.  Reading ­ctl returns a text string representing the number of the connection.  Connections may be used either to listen for incoming calls or to initiate calls to other machines. 

A connection is controlled by writing text strings to the associated ­ctl file.  After a connection has been established data may be read from and written to data.  A connection can be actively established using the ­connect message (see also dial(2)). A connection can be established passively by first using an ­announce message (see dial(2)) to bind to a local port and then opening the ­listen file (see dial(2)) to receive incoming calls.

The following control messages are supported:

connect ipaddress!port!r local
Establish a connection to the remote address ­ipaddress and remote port port. If ­local is specified, it is used as the local port number.  If ­local is not specified but ­!r is, the system will allocate a restricted port number (less than 1024) for the connection to allow communication with Unix ­login and ­exec services.  Otherwise a free port number starting at 5000 is chosen.  The connect fails if the combination of local and remote address/port pairs are already assigned to another port. 

announce X
­X is a decimal port number or ∗.  Set the local port number to ­X and accept calls to X. If ­X is ∗, accept calls for any port that no process has explicitly announced.  The local IP address cannot be set.  ­Announce fails if the connection is already announced or connected. 

bind X
­X is a decimal port number or ∗.  Set the local port number to X. This exists to support emulation of BSD sockets by the APE libraries (see pcc(1)) and is not otherwise used.

backlog n
Set the maximum number of unanswered (queued) incoming connections to an announced port to n. By default ­n is set to five.  If more than ­n connections are pending, further requests for a service will be rejected. 

ttl n Set the time to live IP field in outgoing packets to n.

tos n Set the service type IP field in outgoing packets to n.

Port numbers must be in the range 1 to 32767. 

Several files report the status of a connection.  The ­remote and ­local files contain the IP address and port number for the remote and local side of the connection.  The ­status file contains protocol-dependent information to help debug network connections.  On receiving and error or EOF reading or writing the ­data file, the ­err file contains the reason for error. 

A process may accept incoming connections by open(2)ing the ­listen file.  The ­open will block until a new connection request arrives.  Then ­open will return an open file descriptor which points to the control file of the newly accepted connection.  This procedure will accept all calls for the given protocol.  See dial(2).

TCP

TCP connections are reliable point-to-point byte streams; there are no message delimiters.  A connection is determined by the address and port numbers of the two ends.  TCP ­ctl files support the following additional messages:

­hangup
close down a TCP connection

keepalive n
turn on keep alive messages. N, if given, is the milliseconds between keepalives (default 30000).

UDP

UDP connections carry unreliable and unordered datagrams.  A read from ­data will return the next datagram, discarding anything that doesn’t fit in the read buffer.  A write is sent as a single datagram. 

By default, a UDP connection is a point-to-point link.  Either a ­connect establishes a local and remote address/port pair or after an announce, each datagram coming from a different remote address/port pair establishes a new incoming connection.  However, many-to-one semantics is also possible. 

If, after an announce, one of the following messages is written to ctl, then all messages sent to the announced port are received on the announced connection prefixed with the given structure. 

­headers4

typedef struct Udphdr4 Udphdr4;
struct Udphdr
{
ucharraddr[4];/∗ v4 remote address and port ∗/
ucharladdr[4];/∗ v4 local address and port ∗/
ucharrport[2];
ucharlport[2];
};

­headers

typedef struct Udphdr Udphdr;
struct Udphdr
{
ucharraddr[16];/∗ v6 remote address and port ∗/
ucharladdr[16];/∗ v6 local address and port ∗/
ucharrport[2];
ucharlport[2];
};

The only difference in the two is the type of address, IPv4 or IPv6.  Before a write, a user must prefix a similar structure to each message.  The system overrides the user specified local port with the announced one.  If the user specifies an address that isn’t a unicast address in /net/ipselftab, that too is overridden.  Since the prefixed structure is the same in read and write, it is relatively easy to write a server that responds to client requests by just copying new data into the message body and then writing back the same buffer that was written. 

RUDP

RUDP is a reliable datagram protocol based on UDP.  Packets are delivered in order.  RUDP does not support listen.  One must use either ­connect or ­announce followed immediately by ­headers or headers4. 

Unlike IL or TCP, the reboot of one end of a connection does not force a closing of the connection.  Communications will resume when the rebooted machine resumes talking.  Any unacknowledged packets queued before the reboot will be lost.  A reboot can be detected by reading the ­err file.  It will have the message

hangup address!port

where ­address and ­port are of the far side of the connection.  Retransmitting a datagram more than 10 times is treated like a reboot: all queued messages are dropped, an error is queued to the ­err file, and the conversation resumes. 

IL

IL is a reliable point-to-point datagram protocol.  Like TCP, IL delivers datagrams reliably and in order. Also like TCP, a connection is determined by the address and port numbers of the two ends.  Like UDP, each read and write transfers a single datagram. 

IL is efficient for LANs but doesn’t have the congestion control features needed for use through the Internet. 

GRE

GRE is the encapsulation protocol used by PPTP.  The kernel implements just enough of the protocol to multiplex it.  ­Announce is not allowed in GRE, only connect.  Since GRE has no port numbers, the port number in the connect is actually the 16 bit ­eproto field in the GRE header. 

Reads and writes transfer a GRE datagram starting at the GRE header.  On write, the kernel fills in the ­eproto field with the port number specified in the connect message. 

ESP

ESP is the Encapsulating Security Payload (RFC 1827).  It is used to set up an encrypted tunnel between machines.  Like GRE, ESP has no port numbers.  Instead, the port number in the ­connect message is the SPI (Security Association Identifier (sic)).  IP packets are written to and read from data.  The kernel encrypts any packets written to data, appends a MAC, and prefixes an ESP header before sending to the other end of the tunnel.  Received packets are checked against their MAC’s, decrypted, and queued for reading from data.  The control messages are:

esp alg secret
Encrypt with the algorithm, alg, using ­secret as the key.  Possible algorithms are: null, des_56_cbc, and rc4_128. 

ah alg secret
Use the hash algorithm, alg, with ­secret as the key for generating the MAC.  Possible algorithms are: null, hmac_sha1_96, and hmac_md5_96. 

­header
Turn on header mode.  Every buffer read from ­data starts with 4 unsued bytes, and the first 4 bytes of every buffer written to ­data are ignored. 

­noheader
Turn off header mode.

IP packet filter

The directory ­/net/ipmux looks like another protocol directory.  It is a packet filter built on top of IP.  Each numbered subdirectory represents a different filter.  The connect messages written to the ­ctl file describe the filter. Packets matching the filter can be read on the ­data file.  Packets written to the ­data file are routed to an interface and transmitted. 

A filter is a semicolon-separated list of relations.  Each relation describes a portion of a packet to match.  The possible relations are:

proto=n
the IP protocol number must be n.

dat[n:m]=expr
bytes ­n through ­m following the IP packet must match expr.

ifc=expr
the packet must have been received on an interface whose address matches expr.

src=expr
The source address in the packet must match expr.

dst=expr
The destination address in the packet must match expr.

­Expr is of the form:

­ value

value|value|... 

value&mask

value|value&mask

If a mask is given, the relevant field is first ANDed with the mask.  The result is compared against the value or list of values for a match.  In the case of ifc, dst, and ­src the value is a dot-formatted IP address and the mask is a dot-formatted IP mask.  In the case of dat, both value and mask are strings of 2 character hexadecimal digits representing 8 bit values. 

A packet is delivered to only one filter.  The filters are merged into a single comparison tree.  If two filters match the same packet, the following rules apply in order (here ’>’ means is preferred to):

1)protocol > data > source > destination > interface

2)lower data offsets > higher data offsets

3)longer matches > shorter matches

4)older > younger

So far this has just been used to implement a version of OSPF in Inferno. 

Statistics

The ­stats files are read only and contain statistics useful to network monitoring. 

Reading ­/net/ipifc/stats returns a list of 19 tagged and new line separated fields representing:

forwarding status (0 and 2 mean forwarding off, 1 means on)
default TTL
input packets
input header errors
input address errors
packets forwarded
input packets for unknown protocols
input packets discarded
input packets delivered to higher level protocols
output packets
output packets discarded
output packets with no route
timed out fragments in reassembly queue
requested reassemblies
successful reassemblies
failed reassemblies
successful fragmentations
unsuccessful fragmentations
fragments created

Reading ­/net/icmp/stats returns a list of 25 tagged and new line separated fields representing:

messages received
bad received messages
unreachables received
time exceededs received
input parameter problems received
source quenches received
redirects received
echo requests received
echo replies received
timestamps received
timestamp replies received
address mask requests received
address mask replies received
messages sent
transmission errors
unreachables sent
time exceededs sent
input parameter problems sent
source quenches sent
redirects sent
echo requests sent
echo replies sent
timestamps sent
timestamp replies sent
address mask requests sent
address mask replies sent

Reading ­/net/tcp/stats returns a list of 11 tagged and new line separated fields representing:

maximum number of connections
total outgoing calls
total incoming calls
number of established connections to be reset
number of currently established connections
segments received
segments sent
segments retransmitted
retransmit timeouts
bad received segments
transmission failures

Reading ­/net/udp/stats returns a list of 4 tagged and new line separated fields representing:

datagrams received
datagrams received for bad ports
malformed datagrams received
datagrams sent

Reading ­/net/il/stats returns a list of 7 tagged and new line separated fields representing:

checksum errors
header length errors
out of order messages
retransmitted messages
duplicate messages
duplicate bytes

Reading ­/net/gre/stats returns a list of 1 tagged number representing:

header length errors

SEE ALSO

listen(8), dial(2), ndb(6)

SOURCE

­/sys/src/9/ip

BUGS

­Ipmux has not been heavily used and should be considered experimental.  It may disappear in favor of a more traditional packet filter in the future. 

Plan 9  —  December 20, 2004