Skip to main content

iptables command man page

IPTABLES(8)   IPTABLES(8)

NAME
       iptables - administration tool for IPv4 packet filtering and NAT

SYNOPSIS
       iptables [-t table] -[AD] chain rule-specification [options]
       iptables [-t table] -I chain [rulenum] rule-specification [options]
       iptables [-t table] -R chain rulenum rule-specification [options]
       iptables [-t table] -D chain rulenum [options]
       iptables [-t table] -[LFZ] [chain] [options]
       iptables [-t table] -N chain
       iptables [-t table] -X [chain]
       iptables [-t table] -P chain target [options]
       iptables [-t table] -E old-chain-name new-chain-name
DESCRIPTION
       Iptables is  used  to  set  up, maintain, and inspect the tables of IP
       packet filter rules in the Linux kernel. Several different tables  may
       be  defined.   Each  table contains a number of built-in chains and may
       also contain user-defined chains.

       Each chain is a list of rules which can match a set of  packets.  Each
       rule specifies what to do with a packet that matches.  This is called a
       ‘target’, which may be a jump to a user-defined chain in the  same  ta-
       ble.

TARGETS
       A  firewall rule specifies criteria for a packet, and a target. If the
       packet does not match, the next rule in the chain is the examined;  if
       it does match, then the next rule is specified by the value of the tar-
       get, which can be the name of a user-defined chain or one of  the  spe-
       cial values ACCEPT, DROP, QUEUE, or RETURN.

       ACCEPT  means to let the packet through. DROP means to drop the packet
       on the floor.  QUEUE means to pass the packet to userspace.   (How  the
       packet can be received by a userspace process differs by the particular
       queue handler.  2.4.x and  2.6.x kernels  up  to  2.6.13  include  the
       ip_queue queue handler. Kernels 2.6.14 and later additionally include
       the nfnetlink_queue queue handler.  Packets with a target of QUEUE will
       be  sent to queue number ’0’ in this case. Please also see the NFQUEUE
       target as described  later  in  this  man  page.)   RETURN  means  stop
       traversing  this chain and  resume  at the next rule in the previous
       (calling) chain. If the end of a built-in chain is reached or  a  rule
       in a built-in chain with target RETURN is matched, the target specified
       by the chain policy determines the fate of the packet.

TABLES
       There are currently three independent tables (which tables are  present
       at  any time depends on the kernel configuration options and which mod-
       ules are present).

       -t, --table table
     This option specifies the packet matching table which  the  com-
     mand  should operate on. If the kernel is configured with auto-
     matic module loading, an attempt will be made to load the appro-
     priate module for that table if it is not already there.

     The tables are as follows:

     filter:
 This is  the default table (if no -t option is passed).  It
 contains the built-in chains INPUT (for packets destined  to
 local sockets),  FORWARD  (for packets being routed through
 the box), and OUTPUT (for locally-generated packets).

     nat:
 This table is consulted when a packet that  creates a  new
 connection  is encountered.  It consists of three built-ins:
 PREROUTING (for altering packets as soon as they  come  in),
 OUTPUT  (for altering locally-generated packets before rout-
 ing), and POSTROUTING (for  altering packets as  they  are
 about to go out).

     mangle:
 This table is used for specialized packet alteration. Until
 kernel 2.4.17 it had two built-in  chains:  PREROUTING  (for
 altering  incoming  packets  before routing) and OUTPUT (for
 altering locally-generated packets before  routing). Since
 kernel  2.4.18,  three  other built-in chains are also sup-
 ported: INPUT (for packets coming into the box itself), FOR-
 WARD (for  altering packets being routed through the box),
 and POSTROUTING (for altering packets as they are  about  to
 go out).

     raw:
 This table  is  used mainly for configuring exemptions from
 connection tracking in combination with the NOTRACK  target.
 It registers at the netfilter hooks with higher priority and
 is thus called before ip_conntrack, or any other IP  tables.
 It  provides the following built-in chains: PREROUTING (for
 packets arriving via any  network  interface)  OUTPUT  (for
 packets generated by local processes)

OPTIONS
       The options that are recognized by iptables can be divided into several
       different groups.

   COMMANDS
       These options specify the specific action to perform.  Only one of them
       can  be specified on the command line unless otherwise specified below.
       For all the long versions of the command and option names, you need  to
       use  only  enough  letters to ensure that iptables can differentiate it
       from all other options.

       -A, --append chain rule-specification
     Append one or more rules to the end of the selected chain.  When
     the  source  and/or  destination names resolve to more than one
     address, a rule will be added for each possible address combina-
     tion.

       -D, --delete chain rule-specification
       -D, --delete chain rulenum
     Delete one or more rules from the selected chain. There are two
     versions of this command: the rule can be specified as a number
     in  the  chain  (starting at 1 for the first rule) or a rule to
     match.

       -I, --insert chain [rulenum] rule-specification
     Insert one or more rules in the selected chain as the given rule
     number. So,  if the  rule  number is 1, the rule or rules are
     inserted at the head of the chain.  This is also the default  if
     no rule number is specified.

       -R, --replace chain rulenum rule-specification
     Replace a rule in the selected chain.  If the source and/or des-
     tination names resolve to multiple addresses, the command  will
     fail.  Rules are numbered starting at 1.

       -L, --list [chain]
     List  all rules in the selected chain.  If no chain is selected,
     all chains are listed.  As  every other iptables  command,  it
     applies  to  the specified table (filter is the default), so NAT
     rules get listed by
      iptables -t nat -n -L
     Please note that it is often used with the -n option,  in order
     to  avoid long reverse DNS lookups.  It is legal to specify the
     -Z (zero) option as well, in which case  the  chain(s)  will  be
     atomically  listed  and zeroed.  The exact output is affected by
     the other arguments given. The exact rules are suppressed until
     you use
      iptables -L -v

       -F, --flush [chain]
     Flush the selected chain (all the chains in the table if none is
     given).  This is equivalent to deleting all  the rules  one  by
     one.

       -Z, --zero [chain]
     Zero the packet and byte counters in all chains. It is legal to
     specify the -L, --list (list) option as well, to see  the coun-
     ters immediately before they are cleared. (See above.)

       -N, --new-chain chain
     Create  a new user-defined chain by the given name.  There must
     be no target of that name already.

       -X, --delete-chain [chain]
     Delete the optional user-defined chain specified. There must be
     no  references  to  the chain.  If there are, you must delete or
     replace the referring rules before the  chain  can  be  deleted.
     The  chain  must be  empty,  i.e. not contain any rules. If no
     argument is given, it will attempt to delete  every  non-builtin
     chain in the table.

       -P, --policy chain target
     Set  the policy for the chain to the given target.  See the sec-
     tion TARGETS for the legal targets.   Only  built-in  (non-user-
     defined) chains can  have  policies,  and neither built-in nor
     user-defined chains can be policy targets.

       -E, --rename-chain old-chain new-chain
     Rename the user specified chain to the user supplied name.  This
     is cosmetic, and has no effect on the structure of the table.

       -h     Help.   Give a (currently very brief) description of the command
     syntax.

   PARAMETERS
       The following parameters make up a rule specification (as used  in  the
       add, delete, insert, replace and append commands).

       -p, --protocol [!] protocol
     The  protocol of the rule or of the packet to check.  The speci-
     fied protocol can be one of tcp, udp, icmp, or all, or it can be
     a numeric  value, representing one of these protocols or a dif-
     ferent  one.   A protocol  name from  /etc/protocols  is  also
     allowed. A  "!" argument before the protocol inverts the test.
     The number zero is equivalent to all.  Protocol all  will match
     with  all protocols and is taken as default when this option is
     omitted.

       -s, --source [!] address[/mask]
     Source specification.  Address can be either a network  name,  a
     hostname (please note  that specifying any name to be resolved
     with a remote query such as DNS is a really bad idea), a network
     IP address (with /mask), or a plain IP address.  The mask can be
     either a network mask or a plain number, specifying  the number
     of 1’s at the left side of the network mask.  Thus, a mask of 24
     is equivalent to 255.255.255.0. A  "!" argument  before  the
     address specification inverts the sense of the address. The flag
     --src is an alias for this option.

       -d, --destination [!] address[/mask]
     Destination  specification.   See the  description  of  the  -s
     (source) flag  for  a  detailed description of the syntax.  The
     flag --dst is an alias for this option.

       -j, --jump target
     This specifies the target of the rule; i.e., what to do  if  the
     packet  matches  it.   The  target  can  be a user-defined chain
     (other than the one this rule is in), one of the special builtin
     targets  which  decide the fate of the packet immediately, or an
     extension (see EXTENSIONS below). If this option is omitted  in
     a rule (and -g is not used), then matching the rule will have no
     effect on the packet’s fate, but the counters on the  rule  will
     be incremented.

       -g, --goto chain
     This  specifies  that  the  processing should continue in a user
     specified chain. Unlike the --jump option return will  not  con-
     tinue  processing in  this  chain but instead in the chain that
     called us via --jump.

       -i, --in-interface [!] name
     Name of an interface via which a packet was received  (only  for
     packets  entering the  INPUT,  FORWARD  and PREROUTING chains).
     When the "!" argument is used before  the interface  name,  the
     sense  is inverted.   If the interface name ends in a "+", then
     any interface which begins with this name will match.   If  this
     option is omitted, any interface name will match.

       -o, --out-interface [!] name
     Name of an interface via which a packet is going to be sent (for
     packets entering the FORWARD, OUTPUT  and POSTROUTING  chains).
     When  the "!"  argument is used before the interface name, the
     sense is inverted.  If the interface name ends in a  "+",  then
     any  interface  which begins with this name will match.  If this
     option is omitted, any interface name will match.

       [!]  -f, --fragment
     This means that the rule only refers to second and further frag-
     ments  of fragmented packets.  Since there is no way to tell the
     source or destination ports of such a  packet  (or  ICMP type),
     such a packet will not match any rules which specify them.  When
     the "!" argument precedes the "-f"  flag, the  rule  will  only
     match head fragments, or unfragmented packets.

       -c, --set-counters PKTS BYTES
     This enables the administrator to initialize the packet and byte
     counters of a rule (during INSERT, APPEND, REPLACE  operations).

   OTHER OPTIONS
       The following additional options can be specified:

       -v, --verbose
     Verbose  output. This  option  makes the list command show the
     interface name, the rule options (if any), and  the  TOS masks.
     The  packet  and byte counters are also listed, with the suffix
     ’K’, ’M’ or ’G’ for 1000, 1,000,000 and 1,000,000,000  multipli-
     ers  respectively (but  see  the -x flag to change this).  For
     appending, insertion,  deletion  and  replacement,  this causes
     detailed information on the rule or rules to be printed.

       -n, --numeric
     Numeric  output. IP addresses and port numbers will be printed
     in numeric format.  By default, the program will try to  display
     them  as host names, network names, or services (whenever appli-
     cable).

       -x, --exact
     Expand numbers.  Display the exact value of the packet and  byte
     counters, instead  of only the rounded number in K’s (multiples
     of 1000) M’s (multiples of 1000K) or G’s (multiples  of  1000M).
     This option is only relevant for the -L command.

       --line-numbers
     When  listing  rules,  add line numbers to the beginning of each
     rule, corresponding to that rule’s position in the chain.

       --modprobe=command
     When adding or inserting rules into a chain, use command to load
     any necessary modules (targets, match extensions, etc).

MATCH EXTENSIONS
       iptables can use extended packet matching modules.  These are loaded in
       two ways: implicitly, when -p or --protocol is specified, or  with  the
       -m  or  --match options,  followed  by the matching module name; after
       these, various extra command line options become available,  depending
       on  the specific module.  You can specify multiple extended match mod-
       ules in one line, and you can use the -h or --help  options  after  the
       module has been specified to receive help specific to that module.

       The  following  are included in the base package, and most of these can
       be preceded by a !  to invert the sense of the match.

   account
       Account traffic for all hosts in defined network/netmask.

       Features:

       - long (one counter per protocol TCP/UDP/IMCP/Other) and short  statis-
       tics

       - one iptables rule for all hosts in network/netmask

       - loading/saving counters (by reading/writting to procfs entries)


       --aaddr network/netmask
     defines network/netmask for which make statistics.

       --aname name
     defines  name  of list  where statistics will be kept. If no is
     specified DEFAULT will be used.

       --ashort
     table will colect only short  statistics (only  total  counters
     without splitting it into protocols.

       Example usage:

       account traffic for/to 192.168.0.0/24 network into table mynetwork:

       #   iptables   -A   FORWARD   -m  account  --aname  mynetwork  --aaddr
       192.168.0.0/24

       account traffic for/to WWW serwer for 192.168.0.0/24 network into table
       mywwwserver:

       # iptables -A INPUT -p tcp --dport 80
-m account --aname mywwwserver --aaddr 192.168.0.0/24 --ashort

       # iptables -A OUTPUT -p tcp --sport 80
-m account --aname mywwwserver --aaddr 192.168.0.0/24 --ashort

       read counters:

       # cat   /proc/net/ipt_account/mynetwork  #   cat
       /proc/net/ipt_account/mywwwserver

       set counters:

       #    echo    "ip   = 192.168.0.1 packets_src =     0"     >
       /proc/net/ipt_account/mywwserver

       Webpage:
http://www.barbara.eu.org/~quaker/ipt_account/

   addrtype
       This module matches packets based on their address type. Address types
       are used within the kernel networking stack  and categorize  addresses
       into various groups.  The exact definition of that group depends on the
       specific layer three protocol.

       The following address types are possible:

       UNSPEC an unspecified address (i.e. 0.0.0.0) UNICAST an unicast address
     LOCAL  a local address BROADCAST a broadcast address ANYCAST an
     anycast packet MULTICAST a multicast address BLACKHOLE a black-
     hole  address UNREACHABLE an unreachable address PROHIBIT a pro-
     hibited address THROW FIXME NAT FIXME XRESOLVE FIXME

       --src-type type
     Matches if the source address is of given type

       --dst-type type
     Matches if the destination address is of given type

   ah
       This module matches the SPIs in Authentication header of IPsec packets.

       --ahspi [!] spi[:spi]

   childlevel
       This  is an  experimental module.  It matches on whether the packet is
       part of a master connection or one of its children  (or grandchildren,
       etc).   For  instance,  most packets are level 0.  FTP data transfer is
       level 1.

       --childlevel [!] level

   comment
       Allows you to add comments (up to 256 characters) to any rule.

       --comment comment

       Example:
     iptables -A INPUT -s 192.168.0.0/16 -m comment --comment "A pri-
     vatized IP block"

   condition
       This matches if a specific /proc filename is ’0’ or ’1’.

       --condition [!] filename
     Match  on boolean value stored in /proc/net/ipt_condition/file-
     name file

   connbytes
       Match by how many bytes or packets a connection (or  one  of  the  two
       flows  constituting the connection) have tranferred so far, or by aver-
       age bytes per packet.

       The counters are 64bit and are thus not expected to overflow ;)

       The primary use is to detect long-lived downloads and mark them to  be
       scheduled using a lower priority band in traffic control.

       The  transfered bytes  per  connection can  also  be  viewed  through
       /proc/net/ip_conntrack and accessed via ctnetlink

       [!] --connbytes from:[to]
     match packets  from  a  connection  whose packets/bytes/average
     packet size is more than FROM and less than TO bytes/packets. if
     TO is omitted only FROM check is done.  "!"  is  used  to match
     packets not falling in the range.

       --connbytes-dir [original|reply|both]
     which packets to consider

       --connbytes-mode [packets|bytes|avgpkt]
     whether  to  check the amount of packets, number of bytes trans-
     ferred or the average size (in bytes) of all packets received so
     far.  Note  that when "both" is used together with "avgpkt", and
     data is going (mainly) only in one direction (for example HTTP),
     the  average  packet  size will be about half of the actual data
     packets.

       Example:
     iptables .. -m connbytes --connbytes  10000:100000  --connbytes-
     dir both --connbytes-mode bytes ...

   connlimit
       Allows  you  to restrict  the  number of parallel TCP connections to a
       server per client IP address (or address block).

       [!] --connlimit-above n
     match if the number of existing tcp connections is (not) above n

       --connlimit-mask bits
     group hosts using mask

       Examples:

       # allow 2 telnet connections per client host
     iptables -p tcp --syn --dport 23 -m connlimit --connlimit-above
     2 -j REJECT

       # you can also match the other way around:
     iptables -p tcp --syn --dport 23 -m  connlimit  !  --connlimit-
     above 2 -j ACCEPT

       #  limit the nr of parallel http requests to 16 per class C sized net-
       work (24 bit netmask)
     iptables -p tcp --syn --dport 80 -m connlimit --connlimit-above
     16 --connlimit-mask 24 -j REJECT

   connmark
       This module matches the netfilter mark field associated with a  connec-
       tion (which can be set using the CONNMARK target below).

       --mark value[/mask]
     Matches  packets in connections with the given mark value (if a
     mask is specified, this is logically ANDed with the mark before
     the comparison).

   connrate
       This module matches the current transfer rate in a connection.

       --connrate [!] [from]:[to]
     Match  against the current connection transfer rate being within
     ’from’ and ’to’ bytes per second. When the "!" argument is  used
     before the range, the sense of the match is inverted.

   conntrack
       This  module,  when combined with connection tracking, allows access to
       more connection tracking information than  the  "state" match. (this
       module is present only if iptables was compiled under a kernel support-
       ing this feature)

       --ctstate state
     Where state is a comma separated list of the  connection states
     to  match.   Possible states are INVALID meaning that the packet
     is associated with no known connection, ESTABLISHED meaning that
     the  packet is associated with a connection which has seen pack-
     ets in both directions, NEW meaning that the packet has  started
     a new  connection,  or  otherwise  associated with a connection
     which has not seen packets in both directions, and RELATED mean-
     ing that the packet is starting a new connection, but is associ-
     ated with an existing connection, such as an FTP data  transfer,
     or  an ICMP error.  SNAT A virtual state, matching if the origi-
     nal source address differs from the reply destination.   DNAT  A
     virtual state, matching if the original destination differs from
     the reply source.

       --ctproto proto
     Protocol to match (by number or name)

       --ctorigsrc [!] address[/mask]
     Match against original source address

       --ctorigdst [!] address[/mask]
     Match against original destination address

       --ctreplsrc [!] address[/mask]
     Match against reply source address

       --ctrepldst [!] address[/mask]
     Match against reply destination address

       --ctstatus [NONE|EXPECTED|SEEN_REPLY|ASSURED][,...]
     Match against internal conntrack states

       --ctexpire time[:time]
     Match remaining lifetime in seconds against given value or range
     of values (inclusive)

   dccp
       --source-port,--sport [!] port[:port]

       --destination-port,--dport [!] port[:port]

       --dccp-types [!] mask
     Match  when  the DCCP packet type is one of ’mask’. ’mask’ is a
     comma-separated list of packet types.  Packet types are: REQUEST
     RESPONSE DATA  ACK  DATAACK  CLOSEREQ  CLOSE RESET SYNC SYNCACK
     INVALID.

       --dccp-option [!] number
     Match if DCP option set.

   dscp
       This module matches the 6 bit DSCP field within the TOS field in the IP
       header. DSCP has superseded TOS within the IETF.

       --dscp value
     Match against a numeric (decimal or hex) value [0-63].

       --dscp-class DiffServ Class
     Match  the  DiffServ class. This value may be any of the BE, EF,
     AFxx or CSx classes.   It will  then  be converted  into  it’s
     according numeric value.

   dstlimit
       This  module  allows you to limit the packet per second (pps) rate on a
       per destination IP or per destination port base.  As  opposed  to  the
       ‘limit’ match, every  destination  ip / destination port has it’s own
       limit.

       THIS MODULE IS DEPRECATED AND HAS BEEN REPLACED BY ‘‘hashlimit’’

       --dstlimit avg
     Maximum average match rate (packets per second  unless  followed
     by /sec /minute /hour /day postfixes).

       --dstlimit-mode mode
     The limiting hashmode.  Is the specified limit per dstip, dstip-
     dstport tuple,  srcip-dstip  tuple,  or  per  srcipdstip-dstport
     tuple.

       --dstlimit-name name
     Name for /proc/net/ipt_dstlimit/* file entry

       [--dstlimit-burst burst]
     Number of packets to match in a burst.  Default: 5

       [--dstlimit-htable-size size]
     Number of buckets in the hashtable

       [--dstlimit-htable-max max]
     Maximum number of entries in the hashtable

       [--dstlimit-htable-gcinterval interval]
     Interval between garbage  collection runs of the hashtable (in
     miliseconds).  Default is 1000 (1 second).

       [--dstlimit-htable-expire time
     After which time are idle entries expired  from hashtable  (in
     miliseconds)?  Default is 10000 (10 seconds).

   ecn
       This  allows you to match the ECN bits of the IPv4 and TCP header.  ECN
       is the Explicit Congestion  Notification  mechanism  as specified  in
       RFC3168

       --ecn-tcp-cwr
     This matches if the TCP ECN CWR (Congestion Window Received) bit
     is set.

       --ecn-tcp-ece
     This matches if the TCP ECN ECE (ECN Echo) bit is set.

       --ecn-ip-ect num
     This matches a particular IPv4 ECT (ECN-Capable Transport).  You
     have to specify a number between ‘0’ and ‘3’.

   esp
       This module matches the SPIs in ESP header of IPsec packets.

       --espspi [!] spi[:spi]

   fuzzy
       This  module  matches  a rate  limit based on a fuzzy logic controller
       [FLC]

       --lower-limit number
     Specifies the lower limit (in packets per second).

       --upper-limit number
     Specifies the upper limit (in packets per second).

   hashlimit
       This patch adds a new match called ’hashlimit’. The idea  is  to  have
       something  like ’limit’,  but  either  per destination-ip or per (des-
       tip,destport) tuple.

       It gives you the ability to express

      ’1000 packets per second for every host in 192.168.0.0/16’

      ’100 packets per second for every service of 192.168.1.1’

       with a single iptables rule.

       --hashlimit rate
     A rate just like the limit match

       --hashlimit-burst num
     Burst value, just like limit match

       --hashlimit-mode destip | destip-destport
     Limit per IP or per port

       --hashlimit-name foo
     The name for the /proc/net/ipt_hashlimit/foo entry

       --hashlimit-htable-size num
     The number of buckets of the hash table

       --hashlimit-htable-max num
     Maximum entries in the hash

       --hashlimit-htable-expire num
     After how many miliseconds do hash entries expire

       --hashlimit-htable-gcinterval num
     How many miliseconds between garbage collection intervals

   helper
       This module matches packets related to a specific conntrack-helper.

       --helper string
     Matches packets related to the specified conntrack-helper.

     string can be "ftp" for packets  related to  a  ftp-session  on
     default  port.  For other ports append -portnr to the value, ie.
     "ftp-2121".

     Same rules apply for other conntrack-helpers.

   icmp
       This extension is loaded if ‘--protocol icmp’ is specified.   It  pro-
       vides the following option:

       --icmp-type [!] typename
     This  allows  specification  of  the  ICMP  type, which can be a
     numeric ICMP type, or one of the ICMP type names shown  by  the
     command
      iptables -p icmp -h

   iprange
       This matches on a given arbitrary range of IPv4 addresses

       [!]--src-range ip-ip
     Match source IP in the specified range.

       [!]--dst-range ip-ip
     Match destination IP in the specified range.

   ipv4options
       Match  on IPv4 header options like source routing, record route, times-
       tamp and router-alert.

       --ssrr To match packets with the flag strict source routing.

       --lsrr To match packets with the flag loose source routing.

       --no-srr
     To match packets with no flag for source routing.

       [!] --rr
     To match packets with the RR flag.

       [!] --ts
     To match packets with the TS flag.

       [!] --ra
     To match packets with the router-alert option.

       [!] --any-opt
     To match a packet with at least one IP option, or no  IP option
     at all if ! is chosen.

       Examples:

       $ iptables -A input -m ipv4options --rr -j DROP
     will drop packets with the record-route flag.

       $ iptables -A input -m ipv4options --ts -j DROP
     will drop packets with the timestamp flag.

   length
       This  module matches the length of a packet against a specific value or
       range of values.

       --length [!] length[:length]

   limit
       This module matches at a limited rate using a token bucket  filter.   A
       rule  using  this  extension  will  match  until this limit is reached
       (unless the ‘!’ flag is used).  It can be used in combination with  the
       LOG target to give limited logging, for example.

       --limit rate
     Maximum  average matching  rate: specified as a number, with an
     optional ‘/second’, ‘/minute’, ‘/hour’, or  ‘/day’  suffix;  the
     default is 3/hour.

       --limit-burst number
     Maximum  initial number of  packets to match: this number gets
     recharged by one every time the limit  specified above  is  not
     reached, up to this number; the default is 5.

   mac
       --mac-source [!] address
     Match   source   MAC   address. It   must   be of  the  form
     XX:XX:XX:XX:XX:XX.  Note that this only makes sense for  packets
     coming from an Ethernet device and entering the PREROUTING, FOR-
     WARD or INPUT chains.

   mark
       This module matches the netfilter mark field associated with  a packet
       (which can be set using the MARK target below).

       --mark value[/mask]
     Matches packets with the given unsigned mark value (if a mask is
     specified, this is logically ANDed with the mask before the com-
     parison).

   mport
       This  module  matches  a set of source or destination ports.  Up to 15
       ports can be specified. It can only be used in conjunction with -p tcp
       or -p udp.

       --source-ports port[,port[,port...]]
     Match  if the  source port is one of the given ports.  The flag
     --sports is a convenient alias for this option.

       --destination-ports port[,port[,port...]]
     Match if the destination port is one of the  given  ports.   The
     flag --dports is a convenient alias for this option.

       --ports port[,port[,port...]]
     Match  if the both the source and destination ports are equal to
     each other and to one of the given ports.

   multiport
       This module matches a set of source or destination  ports.   Up to  15
       ports  can be specified. A port range (port:port) counts as two ports.
       It can only be used in conjunction with -p tcp or -p udp.

       --source-ports [!] port[,port[,port:port...]]
     Match if the source port is one of the given  ports.   The  flag
     --sports is a convenient alias for this option.

       --destination-ports [!] port[,port[,port:port...]]
     Match  if the  destination port is one of the given ports.  The
     flag --dports is a convenient alias for this option.

       --ports [!] port[,port[,port:port...]]
     Match if either the source or destination ports are equal to one
     of the given ports.

   nth
       This module matches every ‘n’th packet

       --every value
     Match every ‘value’ packet

       [--counter num]
     Use internal counter number ‘num’.  Default is ‘0’.

       [--start num]
     Initialize the counter at the number ‘num’ insetad of ‘0’.  Most
     between ‘0’ and ‘value’-1.

       [--packet num]
     Match on ‘num’ packet.  Most be between ‘0’ and ‘value’-1.

   osf
       The idea of passive OS fingerprint matching exists  for quite  a  long
       time,  but  was created as extension fo OpenBSD pf only some weeks ago.
       Original idea  was  lurked  in some  OpenBSD  mailing list   (thanks
       grange@open...) and  than  adopted for Linux netfilter in form of this
       code.

       Original  fingerprint table was   created by   Michal   Zalewski
       <lcamtuf@coredump.cx>.

       This module compares some data(WS, MSS, options and it’s order, ttl, df
       and others) from first SYN packet (actually from packets with  SYN  bit
       set) with dynamically loaded OS fingerprints.

       --log 1/0
     If  present,  OSF will log determined genres even if they don’t
     match desired one.    0 - log all determined entries, 1  -  only
     first one.

     In syslog you find something like this:

     ipt_osf: Windows [2000:SP3:Windows  XP Pro  SP1,  2000 SP3]:
     11.22.33.55:4024 -> 11.22.33.44:139

     ipt_osf:      Unknown:       16384:106:1:48:020405B401010402
     44.33.22.11:1239 -> 11.22.33.44:80

       --smart
     if  present, OSF will use some smartness to determine remote OS.
     OSF will use initial TTL only if source of connection is in  our
     local network.

       --netlink
     If  present,  OSF will  log  all events also  through netlink
     NETLINK_NFLOG groupt 1.

       --genre [!] string
     Match a OS genre by passive fingerprinting

       Example:

       #iptables -I INPUT -j ACCEPT -p tcp  -m osf  --genre  Linux  --log  1
       --smart

       NOTE: -p tcp is obviously required as it is a TCP match.

       Fingerprints  can  be  loaded  and  read through /proc/sys/net/ipv4/osf
       file.  One can flush all fingerprints with following command:

     echo -en FLUSH > /proc/sys/net/ipv4/osf

       Only one fingerprint per open/write/close.

       Fingerprints  can  be   downloaded   from   http://www.openbsd.org/cgi-
       bin/cvsweb/src/etc/pf.os

   owner
       This  module  attempts  to  match various characteristics of the packet
       creator, for locally-generated packets. It is only valid in the OUTPUT
       chain,  and  even  this some packets (such as ICMP ping responses) may
       have no owner, and hence never match.

       --uid-owner userid
     Matches if the packet was created by a process  with  the given
     effective user id.

       --gid-owner groupid
     Matches  if  the packet was created by a process with the given
     effective group id.

       --pid-owner processid
     Matches if the packet was created by a process  with  the given
     process id.

       --sid-owner sessionid
     Matches if the packet was created by a process in the given ses-
     sion group.

       --cmd-owner name
     Matches if the packet was created by a process  with  the given
     command name.  (this option is present only if iptables was com-
     piled under a kernel supporting this feature)

       NOTE: pid, sid and command matching are broken on SMP

   physdev
       This module matches  on the  bridge  port  input  and  output  devices
       enslaved to  a bridge device. This module is a part of the infrastruc-
       ture that enables a transparent bridging IP firewall and is only useful
       for kernel versions above version 2.5.44.

       --physdev-in [!] name
     Name  of a bridge port via which a packet is received (only for
     packets entering the INPUT, FORWARD and PREROUTING  chains).  If
     the  interface  name  ends  in  a "+", then any interface which
     begins with this name will match. If the packet didn’t arrive
     through  a  bridge  device, this packet won’t match this option,
     unless ’!’ is used.

       --physdev-out [!] name
     Name of a bridge port via which a packet is  going  to  be  sent
     (for  packets  entering  the  FORWARD,  OUTPUT  and  POSTROUTING
     chains). If the interface name ends in a "+", then  any inter-
     face  which  begins  with this name will match. Note that in the
     nat and mangle OUTPUT chains one cannot match on the bridge out-
     put  port,  however  one can in the filter OUTPUT chain. If the
     packet won’t leave by a bridge device or it is yet unknown  what
     the  output  device  will be,  then the packet won’t match this
     option, unless

       [!] --physdev-is-in
     Matches if the packet has entered through a bridge interface.

       [!] --physdev-is-out
     Matches if the packet will leave through a bridge interface.

       [!] --physdev-is-bridged
     Matches if the packet is being  bridged  and  therefore  is  not
     being  routed.  This is only useful in the FORWARD and POSTROUT-
     ING chains.

   pkttype
       This module matches the link-layer packet type.

       --pkt-type [unicast|broadcast|multicast]

   policy
       This modules matches the policy used by IPsec for handling a packet.

       --dir in|out
     Used to select whether to match the policy used  for  decapsula-
     tion  or the policy that will be used for encapsulation. in is
     valid in the PREROUTING, INPUT and FORWARD chains, out is valid
     in the POSTROUTING, OUTPUT and FORWARD chains.

       --pol none|ipsec
     Matches if the packet is subject to IPsec processing.

       --strict
     Selects  whether to match the exact policy or match if any rule
     of the policy matches the given policy.

       --reqid id
     Matches the reqid of the policy rule. The reqid can be specified
     with setkey(8) using unique:id as level.

       --spi spi
     Matches the SPI of the SA.

       --proto ah|esp|ipcomp
     Matches the encapsulation protocol.

       --mode tunnel|transport
     Matches the encapsulation mode.

       --tunnel-src addr[/mask]
     Matches  the source end-point address of a tunnel mode SA.  Only
     valid with --mode tunnel.

       --tunnel-dst addr[/mask]
     Matches the destination end-point address of a tunnel  mode  SA.
     Only valid with --mode tunnel.

       --next Start  the next element in the policy specification. Can only be
     used with --strict

   psd
       Attempt to detect TCP and UDP port scans. This match was derived  from
       Solar Designer’s scanlogd.

       --psd-weight-threshold threshold
     Total weight of the latest TCP/UDP packets with different desti-
     nation ports coming from the same host to be  treated  as  port
     scan sequence.

       --psd-delay-threshold delay
     Delay  (in  hundredths of second) for the packets with different
     destination ports coming from the same host  to  be  treated  as
     possible port scan subsequence.

       --psd-lo-ports-weight weight
     Weight  of the packet with privileged (<=1024) destination port.

       --psd-hi-ports-weight weight
     Weight of the packet with non-priviliged destination port.

   quota
       Implements network quotas by decrementing  a  byte  counter  with  each
       packet.

       --quota bytes
     The quota in bytes.

       KNOWN BUGS: this does not work on SMP systems.

   random
       This module randomly matches a certain percentage of all packets.

       --average percent
     Matches  the given percentage.  If omitted, a probability of 50%
     is set.

   realm
       This matches the routing realm. Routing realms are  used  in  complex
       routing setups involving dynamic routing protocols like BGP.

       --realm [!]value[/mask]
     Matches a given realm number (and optionally mask).

   recent
       Allows  you to dynamically create a list of IP addresses and then match
       against that list in a few different ways.

       For example, you can create a ‘badguy’ list out of people attempting to
       connect to  port 139 on your firewall and then DROP all future packets
       from them without considering them.

       --name name
     Specify the list to use for the commands. If no  name  is given
     then ’DEFAULT’ will be used.

       [!] --set
     This  will  add the source address of the packet to the list. If
     the source address is already in the list, this will update  the
     existing entry. This will always return success (or failure if
     ‘!’ is passed in).

       [!] --rcheck
     Check if the source address of the packet is  currently  in  the
     list.

       [!] --update
     Like  --rcheck,  except it will update the "last seen" timestamp
     if it matches.

       [!] --remove
     Check if the source address of the packet is  currently  in  the
     list  and if  so that address will be removed from the list and
     the rule will return true. If the address is not found, false is
     returned.

       [!] --seconds seconds
     This  option must be used in conjunction with one of --rcheck or
     --update. When used, this will narrow the match to  only happen
     when  the address  is  in the list and was seen within the last
     given number of seconds.

       [!] --hitcount hits
     This option must be used in conjunction with one of --rcheck  or
     --update. When  used, this will narrow the match to only happen
     when the address is in the list and packets  had been  received
     greater  than  or equal to the given value. This option may be
     used along with --seconds to  create  an even  narrower match
     requiring a certain number of hits within a specific time frame.

       --rttl This option must be used in conjunction with one of --rcheck  or
     --update. When  used, this will narrow the match to only happen
     when the address is in the list  and  the TTL  of  the  current
     packet matches that of the packet which hit the --set rule. This
     may be useful if you have problems  with people faking their
     source  address in order to DoS you via this module by disallow-
     ing others access to your site by sending bogus packets to  you.

       Examples:

     # iptables -A FORWARD -m recent --name badguy --rcheck --seconds
     60 -j DROP

     # iptables -A FORWARD -p tcp  -i eth0  --dport  139  -m recent
     --name badguy --set -j DROP

       Official website  (http://snowman.net/projects/ipt_recent/)  also  has
       some examples of usage.

       /proc/net/ipt_recent/* are the current lists of addresses and  informa-
       tion about each entry of each list.

       Each  file in /proc/net/ipt_recent/ can be read from to see the current
       list or written two using the following commands to modify the list:

       echo xx.xx.xx.xx > /proc/net/ipt_recent/DEFAULT
     to Add to the DEFAULT list

       echo -xx.xx.xx.xx > /proc/net/ipt_recent/DEFAULT
     to Remove from the DEFAULT list

       echo clear > /proc/net/ipt_recent/DEFAULT
     to empty the DEFAULT list.

       The module itself accepts parameters, defaults shown:

       ip_list_tot=100
     Number of addresses remembered per table

       ip_pkt_list_tot=20
     Number of packets per address remembered

       ip_list_hash_size=0
     Hash table size. 0 means to calculate it based  on  ip_list_tot,
     default: 512

       ip_list_perms=0644
     Permissions for /proc/net/ipt_recent/* files

       debug=0
     Set to 1 to get lots of debugging info

   sctp
       --source-port,--sport [!] port[:port]

       --destination-port,--dport [!] port[:port]

       --chunk-types [!] all|any|only chunktype[:flags] [...]
     The  flag letter in  upper  case indicates that the flag is to
     match if set, in the lower case indicates to match if unset.

     Chunk types: DATA INIT  INIT_ACK SACK  HEARTBEAT HEARTBEAT_ACK
     ABORT   SHUTDOWN SHUTDOWN_ACK ERROR  COOKIE_ECHO  COOKIE_ACK
     ECN_ECNE ECN_CWR SHUTDOWN_COMPLETE ASCONF ASCONF_ACK

     chunk type    available flags
     DATA    U B E u b e
     ABORT    T t
     SHUTDOWN_COMPLETE    T t

     (lowercase means flag should be "off", uppercase means "on")

       Examples:

       iptables -A INPUT -p sctp --dport 80 -j DROP

       iptables -A INPUT -p sctp --chunk-types any DATA,INIT -j DROP

       iptables -A INPUT -p sctp --chunk-types any DATA:Be -j ACCEPT

   set
       This modules macthes IP sets which can be defined by ipset(8).

       --set setname flag[,flag...]
     where flags are src and/or dst and there can be no more than six
     of them. Hence the command
      iptables -A FORWARD -m set --set test src,dst
     will match packets, for which (depending on the type of the set)
     the source address or port number of the packet can be found  in
     the specified set. If there is a binding belonging to the mached
     set element or there is a default binding for  the  given  set,
     then  the rule  will  match  the packet only  if additionally
     (depending on the type of the set) the  destination  address  or
     port  number  of the packet can be found in the set according to
     the binding.

   state
       This module, when combined with connection tracking, allows  access  to
       the connection tracking state for this packet.

       --state state
     Where  state  is a comma separated list of the connection states
     to match. Possible states are INVALID meaning that  the packet
     could  not  be identified for some reason which includes running
     out of memory and ICMP errors  which  don’t  correspond  to  any
     known connection, ESTABLISHED meaning that the packet is associ-
     ated with a connection which has seen  packets  in  both direc-
     tions, NEW meaning that the packet has started a new connection,
     or otherwise associated with a connection which has  not  seen
     packets  in both directions, and RELATED meaning that the packet
     is starting a new connection, but is associated with an existing
     connection, such as an FTP data transfer, or an ICMP error.

   string
       This  modules  matches  a  given string by using some pattern matching
       strategy. It requires a linux kernel >= 2.6.14.

       --algo  bm|kmp
     Select the pattern matching strategy. (bm = Boyer-Moore, kmp  =
     Knuth-Pratt-Morris)

       --from offset
     Set the offset from which it starts looking for any matching. If
     not passed, default is 0.

       --to offset
     Set the offset from which it starts looking for any matching. If
     not passed, default is the packet size.

       --string pattern
     Matches  the  given  pattern.   --hex-string pattern Matches the
     given pattern in hex notation.

   tcp
       These extensions are loaded if ‘--protocol tcp’ is specified.  It  pro-
       vides the following options:

       --source-port [!] port[:port]
     Source  port  or port range specification. This can either be a
     service name or a port number. An inclusive range can  also  be
     specified,  using the  format  port:port.  If the first port is
     omitted, "0" is assumed; if the  last  is omitted,  "65535"  is
     assumed. If the second port greater then the first they will be
     swapped. The flag  --sport  is  a  convenient  alias  for  this
     option.

       --destination-port [!] port[:port]
     Destination  port or port range specification.  The flag --dport
     is a convenient alias for this option.

       --tcp-flags [!] mask comp
     Match when the TCP flags are as specified.  The  first  argument
     is  the  flags which we should examine, written as a comma-sepa-
     rated list, and the second argument is a comma-separated list of
     flags which must be set. Flags are: SYN ACK FIN RST URG PSH ALL
     NONE.  Hence the command
      iptables -A FORWARD -p tcp --tcp-flags SYN,ACK,FIN,RST SYN
     will only match packets with the SYN flag set, and the ACK,  FIN
     and RST flags unset.

       [!] --syn
     Only  match TCP packets with the SYN bit set and the ACK,RST and
     FIN bits cleared. Such packets are used to request TCP  connec-
     tion initiation; for example, blocking such packets coming in an
     interface will prevent incoming TCP  connections, but  outgoing
     TCP  connections will be unaffected.  It is equivalent to --tcp-
     flags  SYN,RST,ACK,FIN  SYN.   If the  "!"  flag precedes  the
     "--syn", the sense of the option is inverted.

       --tcp-option [!] number
     Match if TCP option set.

       --mss value[:value]
     Match  TCP  SYN  or SYN/ACK packets with the specified MSS value
     (or range), which control the maximum packet size for that  con-
     nection.

   tcpmss
       This  matches  the  TCP MSS  (maximum  segment size) field of the TCP
       header. You can only use this on TCP SYN or SYN/ACK packets, since the
       MSS  is only negotiated during the TCP handshake at connection startup
       time.

       [!] --mss value[:value]"
     Match a given TCP MSS value or range.

   time
       This matches if the packet arrival time/date is within a given range.
       All options are facultative.

--timestart value
     Match  only  if  it is after ‘value’ (Inclusive, format: HH:MM ;
     default 00:00).

       --timestop  value
     Match only if it is before ‘value’ (Inclusive, format:  HH:MM  ;
     default 23:59).

       --days listofdays
     Match  only  if  today  is  one  of  the given  days.  (format:
     Mon,Tue,Wed,Thu,Fri,Sat,Sun ; default everyday)

       --datestart date
     Match  only  if  it  is after ‘date’ (Inclusive,   format:
     YYYY[:MM[:DD[:hh[:mm[:ss]]]]]  ; h,m,s start from 0 ; default to
     1970)

       --datestop date
     Match  only  if  it  is  before ‘date’ (Inclusive,   format:
     YYYY[:MM[:DD[:hh[:mm[:ss]]]]]  ; h,m,s start from 0 ; default to
     2037)

   tos
       This module matches the 8 bits of Type  of  Service  field  in  the  IP
       header (ie. including the precedence bits).

       --tos tos
     The argument is either a standard name, (use
      iptables -m tos -h
     to see the list), or a numeric value to match.

   ttl
       This module matches the time to live field in the IP header.

       --ttl-eq ttl
     Matches the given TTL value.

       --ttl-gt ttl
     Matches if TTL is greater than the given TTL value.

       --ttl-lt ttl
     Matches if TTL is less than the given TTL value.

   u32
       U32  allows  you to extract quantities of up to 4 bytes from a packet,
       AND them with specified masks, shift them by specified amounts and test
       whether the results are in any of a set of specified ranges.  The spec-
       ification of what to extract is general enough  to  skip over  headers
       with lengths stored in the packet, as in IP or TCP header lengths.

       Details and examples are in the kernel module source.

   udp
       These  extensions are loaded if ‘--protocol udp’ is specified.  It pro-
       vides the following options:

       --source-port [!] port[:port]
     Source port or port range specification. See the description of
     the --source-port option of the TCP extension for details.

       --destination-port [!] port[:port]
     Destination  port or port range specification.  See the descrip-
     tion of the --destination-port option of the TCP extension  for
     details.

   unclean
       This  module takes no options, but attempts to match packets which seem
       malformed or unusual.  This is regarded as experimental.

TARGET EXTENSIONS
       iptables can use extended target modules: the following are included in
       the standard distribution.

   BALANCE
       This  allows  you to DNAT connections in a round-robin way over a given
       range of destination addresses.

       --to-destination ipaddr-ipaddr
     Address range to round-robin over.

   CLASSIFY
       This module allows you to set the skb->priority value (and  thus clas-
       sify the packet into a specific CBQ class).

       --set-class MAJOR:MINOR
     Set the major and minor class value.

   CLUSTERIP
       This  module  allows  you  to  configure a simple cluster of nodes that
       share a certain IP and MAC address without an explicit load balancer in
       front  of  them.  Connections  are  statically distributed between the
       nodes in this cluster.

       --new  Create a new ClusterIP.  You always have to  set this  on  the
     first rule for a given ClusterIP.

       --hashmode mode
     Specify  the hashing mode.  Has to be one of sourceip, sourceip-
     sourceport, sourceip-sourceport-destport

       --clustermac mac
     Specify the ClusterIP MAC address.  Has to be a link-layer  mul-
     ticast address

       --total-nodes num
     Number of total nodes within this cluster.

       --local-node num
     Local node number within this cluster.

       --hash-init rnd
     Specify the random seed used for hash initialization.

   CONNMARK
       This module sets the netfilter mark value associated with a connection

       --set-mark mark[/mask]
     Set connection mark. If a mask is specified then only those bits
     set in the mask is modified.

       --save-mark [--mask mask]
     Copy the netfilter packet mark value to the connection mark.  If
     a mask is specified then only those bits are copied.

       --restore-mark [--mask mask]
     Copy the connection mark value to the packet. If a mask is spec-
     ified then only those bits are copied. This is only valid in the
     mangle table.

   DNAT
       This  target is only valid in the nat table, in the PREROUTING and OUT-
       PUT chains, and user-defined chains which are only  called  from those
       chains. It specifies that the destination address of the packet should
       be modified (and all future packets in this  connection will  also  be
       mangled),  and rules should cease being examined.  It takes one type of
       option:

       --to-destination ipaddr[-ipaddr][:port-port]
     which can specify a single new destination IP address, an inclu-
     sive  range of IP addresses, and optionally, a port range (which
     is only valid if the rule also specifies -p tcp or -p udp).   If
     no port range is specified, then the destination port will never
     be modified.

     In Kernels up to 2.6.10 you  can add  several  --to-destination
     options. For those kernels, if you specify more than one desti-
     nation address, either via an address range  or  multiple --to-
     destination  options, a simple round-robin (one after another in
     cycle) load  balancing  takes  place  between  these  addresses.
     Later  Kernels  (>= 2.6.11-rc1) don’t have the ability to NAT to
     multiple ranges anymore.


   DSCP
       This target allows to alter the value of the DSCP bits within  the  TOS
       header  of  the IPv4 packet.  As this manipulates a packet, it can only
       be used in the mangle table.

       --set-dscp value
     Set the DSCP field to a numerical value (can be decimal or hex)

       --set-dscp-class class
     Set the DSCP field to a DiffServ class.

   ECN
       This target allows to selectively work around known ECN blackholes.  It
       can only be used in the mangle table.

       --ecn-tcp-remove
     Remove all ECN bits from the TCP header. Of course, it can only
     be used in conjunction with -p tcp.

   IPMARK
       Allows you to mark a received packet basing on its IP address. This can
       replace many  mangle/mark  entries  with only one, if you use firewall
       based classifier.

       This target is to be used inside the mangle table, in  the  PREROUTING,
       POSTROUTING or FORWARD hooks.

       --addr src/dst
     Use source or destination IP address.

       --and-mask mask
     Perform bitwise ‘and’ on the IP address and this mask.

       --or-mask mask
     Perform bitwise ‘or’ on the IP address and this mask.

       The  order  of  IP  address  bytes  is reversed to meet "human order of
       bytes": 192.168.0.1 is 0xc0a80001. At first the ‘and’ operation is per-
       formed, then ‘or’.

       Examples:

       We create a queue for each user, the queue number is adequate to the IP
       address of the user, e.g.: all packets going  to/from  192.168.5.2  are
       directed to 1:0502 queue, 192.168.5.12 -> 1:050c etc.

       We have one classifier rule:

     tc filter add dev eth3 parent 1:0 protocol ip fw

       Earlier we had many rules just like below:

     iptables -t mangle -A POSTROUTING -o eth3 -d 192.168.5.2 -j MARK
     --set-mark 0x10502

     iptables -t mangle -A POSTROUTING -o eth3 -d 192.168.5.3 -j MARK
     --set-mark 0x10503

       Using  IPMARK target we can replace all the mangle/mark rules with only
       one:

     iptables -t mangle -A POSTROUTING -o eth3 -j  IPMARK  --addr=dst
     --and-mask=0xffff --or-mask=0x10000

       On  the routers with hundreds of users there should be significant load
       decrease (e.g. twice).

   IPV4OPTSSTRIP
       Strip all the IP options from a packet.

       The target doesn’t take any option, and therefore is extremly  easy  to
       use :

       # iptables -t mangle -A PREROUTING -j IPV4OPTSSTRIP

   LOG
       Turn  on kernel logging of matching packets.  When this option is set
       for a rule, the Linux kernel will print some information on all match-
       ing  packets  (like most IP header fields) via the kernel log (where it
       can be read with dmesg or syslogd(8)).  This is a "non-terminating tar-
       get",  i.e.  rule traversal continues at the next rule. So if you want
       to LOG the packets you refuse, use two separate rules  with  the  same
       matching criteria, first using target LOG then DROP (or REJECT).

       --log-level level
     Level of logging (numeric or see syslog.conf(5)).

       --log-prefix prefix
     Prefix  log messages with the specified prefix; up to 29 letters
     long, and useful for distinguishing messages in the logs.

       --log-tcp-sequence
     Log TCP sequence numbers. This is a security risk if the log  is
     readable by users.

       --log-tcp-options
     Log options from the TCP packet header.

       --log-ip-options
     Log options from the IP packet header.

       --log-uid
     Log the userid of the process which generated the packet.

   MARK
       This  is used  to  set the  netfilter mark value associated with the
       packet. It is only valid in the mangle table.  It can for  example  be
       used in conjunction with iproute2.

       --set-mark mark

   MASQUERADE
       This  target  is only valid in the nat table, in the POSTROUTING chain.
       It should only be used with dynamically assigned IP  (dialup)  connec-
       tions: if you have a static IP address, you should use the SNAT target.
       Masquerading is equivalent to specifying a mapping to the IP address of
       the  interface  the  packet  is going out, but also has the effect that
       connections are forgotten when the interface goes down. This  is  the
       correct behavior  when the  next  dialup is unlikely to have the same
       interface address (and hence any established connections are lost  any-
       way).  It takes one option:

       --to-ports port[-port]
     This  specifies  a  range of source ports to use, overriding the
     default SNAT source port-selection heuristics (see above).  This
     is only valid if the rule also specifies -p tcp or -p udp.

   MIRROR
       This  is an experimental demonstration target which inverts the source
       and destination fields in the IP header and retransmits the packet.  It
       is  only valid in the INPUT, FORWARD and PREROUTING chains, and user-
       defined chains which are only called from those chains. Note that  the
       outgoing packets  are NOT seen by any packet filtering chains, connec-
       tion tracking or NAT, to avoid loops and other problems.

   NETMAP
       This target allows you to statically map a whole network of  addresses
       onto  another  network of addresses.  It can only be used from rules in
       the nat table.

       --to address[/mask]
     Network address to map to.  The resulting address will  be  con-
     structed in  the following way: All ’one’ bits in the mask are
     filled in from the new ‘address’. All bits that are zero in the
     mask are filled in from the original address.

   NFQUEUE
       This  target  is an extension of the QUEUE target. As opposed to QUEUE,
       it allows you to put a packet into any specific queue, identified  by
       its 16-bit queue number.

       --queue-num value
     This  specifies the QUEUE number to use. Valud queue numbers are
     0 to 65535. The default value is 0.

       It can only be used with Kernel versions 2.6.14 or  later,  since  it
       requires
     the nfnetlink_queue kernel support.

   NOTRACK
       This target disables connection tracking for all packets matching  that
       rule.

       It can only be used in the
     raw table.

   REDIRECT
       This  target is only valid in the nat table, in the PREROUTING and OUT-
       PUT chains, and user-defined chains which are only  called  from those
       chains. It redirects the packet to the machine itself by changing the
       destination IP  to  the primary address  of  the  incoming  interface
       (locally-generated  packets  are mapped to the 127.0.0.1 address).  It
       takes one option:

       --to-ports port[-port]
     This specifies a destination port or  range  of  ports  to  use:
     without  this,  the  destination port is never altered.  This is
     only valid if the rule also specifies -p tcp or -p udp.

   REJECT
       This is used to send back an error packet in response  to  the  matched
       packet: otherwise it is equivalent to DROP so it is a terminating TAR-
       GET, ending rule traversal.  This target is only valid  in  the INPUT,
       FORWARD and  OUTPUT  chains,  and  user-defined chains which are only
       called from those chains.  The following option controls the nature  of
       the error packet returned:

       --reject-with type
     The type given can be
      icmp-net-unreachable
      icmp-host-unreachable
      icmp-port-unreachable
      icmp-proto-unreachable
      icmp-net-prohibited
      icmp-host-prohibited or
      icmp-admin-prohibited (*)
     which  return  the appropriate ICMP error message (port-unreach-
     able is the default).  The option tcp-reset can be used on rules
     which  only match the TCP protocol: this causes a TCP RST packet
     to be sent back. This  is  mainly  useful  for  blocking ident
     (113/tcp) probes which frequently  occur when sending mail to
     broken mail hosts (which won’t accept your mail otherwise).

       (*) Using icmp-admin-prohibited with kernels that  do  not  support  it
       will result in a plain DROP instead of REJECT

   SAME
       Similar to SNAT/DNAT depending on chain: it takes a range of  addresses
       (‘--to  1.2.3.4-1.2.3.7’)  and gives a client the same source-/destina-
       tion-address for each connection.

       --to <ipaddr>-<ipaddr>
     Addresses to map source to. May be specified more than once  for
     multiple ranges.

       --nodst
     Don’t  use the destination-ip in the calculations when selecting
     the new source-ip

   SET
       This modules adds and/or deletes entries from  IP  sets which  can  be
       defined by ipset(8).

       --add-set setname flag[,flag...]
     add the address(es)/port(s) of the packet to the sets

       --del-set setname flag[,flag...]
     delete  the  address(es)/port(s) of  the packet from the sets,
     where flags are src and/or dst and there can be no more than six
     of them.

       The bindings to follow must previously be defined in order to use
     multilevel adding/deleting by the SET target.

   SNAT
       This  target  is only valid in the nat table, in the POSTROUTING chain.
       It specifies that the source address of the packet should  be  modified
       (and  all  future packets in this connection will also be mangled), and
       rules should cease being examined.  It takes one type of option:

       --to-source  ipaddr[-ipaddr][:port-port]
     which can specify a single new source IP address, an  inclusive
     range  of IP  addresses, and optionally, a port range (which is
     only valid if the rule also specifies -p tcp or -p udp). If  no
     port  range  is  specified,  then source ports below 512 will be
     mapped to other ports below 512: those  between 512  and  1023
     inclusive will  be  mapped to ports below 1024, and other ports
     will be mapped to 1024 or above. Where possible, no port alter-
     ation will occur.

     In  Kernels  up  to  2.6.10,  you can  add  several --to-source
     options. For those kernels, if you specify more than one source
     address, either via  an address range or multiple --to-source
     options, a simple round-robin (one after another in cycle) takes
     place  between  these  addresses. Later Kernels (>= 2.6.11-rc1)
     don’t have the ability to NAT to multiple ranges anymore.

   TARPIT
       Captures and holds incoming TCP connections using no local  per-connec-
       tion  resources. Connections are accepted, but immediately switched to
       the persist state (0 byte window), in which the remote side stops send-
       ing  data and asks to continue every 60-240 seconds.  Attempts to close
       the connection are ignored, forcing the remote side  to time  out  the
       connection in 12-24 minutes.

       This   offers   similar functionality to  LaBrea  <http://www.hack-
       busters.net/LaBrea/> but doesn’t require dedicated hardware or IPs. Any
       TCP  port  that you would normally DROP or REJECT can instead become a
       tarpit.

       To tarpit connections to TCP port 80 destined for the current machine:

     iptables -A INPUT -p tcp -m tcp --dport 80 -j TARPIT

       To significantly slow down Code Red/Nimda-style scans of unused address
       space,  forward unused ip  addresses  to  a Linux box not acting as a
       router (e.g. "ip route 10.0.0.0 255.0.0.0 ip.of.linux.box" on a Cisco),
       enable IP forwarding on the Linux box, and add:

     iptables -A FORWARD -p tcp -j TARPIT

     iptables -A FORWARD -j DROP

       NOTE:  If  you use the conntrack module while you are using TARPIT, you
     should also use the NOTRACK target, or the kernel will  unneces-
     sarily  allocate resources  for each  TARPITted connection. To
     TARPIT incoming connections to the standard IRC port while using
     conntrack, you could:

     iptables -t raw -A PREROUTING -p tcp --dport 6667 -j NOTRACK

     iptables -A INPUT -p tcp --dport 6667 -j TARPIT

   TCPMSS
       This  target  allows to alter the MSS value of TCP SYN packets, to con-
       trol the maximum size for that connection (usually limiting it to  your
       outgoing interface’s MTU minus 40).  Of course, it can only be used in
       conjunction with -p tcp. It is only valid in the mangle table.
       This target is used to overcome criminally braindead  ISPs  or  servers
       which  block  ICMP  Fragmentation Needed packets.  The symptoms of this
       problem are that everything works fine from your Linux firewall/router,
       but machines behind it can never exchange large packets:
1) Web browsers connect, then hang with no data received.
2) Small mail works fine, but large emails hang.
3) ssh works fine, but scp hangs after initial handshaking.
       Workaround:  activate  this option and add a rule to your firewall con-
       figuration like:
iptables -t mangle -A FORWARD -p tcp --tcp-flags SYN,RST SYN \
   -j TCPMSS --clamp-mss-to-pmtu

       --set-mss value
     Explicitly set MSS option to specified value.

       --clamp-mss-to-pmtu
     Automatically clamp MSS value to (path_MTU - 40).

       These options are mutually exclusive.

   TOS
       This is used to set the 8-bit Type of Service field in the  IP  header.
       It is only valid in the mangle table.

       --set-tos tos
     You can use a numeric TOS values, or use
      iptables -j TOS -h
     to see the list of valid TOS names.

   TRACE
       This  target  has  no options.  It just turns on packet tracing for all
       packets that match this rule.

   TTL
       This is used to modify the IPv4 TTL header field.  The TTL field deter-
       mines  how many hops (routers) a packet can traverse until it’s time to
       live is exceeded.

       Setting or incrementing the TTL field can potentially be very  danger-
       ous,
     so it should be avoided at any cost.

       Don’t ever set or increment the value on packets that leave your local
       network!
     mangle table.

       --ttl-set value
     Set the TTL value to ‘value’.

       --ttl-dec value
     Decrement the TTL value ‘value’ times.

       --ttl-inc value
     Increment the TTL value ‘value’ times.

   ULOG
       This target provides userspace logging of matching packets.  When  this
       target  is  set for a rule, the Linux kernel will multicast this packet
       through a netlink socket. One or more userspace processes may then sub-
       scribe  to various multicast groups and receive the packets.  Like LOG,
       this is a "non-terminating target", i.e. rule  traversal continues  at
       the next rule.

       --ulog-nlgroup nlgroup
     This  specifies  the netlink group (1-32) to which the packet is
     sent.  Default value is 1.

       --ulog-prefix prefix
     Prefix log messages with the specified prefix; up to 32  charac-
     ters long, and useful for distinguishing messages in the logs.

       --ulog-cprange size
     Number  of bytes to be copied to userspace.  A value of 0 always
     copies the entire packet, regardless of its size. Default is 0.

       --ulog-qthreshold size
     Number of packet to queue inside kernel. Setting this value to,
     e.g. 10 accumulates ten packets inside the kernel and  transmits
     them  as one netlink multipart message to userspace.  Default is
     1 (for backwards compatibility).

   XOR
       Encrypt TCP and UDP traffic using a simple XOR encryption

       --key string
     Set key to "string"

       --block-size
     Set block size

DIAGNOSTICS
       Various error messages are printed to standard error.  The exit code is
       0 for correct functioning.  Errors which appear to be caused by invalid
       or abused command line parameters cause an exit code of 2,  and other
       errors cause an exit code of 1.

BUGS
       Bugs?   What’s  this?  ;-)  Well,  you  might  want  to have a look at
       http://bugzilla.netfilter.org/

COMPATIBILITY WITH IPCHAINS
       This iptables is very similar to ipchains by Rusty Russell.   The  main
       difference  is  that the chains INPUT and OUTPUT are only traversed for
       packets coming into the local host and originating from the local  host
       respectively.   Hence every packet only passes through one of the three
       chains (except loopback traffic, which involves both INPUT  and OUTPUT
       chains); previously a forwarded packet would pass through all three.

       The  other main difference is that -i refers to the input interface; -o
       refers to the output interface, and  both  are  available  for  packets
       entering the FORWARD chain.

       iptables is a pure packet filter when using the default ‘filter’ table,
       with optional extension modules. This should simplify much of the pre-
       vious confusion over the combination of IP masquerading and packet fil-
       tering seen previously. So the following options are  handled  differ-
       ently:
-j MASQ
-M -S
-M -L
       There are several other changes in iptables.

SEE ALSO
       iptables-save(8), iptables-restore(8), ip6tables(8), ip6tables-save(8),
       ip6tables-restore(8), libipq(3).

       The packet-filtering-HOWTO details iptables usage for packet filtering,
       the  NAT-HOWTO  details NAT, the netfilter-extensions-HOWTO details the
       extensions that are not in the standard distribution, and  the  netfil-
       ter-hacking-HOWTO details the netfilter internals.
       See http://www.netfilter.org/.

AUTHORS
       Rusty  Russell  originally  wrote  iptables, in early consultation with
       Michael Neuling.

       Marc Boucher made Rusty abandon ipnatctl by  lobbying  for  a  generic
       packet  selection  framework  in iptables, then wrote the mangle table,
       the owner match, the mark stuff, and ran around doing cool stuff every-
       where.

       James Morris wrote the TOS target, and tos match.

       Jozsef Kadlecsik wrote the REJECT target.

       Harald  Welte  wrote  the  ULOG and NFQUEUE target, the new libiptc, as
       well as the TTL, DSCP, ECN matches and targets.

       The Netfilter Core Team is:  Marc  Boucher,  Martin  Josefsson, Jozsef
       Kadlecsik,  Patrick  McHardy, James Morris, Harald Welte and Rusty Rus-
       sell.

       Man page originally written by Herve Eychenne <rv@wallfire.org>.

Mar 09, 2002   IPTABLES(8)

Comments

Popular posts from this blog