2 .\" $FreeBSD: src/sbin/ipfw/ipfw.8,v 1.63.2.33 2003/02/04 01:36:02 brueffer Exp $
9 .Nd IP firewall and traffic shaper control program
24 .Brq Cm delete | zero | resetlog
29 .Brq Cm firewall | one_pass | debug | verbose | dyn_keepalive
32 .Brq Cm firewall | one_pass | debug | verbose | dyn_keepalive
35 .Cm set Oo Cm disable Ar number ... Oc Op Cm enable Ar number ...
39 .Ar number Cm to Ar number
41 .Cm set swap Ar number number
53 .Brq Cm delete | list | show
61 .Ar macro Ns Op = Ns Ar value
69 utility is the user interface for controlling the
76 .Bd -ragged -offset XXXX
78 this manual page documents the newer version of
82 CURRENT in July 2002, also known as
85 is a superset of the old firewall,
87 The differences between the two are listed in Section
88 .Sx IPFW2 ENHANCEMENTS ,
89 which you are encouraged to read to revise older rulesets and possibly
90 write them more efficiently.
92 .Sx USING IPFW2 IN FreeBSD-STABLE
93 for instructions on how to run
106 numbered from 1 to 65535.
107 Packets are passed to
109 from a number of different places in the protocol stack
110 (depending on the source and destination of the packet,
113 is invoked multiple times on the same packet).
114 The packet passed to the firewall is compared
115 against each of the rules in the firewall
117 When a match is found, the action corresponding to the
118 matching rule is performed.
120 Depending on the action and certain system settings, packets
121 can be reinjected into the firewall at some rule after the
122 matching one for further processing.
126 ruleset always includes a
128 rule (numbered 65535) which cannot be modified,
129 and matches all packets.
130 The action associated with the
136 depending on how the kernel is configured.
138 If the ruleset includes one or more rules with the
146 behaviour, i.e. upon a match it will create dynamic rules matching
147 the exact parameters (addresses and ports) of the matching packet.
149 These dynamic rules, which have a limited lifetime, are checked
150 at the first occurrence of a
155 rule, and are typically used to open the firewall on-demand to
156 legitimate traffic only.
158 .Sx STATEFUL FIREWALL
161 Sections below for more information on the stateful behaviour of
164 All rules (including dynamic ones) have a few associated counters:
165 a packet count, a byte count, a log count and a timestamp
166 indicating the time of the last match.
167 Counters can be displayed or reset with
171 Rules can be added with the
173 command; deleted individually or in groups with the
175 command, and globally with the
177 command; displayed, optionally with the content of the
183 Finally, counters can be reset with the
189 Also, each rule belongs to one of 32 different
193 commands to atomically manipulate sets, such as enable,
194 disable, swap sets, move all rules in a set to another
195 one, delete all rules in a set. These can be useful to
196 install temporary configurations, or to test them.
199 for more information on
202 The following options are available:
203 .Bl -tag -width indent
205 While listing, show counter values.
208 command just implies this option.
210 When entering or showing rules, print them in compact form,
211 i.e. without the optional "ip from any to any" string
212 when this does not carry any additional information.
214 While listing, show dynamic rules in addition to static ones.
216 While listing, if the
218 option was specified, also show expired dynamic rules.
220 Don't ask for confirmation for commands that can cause problems
223 If there is no tty associated with the process, this is implied.
225 Try to resolve addresses and service names in output.
233 be quiet about actions
236 This is useful for adjusting rules by executing multiple
240 .Ql sh\ /etc/rc.firewall ) ,
241 or by processing a file of many
243 rules across a remote login session.
246 is performed in normal (verbose) mode (with the default kernel
247 configuration), it prints a message.
248 Because all rules are flushed, the message might not be delivered
249 to the login session, causing the remote login session to be closed
250 and the remainder of the ruleset to not be processed.
251 Access to the console would then be required to recover.
253 While listing rules, show the
255 each rule belongs to.
256 If this flag is not specified, disabled rules will not be
259 While listing pipes, sort according to one of the four
260 counters (total or current packets or bytes).
262 While listing, show last match timestamp.
265 To ease configuration, rules can be put into a file which is
268 as shown in the last synopsis line.
272 The file will be read line by line and applied as arguments to the
276 Optionally, a preprocessor can be specified using
280 is to be piped through.
281 Useful preprocessors include
287 doesn't start with a slash
289 as its first character, the usual
291 name search is performed.
292 Care should be taken with this in environments where not all
293 file systems are mounted (yet) by the time
295 is being run (e.g. when they are mounted over NFS).
298 has been specified, optional
302 specifications can follow and will be passed on to the preprocessor.
303 This allows for flexible configuration files (like conditionalizing
304 them on the local hostname) and the use of macros to centralize
305 frequently required arguments like IP addresses.
312 commands are used to configure the traffic shaper, as shown in the
313 .Sx TRAFFIC SHAPER (DUMMYNET) CONFIGURATION
316 If the world and the kernel get out of sync the
318 ABI may break, preventing you from being able to add any rules. This can
319 adversely effect the booting process. You can use
323 to temporarily disable the firewall to regain access to the network,
324 allowing you to fix the problem.
326 A packet is checked against the active ruleset in multiple places
327 in the protocol stack, under control of several sysctl variables.
328 These places and variables are shown below, and it is important to
329 have this picture in mind in order to design a correct ruleset.
330 .Bd -literal -offset indent
333 +----------->-----------+
335 [ip_input] [ip_output] net.inet.ip.fw.enable=1
338 [ether_demux] [ether_output_frame] net.link.ether.ipfw=1
340 +-->--[bdg_forward]-->--+ net.link.ether.bridge_ipfw=1
345 As can be noted from the above picture, the number of
346 times the same packet goes through the firewall can
347 vary between 0 and 4 depending on packet source and
348 destination, and system configuration.
350 Note that as packets flow through the stack, headers can be
351 stripped or added to it, and so they may or may not be available
353 E.g., incoming packets will include the MAC header when
357 but the same packets will have the MAC header stripped off when
362 Also note that each packet is always checked against the complete ruleset,
363 irrespective of the place where the check occurs, or the source of the packet.
364 If a rule contains some match patterns or actions which are not valid
365 for the place of invocation (e.g. trying to match a MAC header within
367 ), the match pattern will not match, but a
369 operator in front of such patterns
373 match on those packets.
374 It is thus the responsibility of
375 the programmer, if necessary, to write a suitable ruleset to
376 differentiate among the possible places.
378 rules can be useful here, as an example:
379 .Bd -literal -offset indent
380 # packets from ether_demux or bdg_forward
381 ipfw add 10 skipto 1000 all from any to any layer2 in
382 # packets from ip_input
383 ipfw add 10 skipto 2000 all from any to any not layer2 in
384 # packets from ip_output
385 ipfw add 10 skipto 3000 all from any to any not layer2 out
386 # packets from ether_output_frame
387 ipfw add 10 skipto 4000 all from any to any layer2 out
390 (yes, at the moment there is no way to differentiate between
391 ether_demux and bdg_forward).
395 rules is the following:
396 .Bd -ragged -offset indent
398 .Op Cm set Ar set_number
399 .Op Cm prob Ar match_probability
402 .Op Cm log Op Cm logamount Ar number
406 where the body of the rule specifies which information is used
407 for filtering packets, among the following:
409 .Bl -tag -width "Source and dest. addresses and ports" -offset XXX -compact
410 .It Layer-2 header fields
414 .It Source and dest. addresses and ports
418 .It Transmit and receive interface
420 .It Misc. IP header fields
421 Version, type of service, datagram length, identification,
422 fragment flag (non-zero IP offset),
425 .It Misc. TCP header fields
426 TCP flags (SYN, FIN, ACK, RST, etc.),
427 sequence number, acknowledgment number,
433 When the packet can be associated with a local socket.
436 Note that some of the above information, e.g. source MAC or IP addresses and
437 TCP/UDP ports, could easily be spoofed, so filtering on those fields
438 alone might not guarantee the desired results.
439 .Bl -tag -width indent
441 Each rule is associated with a
443 in the range 1..65535, with the latter reserved for the
446 Rules are checked sequentially by rule number.
447 Multiple rules can have the same number, in which case they are
448 checked (and listed) according to the order in which they have
450 If a rule is entered without specifying a number, the kernel will
451 assign one in such a way that the rule becomes the last one
455 Automatic rule numbers are assigned by incrementing the last
456 non-default rule number by the value of the sysctl variable
457 .Ar net.inet.ip.fw.autoinc_step
458 which defaults to 100.
459 If this is not possible (e.g. because we would go beyond the
460 maximum allowed rule number), the number of the last
461 non-default value is used instead.
462 .It Cm set Ar set_number
463 Each rule is associated with a
465 in the range 0..31, with the latter reserved for the
468 Sets can be individually disabled and enabled, so this parameter
469 is of fundamental importance for atomic ruleset manipulation.
470 It can be also used to simplify deletion of groups of rules.
471 If a rule is entered without specifying a set number,
473 .It Cm prob Ar match_probability
474 A match is only declared with the specified probability
475 (floating point number between 0 and 1).
476 This can be useful for a number of applications such as
477 random packet drop or
480 to simulate the effect of multiple paths leading to out-of-order
482 .It Cm log Op Cm logamount Ar number
483 When a packet matches a rule with the
485 keyword, a message will be
491 The logging only occurs if the sysctl variable
492 .Em net.inet.ip.fw.verbose
494 (which is the default when the kernel is compiled with
495 .Dv IPFIREWALL_VERBOSE
496 ) and the number of packets logged so far for that
497 particular rule does not exceed the
502 is specified, the limit is taken from the sysctl variable
503 .Em net.inet.ip.fw.verbose_limit .
504 In both cases, a value of 0 removes the logging limit.
506 Once the limit is reached, logging can be re-enabled by
507 clearing the logging counter or the packet counter for that entry, see the
513 A rule can be associated with one of the following actions, which
514 will be executed when the packet matches the body of the rule.
515 .Bl -tag -width indent
516 .It Cm allow | accept | pass | permit
517 Allow packets that match rule.
518 The search terminates.
520 Checks the packet against the dynamic ruleset.
521 If a match is found, execute the action associated with
522 the rule which generated this dynamic rule, otherwise
523 move to the next rule.
526 rules do not have a body.
529 rule is found, the dynamic ruleset is checked at the first
535 Update counters for all packets that match rule.
536 The search continues with the next rule.
538 Discard packets that match this rule.
539 The search terminates.
540 .It Cm divert Ar port
541 Divert packets that match this rule to the
545 The search terminates.
546 .It Cm fwd | forward Ar ipaddr Ns Op , Ns Ar port
547 Change the next-hop on matching packets to
549 which can be an IP address in dotted quad format or a host name.
550 The search terminates if this rule matches.
554 is a local address, then matching packets will be forwarded to
556 (or the port number in the packet if one is not specified in the rule)
557 on the local machine.
561 is not a local address, then the port number
562 (if specified) is ignored, and the packet will be
563 forwarded to the remote address, using the route as found in
564 the local routing table for that IP.
568 rule will not match layer-2 packets (those received
569 on ether_input, ether_output, or bridged).
573 action does not change the contents of the packet at all.
574 In particular, the destination address remains unmodified, so
575 packets forwarded to another system will usually be rejected by that system
576 unless there is a matching rule on that system to capture them.
577 For packets forwarded locally,
578 the local address of the socket will be
579 set to the original destination address of the packet.
582 entry look rather weird but is intended for
583 use with transparent proxy servers.
584 .It Cm pipe Ar pipe_nr
588 (for bandwidth limitation, delay, etc.).
590 .Sx TRAFFIC SHAPER (DUMMYNET) CONFIGURATION
591 Section for further information.
592 The search terminates; however, on exit from the pipe and if
596 .Em net.inet.ip.fw.one_pass
597 is not set, the packet is passed again to the firewall code
598 starting from the next rule.
599 .It Cm queue Ar queue_nr
603 (for bandwidth limitation using WF2Q+).
609 Discard packets that match this rule, and if the
610 packet is a TCP packet, try to send a TCP reset (RST) notice.
611 The search terminates.
612 .It Cm skipto Ar number
613 Skip all subsequent rules numbered less than
615 The search continues with the first rule numbered
619 Send a copy of packets matching this rule to the
623 The search terminates and the original packet is accepted
627 .It Cm unreach Ar code
628 Discard packets that match this rule, and try to send an ICMP
629 unreachable notice with code
633 is a number from 0 to 255, or one of these aliases:
634 .Cm net , host , protocol , port ,
635 .Cm needfrag , srcfail , net-unknown , host-unknown ,
636 .Cm isolated , net-prohib , host-prohib , tosnet ,
637 .Cm toshost , filter-prohib , host-precedence
639 .Cm precedence-cutoff .
640 The search terminates.
643 The body of a rule contains zero or more patterns (such as
644 specific source and destination addresses or ports,
645 protocol options, incoming or outgoing interfaces, etc.)
646 that the packet must match in order to be recognised.
647 In general, the patterns are connected by (implicit)
649 operators -- i.e. all must match in order for the
651 Individual patterns can be prefixed by the
653 operator to reverse the result of the match, as in
655 .Dl "ipfw add 100 allow ip from not 1.2.3.4 to any"
657 Additionally, sets of alternative match patterns (
659 ) can be constructed by putting the patterns in
660 lists enclosed between parentheses ( ) or braces { }, and
665 .Dl "ipfw add 100 allow ip from { x or not y or z } to any"
667 Only one level of parentheses is allowed.
668 Beware that most shells have special meanings for parentheses
669 or braces, so it is advisable to put a backslash \\ in front of them
670 to prevent such interpretations.
672 The body of a rule must in general include a source and destination
676 can be used in various places to specify that the content of
677 a required field is irrelevant.
679 The rule body has the following format:
680 .Bd -ragged -offset indent
681 .Op Ar proto Cm from Ar src Cm to Ar dst
685 The first part (protocol from src to dst) is for backward
690 any match pattern (including MAC headers, IPv4 protocols,
691 addresses and ports) can be specified in the
695 Rule fields have the following meaning:
696 .Bl -tag -width indent
697 .It Ar proto : protocol | Cm { Ar protocol Cm or ... }
698 An IPv4 protocol (or an
700 with multiple protocols) specified by number or name
701 (for a complete list see
702 .Pa /etc/protocols ) .
707 keywords mean any protocol will match.
708 .It Ar src No and Ar dst : ip-address | Cm { Ar ip-address Cm or ... } Op Ar ports
713 containing one or more of them,
714 optionally followed by
718 An address (or set of addresses) specified in one of the following
719 ways, optionally preceded by a
722 .Bl -tag -width indent
724 matches any IP address.
726 matches any IP address configured on an interface in the system.
727 The address list is evaluated at the time the packet is
729 .It Ar numeric-ip | hostname
730 Matches a single IPv4 address, specified as dotted-quad or a hostname.
731 Hostnames are resolved at the time the rule is added to the firewall list.
732 .It Ar addr Ns / Ns Ar masklen
733 Matches all addresses with base
735 (specified as a dotted quad or a hostname)
739 As an example, 1.2.3.4/25 will match
740 all IP numbers from 1.2.3.0 to 1.2.3.127 .
741 .It Ar addr Ns / Ns Ar masklen Ns Cm { Ns Ar num,num,... Ns Cm }
742 Matches all addresses with base address
744 (specified as a dotted quad or a hostname)
745 and whose last byte is in the list between braces { } .
746 Note that there must be no spaces between braces, commas and
750 field is used to limit the size of the set of addresses,
751 and can have any value between 24 and 32.
753 As an example, an address specified as 1.2.3.4/24{128,35,55,89}
754 will match the following IP addresses:
756 1.2.3.128 1.2.3.35 1.2.3.55 1.2.3.89 .
758 This format is particularly useful to handle sparse address sets
759 within a single rule. Because the matching occurs using a
760 bitmask, it takes constant time and dramatically reduces
761 the complexity of rulesets.
762 .It Ar addr Ns : Ns Ar mask
763 Matches all addresses with base
765 (specified as a dotted quad or a hostname)
768 specified as a dotted quad.
769 As an example, 1.2.3.4/255.0.255.0 will match
771 We suggest to use this form only for non-contiguous
772 masks, and resort to the
773 .Ar addr Ns / Ns Ar masklen
774 format for contiguous masks, which is more compact and less
777 .It Ar ports : Oo Cm not Oc Bro Ar port | port Ns \&- Ns Ar port Ns Brc Op , Ns Ar ...
778 For protocols which support port numbers (such as TCP and UDP), optional
780 may be specified as one or more ports or port ranges, separated
781 by commas but no spaces, and an optional
786 notation specifies a range of ports (including boundaries).
790 may be used instead of numeric port values.
791 The length of the port list is limited to 30 ports or ranges,
792 though one can specify larger ranges by using an
800 can be used to escape the dash
802 character in a service name (from a shell, the backslash must be
803 typed twice to avoid the shell itself interpreting it as an escape
806 .Dl "ipfw add count tcp from any ftp\e\e-data-ftp to any"
808 Fragmented packets which have a non-zero offset (i.e. not the first
809 fragment) will never match a rule which has one or more port
813 option for details on matching fragmented packets.
815 .Ss RULE OPTIONS (MATCH PATTERNS)
816 Additional match patterns can be used within
817 rules. Zero or more of these so-called
819 can be present in a rule, optionally prefixed by the
821 operand, and possibly grouped into
824 The following match patterns can be used (listed in alphabetical order):
825 .Bl -tag -width indent
827 Matches only bridged packets.
828 .It Cm dst-ip Ar ip address
829 Matches IP packets whose destination IP is one of the address(es)
830 specified as argument.
831 .It Cm dst-port Ar source ports
832 Matches IP packets whose destination port is one of the port(s)
833 specified as argument.
835 Matches TCP packets that have the RST or ACK bits set.
837 Matches packets that are fragments and not the first
838 fragment of an IP datagram. Note that these packets will not have
839 the next protocol header (e.g. TCP, UDP) so options that look into
840 these headers cannot match.
842 Matches all TCP or UDP packets sent by or received for a
846 may be specified by name or number.
847 .It Cm icmptypes Ar types
848 Matches ICMP packets whose ICMP type is in the list
850 The list may be specified as any combination of ranges or
851 individual types separated by commas.
852 The supported ICMP types are:
856 destination unreachable
868 time-to-live exceeded
882 and address mask reply
885 Matches incoming or outgoing packets, respectively.
889 are mutually exclusive (in fact,
894 Matches IP packets whose
899 Matches IP packets whose total length, including header and data, is
902 .It Cm ipoptions Ar spec
903 Matches packets whose IP header contains the comma separated list of
906 The supported IP options are:
909 (strict source route),
911 (loose source route),
913 (record packet route) and
916 The absence of a particular option may be denoted
919 .It Cm ipprecedence Ar precedence
920 Matches IP packets whose precedence field is equal to
923 Matches IP packets whose
925 field contains the comma separated list of
926 service types specified in
928 The supported IP types of service are:
931 .Pq Dv IPTOS_LOWDELAY ,
933 .Pq Dv IPTOS_THROUGHPUT ,
935 .Pq Dv IPTOS_RELIABILITY ,
937 .Pq Dv IPTOS_MINCOST ,
940 The absence of a particular type may be denoted
944 Matches IP packets whose time to live is
946 .It Cm ipversion Ar ver
947 Matches IP packets whose IP version field is
950 Upon a match, the firewall will create a dynamic rule, whose
951 default behaviour is to match bidirectional traffic between
952 source and destination IP/port using the same protocol.
953 The rule has a limited lifetime (controlled by a set of
955 variables), and the lifetime is refreshed every time a matching
958 Matches only layer2 packets, i.e. those passed to
960 from ether_demux() and ether_output_frame().
961 .It Cm limit Bro Cm src-addr | src-port | dst-addr | dst-port Brc Ar N
962 The firewall will only allow
964 connections with the same
965 set of parameters as specified in the rule.
967 of source and destination addresses and ports can be
969 .It Cm { MAC | mac } Ar dst-mac src-mac
970 Match packets with a given
974 addresses, specified as the
976 keyword (matching any MAC address), or six groups of hex digits
978 and optionally followed by a mask indicating how many bits are
981 .Dl "MAC 10:20:30:40:50:60/33 any"
983 Note that the order of MAC addresses (destination first,
985 the same as on the wire, but the opposite of the one used for
987 .It Cm mac-type Ar mac-type
988 Matches packets whose Ethernet Type field
989 corresponds to one of those specified as argument.
991 is specified in the same way as
993 (i.e. one or more comma-separated single values or ranges).
994 You can use symbolic names for known values such as
995 .Em vlan , ipv4, ipv6 .
996 Values can be entered as decimal or hexadecimal (if prefixed by 0x),
997 and they are always printed as hexadecimal (unless the
999 option is used, in which case symbolic resolution will be attempted).
1000 .It Cm proto Ar protocol
1001 Matches packets with the corresponding IPv4 protocol.
1002 .It Cm recv | xmit | via Brq Ar ifX | Ar if Ns Cm * | Ar ipno | Ar any
1003 Matches packets received, transmitted or going through,
1004 respectively, the interface specified by exact name
1005 .Ns No ( Ar ifX Ns No ),
1007 .Ns No ( Ar if Ns Ar * Ns No ),
1008 by IP address, or through some interface.
1012 keyword causes the interface to always be checked.
1019 then only the receive or transmit interface (respectively)
1021 By specifying both, it is possible to match packets based on
1022 both receive and transmit interface, e.g.:
1024 .Dl "ipfw add deny ip from any to any out recv ed0 xmit ed1"
1028 interface can be tested on either incoming or outgoing packets,
1031 interface can only be tested on outgoing packets.
1036 is invalid) whenever
1040 A packet may not have a receive or transmit interface: packets
1041 originating from the local host have no receive interface,
1042 while packets destined for the local host have no transmit
1045 Matches TCP packets that have the SYN bit set but no ACK bit.
1046 This is the short form of
1047 .Dq Li tcpflags\ syn,!ack .
1048 .It Cm src-ip Ar ip-address
1049 Matches IP packets whose source IP is one of the address(es)
1050 specified as argument.
1051 .It Cm src-port Ar ports
1052 Matches IP packets whose source port is one of the port(s)
1053 specified as argument.
1054 .It Cm tcpack Ar ack
1056 Match if the TCP header acknowledgment number field is set to
1058 .It Cm tcpflags Ar spec
1060 Match if the TCP header contains the comma separated list of
1063 The supported TCP flags are:
1072 The absence of a particular flag may be denoted
1075 A rule which contains a
1077 specification can never match a fragmented packet which has
1081 option for details on matching fragmented packets.
1082 .It Cm tcpseq Ar seq
1084 Match if the TCP header sequence number field is set to
1086 .It Cm tcpwin Ar win
1088 Match if the TCP header window field is set to
1090 .It Cm tcpoptions Ar spec
1092 Match if the TCP header contains the comma separated list of
1093 options specified in
1095 The supported TCP options are:
1098 (maximum segment size),
1100 (tcp window advertisement),
1104 (rfc1323 timestamp) and
1106 (rfc1644 t/tcp connection count).
1107 The absence of a particular option may be denoted
1111 Match all TCP or UDP packets sent by or received for a
1115 may be matched by name or identification number.
1118 Each rule belongs to one of 32 different
1121 Set 31 is reserved for the default rule.
1123 By default, rules are put in set 0, unless you use the
1125 attribute when entering a new rule.
1126 Sets can be individually and atomically enabled or disabled,
1127 so this mechanism permits an easy way to store multiple configurations
1128 of the firewall and quickly (and atomically) switch between them.
1129 The command to enable/disable sets is
1130 .Bd -ragged -offset indent
1132 .Cm set Oo Cm disable Ar number ... Oc Op Cm enable Ar number ...
1139 sections can be specified.
1140 Command execution is atomic on all the sets specified in the command.
1141 By default, all sets are enabled.
1143 When you disable a set, its rules behave as if they do not exist
1144 in the firewall configuration, with only one exception:
1145 .Bd -ragged -offset indent
1146 dynamic rules created from a rule before it had been disabled
1147 will still be active until they expire. In order to delete
1148 dynamic rules you have to explicitly delete the parent rule
1149 which generated them.
1152 The set number of rules can be changed with the command
1153 .Bd -ragged -offset indent
1156 .Brq Cm rule Ar rule-number | old-set
1160 Also, you can atomically swap two rulesets with the command
1161 .Bd -ragged -offset indent
1163 .Cm set swap Ar first-set second-set
1168 Section on some possible uses of sets of rules.
1169 .Sh STATEFUL FIREWALL
1170 Stateful operation is a way for the firewall to dynamically
1171 create rules for specific flows when packets that
1172 match a given pattern are detected. Support for stateful
1173 operation comes through the
1174 .Cm check-state , keep-state
1180 Dynamic rules are created when a packet matches a
1184 rule, causing the creation of a
1186 rule which will match all and only packets with
1190 .Em src-ip/src-port dst-ip/dst-port
1195 are used here only to denote the initial match addresses, but they
1196 are completely equivalent afterwards).
1197 Dynamic rules will be checked at the first
1198 .Cm check-state, keep-state
1201 occurrence, and the action performed upon a match will be the same
1202 as in the parent rule.
1204 Note that no additional attributes other than protocol and IP addresses
1205 and ports are checked on dynamic rules.
1207 The typical use of dynamic rules is to keep a closed firewall configuration,
1208 but let the first TCP SYN packet from the inside network install a
1209 dynamic rule for the flow so that packets belonging to that session
1210 will be allowed through the firewall:
1212 .Dl "ipfw add check-state"
1213 .Dl "ipfw add allow tcp from my-subnet to any setup keep-state"
1214 .Dl "ipfw add deny tcp from any to any"
1216 A similar approach can be used for UDP, where an UDP packet coming
1217 from the inside will install a dynamic rule to let the response through
1220 .Dl "ipfw add check-state"
1221 .Dl "ipfw add allow udp from my-subnet to any keep-state"
1222 .Dl "ipfw add deny udp from any to any"
1224 Dynamic rules expire after some time, which depends on the status
1225 of the flow and the setting of some
1229 .Sx SYSCTL VARIABLES
1231 For TCP sessions, dynamic rules can be instructed to periodically
1232 send keepalive packets to refresh the state of the rule when it is
1237 for more examples on how to use dynamic rules.
1238 .Sh TRAFFIC SHAPER (DUMMYNET) CONFIGURATION
1240 is also the user interface for the
1245 operates by first using the firewall to classify packets and divide them into
1247 using any match pattern that can be used in
1250 Depending on local policies, a flow can contain packets for a single
1251 TCP connection, or from/to a given host, or entire subnet, or a
1254 Packets belonging to the same flow are then passed to either of two
1255 different objects, which implement the traffic regulation:
1256 .Bl -hang -offset XXXX
1258 A pipe emulates a link with given bandwidth, propagation delay,
1259 queue size and packet loss rate.
1260 Packets are queued in front of the pipe as they come out from the classifier,
1261 and then transferred to the pipe according to the pipe's parameters.
1265 is an abstraction used to implement the WF2Q+
1266 (Worst-case Fair Weighted Fair Queueing) policy, which is
1267 an efficient variant of the WFQ policy.
1269 The queue associates a
1271 and a reference pipe to each flow, and then all backlogged (i.e.,
1272 with packets queued) flows linked to the same pipe share the pipe's
1273 bandwidth proportionally to their weights.
1274 Note that weights are not priorities; a flow with a lower weight
1275 is still guaranteed to get its fraction of the bandwidth even if a
1276 flow with a higher weight is permanently backlogged.
1281 can be used to set hard limits to the bandwidth that a flow can use, whereas
1283 can be used to determine how different flow share the available bandwidth.
1289 configuration commands are the following:
1290 .Bd -ragged -offset indent
1291 .Cm pipe Ar number Cm config Ar pipe-configuration
1293 .Cm queue Ar number Cm config Ar queue-configuration
1296 The following parameters can be configured for a pipe:
1298 .Bl -tag -width indent -compact
1299 .It Cm bw Ar bandwidth | device
1300 Bandwidth, measured in
1303 .Brq Cm bit/s | Byte/s .
1306 A value of 0 (default) means unlimited bandwidth.
1307 The unit must immediately follow the number, as in
1309 .Dl "ipfw pipe 1 config bw 300Kbit/s"
1311 If a device name is specified instead of a numeric value, as in
1313 .Dl "ipfw pipe 1 config bw tun0"
1315 then the transmit clock is supplied by the specified device.
1316 At the moment only the
1318 device supports this
1319 functionality, for use in conjunction with
1322 .It Cm delay Ar ms-delay
1323 Propagation delay, measured in milliseconds.
1324 The value is rounded to the next multiple of the clock tick
1325 (typically 10ms, but it is a good practice to run kernels
1327 .Dq "options HZ=1000"
1329 the granularity to 1ms or less).
1330 Default value is 0, meaning no delay.
1333 The following parameters can be configured for a queue:
1335 .Bl -tag -width indent -compact
1336 .It Cm pipe Ar pipe_nr
1337 Connects a queue to the specified pipe.
1338 Multiple queues (with the same or different weights) can be connected to
1339 the same pipe, which specifies the aggregate rate for the set of queues.
1341 .It Cm weight Ar weight
1342 Specifies the weight to be used for flows matching this queue.
1343 The weight must be in the range 1..100, and defaults to 1.
1346 Finally, the following parameters can be configured for both
1349 .Bl -tag -width XXXX -compact
1351 .It Cm buckets Ar hash-table-size
1352 Specifies the size of the hash table used for storing the
1354 Default value is 64 controlled by the
1357 .Em net.inet.ip.dummynet.hash_size ,
1358 allowed range is 16 to 65536.
1360 .It Cm mask Ar mask-specifier
1361 Packets sent to a given pipe or queue by an
1363 rule can be further classified into multiple flows, each of which is then
1367 A flow identifier is constructed by masking the IP addresses,
1368 ports and protocol types as specified with the
1370 options in the configuration of the pipe or queue.
1371 For each different flow identifier, a new pipe or queue is created
1372 with the same parameters as the original object, and matching packets
1377 are used, each flow will get the same bandwidth as defined by the pipe,
1380 are used, each flow will share the parent's pipe bandwidth evenly
1381 with other flows generated by the same queue (note that other queues
1382 with different weights might be connected to the same pipe).
1384 Available mask specifiers are a combination of one or more of the following:
1386 .Cm dst-ip Ar mask ,
1387 .Cm src-ip Ar mask ,
1388 .Cm dst-port Ar mask ,
1389 .Cm src-port Ar mask ,
1394 where the latter means all bits in all fields are significant.
1397 When a packet is dropped by a dummynet queue or pipe, the error
1398 is normally reported to the caller routine in the kernel, in the
1399 same way as it happens when a device queue fills up. Setting this
1400 option reports the packet as successfully delivered, which can be
1401 needed for some experimental setups where you want to simulate
1402 loss or congestion at a remote router.
1404 .It Cm plr Ar packet-loss-rate
1407 .Ar packet-loss-rate
1408 is a floating-point number between 0 and 1, with 0 meaning no
1409 loss, 1 meaning 100% loss.
1410 The loss rate is internally represented on 31 bits.
1412 .It Cm queue Brq Ar slots | size Ns Cm Kbytes
1417 Default value is 50 slots, which
1418 is the typical queue size for Ethernet devices.
1419 Note that for slow speed links you should keep the queue
1420 size short or your traffic might be affected by a significant
1422 E.g., 50 max-sized ethernet packets (1500 bytes) mean 600Kbit
1423 or 20s of queue on a 30Kbit/s pipe.
1424 Even worse effect can result if you get packets from an
1425 interface with a much larger MTU, e.g. the loopback interface
1426 with its 16KB packets.
1428 .It Cm red | gred Ar w_q Ns / Ns Ar min_th Ns / Ns Ar max_th Ns / Ns Ar max_p
1429 Make use of the RED (Random Early Detection) queue management algorithm.
1434 point numbers between 0 and 1 (0 not included), while
1438 are integer numbers specifying thresholds for queue management
1439 (thresholds are computed in bytes if the queue has been defined
1440 in bytes, in slots otherwise).
1443 also supports the gentle RED variant (gred).
1446 variables can be used to control the RED behaviour:
1447 .Bl -tag -width indent
1448 .It Em net.inet.ip.dummynet.red_lookup_depth
1449 specifies the accuracy in computing the average queue
1450 when the link is idle (defaults to 256, must be greater than zero)
1451 .It Em net.inet.ip.dummynet.red_avg_pkt_size
1452 specifies the expected average packet size (defaults to 512, must be
1454 .It Em net.inet.ip.dummynet.red_max_pkt_size
1455 specifies the expected maximum packet size, only used when queue
1456 thresholds are in bytes (defaults to 1500, must be greater than zero).
1460 Here are some important points to consider when designing your
1464 Remember that you filter both packets going
1468 Most connections need packets going in both directions.
1470 Remember to test very carefully.
1471 It is a good idea to be near the console when doing this.
1472 If you cannot be near the console,
1473 use an auto-recovery script such as the one in
1474 .Pa /usr/share/examples/ipfw/change_rules.sh .
1476 Don't forget the loopback interface.
1481 There are circumstances where fragmented datagrams are unconditionally
1483 TCP packets are dropped if they do not contain at least 20 bytes of
1484 TCP header, UDP packets are dropped if they do not contain a full 8
1485 byte UDP header, and ICMP packets are dropped if they do not contain
1486 4 bytes of ICMP header, enough to specify the ICMP type, code, and
1488 These packets are simply logged as
1490 since there may not be enough good data in the packet to produce a
1491 meaningful log entry.
1493 Another type of packet is unconditionally dropped, a TCP packet with a
1494 fragment offset of one.
1495 This is a valid packet, but it only has one use, to try
1496 to circumvent firewalls.
1497 When logging is enabled, these packets are
1498 reported as being dropped by rule -1.
1500 If you are logged in over a network, loading the
1504 is probably not as straightforward as you would think.
1505 I recommend the following command line:
1506 .Bd -literal -offset indent
1507 kldload /modules/ipfw.ko && \e
1508 ipfw add 32000 allow ip from any to any
1511 Along the same lines, doing an
1512 .Bd -literal -offset indent
1516 in similar surroundings is also a bad idea.
1520 filter list may not be modified if the system security level
1521 is set to 3 or higher
1524 for information on system security levels).
1526 .Sh PACKET DIVERSION
1529 socket bound to the specified port will receive all packets
1530 diverted to that port.
1531 If no socket is bound to the destination port, or if the kernel
1532 wasn't compiled with divert socket support, the packets are
1534 .Sh SYSCTL VARIABLES
1537 variables controls the behaviour of the firewall and
1538 associated modules (
1539 .Nm dummynet, bridge
1541 These are shown below together with their default value
1542 (but always check with the
1544 command what value is actually in use) and meaning:
1545 .Bl -tag -width indent
1546 .It Em net.inet.ip.dummynet.expire : No 1
1547 Lazily delete dynamic pipes/queue once they have no pending traffic.
1548 You can disable this by setting the variable to 0, in which case
1549 the pipes/queues will only be deleted when the threshold is reached.
1550 .It Em net.inet.ip.dummynet.hash_size : No 64
1551 Default size of the hash table used for dynamic pipes/queues.
1552 This value is used when no
1554 option is specified when configuring a pipe/queue.
1555 .It Em net.inet.ip.dummynet.max_chain_len : No 16
1556 Target value for the maximum number of pipes/queues in a hash bucket.
1558 .Cm max_chain_len*hash_size
1559 is used to determine the threshold over which empty pipes/queues
1560 will be expired even when
1561 .Cm net.inet.ip.dummynet.expire=0 .
1562 .It Em net.inet.ip.dummynet.red_lookup_depth : No 256
1563 .It Em net.inet.ip.dummynet.red_avg_pkt_size : No 512
1564 .It Em net.inet.ip.dummynet.red_max_pkt_size : No 1500
1565 Parameters used in the computations of the drop probability
1566 for the RED algorithm.
1567 .It Em net.inet.ip.fw.autoinc_step : No 100
1568 Delta between rule numbers when auto-generating them.
1569 The value must be in the range 1..1000.
1570 .It Em net.inet.ip.fw.curr_dyn_buckets : Em net.inet.ip.fw.dyn_buckets
1571 The current number of buckets in the hash table for dynamic rules
1573 .It Em net.inet.ip.fw.debug : No 1
1574 Controls debugging messages produced by
1576 .It Em net.inet.ip.fw.dyn_buckets : No 256
1577 The number of buckets in the hash table for dynamic rules.
1578 Must be a power of 2, up to 65536.
1579 It only takes effect when all dynamic rules have expired, so you
1580 are advised to use a
1582 command to make sure that the hash table is resized.
1583 .It Em net.inet.ip.fw.dyn_count : No 3
1584 Current number of dynamic rules
1586 .It Em net.inet.ip.fw.dyn_keepalive : No 1
1587 Enables generation of keepalive packets for
1589 rules on TCP sessions. A keepalive is generated to both
1590 sides of the connection every 5 seconds for the last 20
1591 seconds of the lifetime of the rule.
1592 .It Em net.inet.ip.fw.dyn_max : No 8192
1593 Maximum number of dynamic rules.
1594 When you hit this limit, no more dynamic rules can be
1595 installed until old ones expire.
1596 .It Em net.inet.ip.fw.dyn_ack_lifetime : No 300
1597 .It Em net.inet.ip.fw.dyn_syn_lifetime : No 20
1598 .It Em net.inet.ip.fw.dyn_fin_lifetime : No 1
1599 .It Em net.inet.ip.fw.dyn_rst_lifetime : No 1
1600 .It Em net.inet.ip.fw.dyn_udp_lifetime : No 5
1601 .It Em net.inet.ip.fw.dyn_short_lifetime : No 30
1602 These variables control the lifetime, in seconds, of dynamic
1604 Upon the initial SYN exchange the lifetime is kept short,
1605 then increased after both SYN have been seen, then decreased
1606 again during the final FIN exchange or when a RST is received.
1608 .Em dyn_fin_lifetime
1610 .Em dyn_rst_lifetime
1611 must be strictly lower than 5 seconds, the period of
1612 repetition of keepalives. The firewall enforces that.
1613 .It Em net.inet.ip.fw.enable : No 1
1614 Enables the firewall.
1615 Setting this variable to 0 lets you run your machine without
1616 firewall even if compiled in.
1617 .It Em net.inet.ip.fw.one_pass : No 1
1618 When set, the packet exiting from the
1620 pipe is not passed though the firewall again.
1621 Otherwise, after a pipe action, the packet is
1622 reinjected into the firewall at the next rule.
1624 Note: bridged and layer 2 packets coming out of a pipe
1625 are never reinjected in the firewall irrespective of the
1626 value of this variable.
1627 .It Em net.inet.ip.fw.verbose : No 1
1628 Enables verbose messages.
1629 .It Em net.inet.ip.fw.verbose_limit : No 0
1630 Limits the number of messages produced by a verbose firewall.
1631 .It Em net.link.ether.ipfw : No 0
1632 Controls whether layer-2 packets are passed to
1635 .It Em net.link.ether.bridge_ipfw : No 0
1636 Controls whether bridged packets are passed to
1640 .Sh USING IPFW2 IN FreeBSD-STABLE
1648 unless the kernel is compiled with
1653 .Nm /usr/lib/libalias
1656 and reinstalled (the same effect can be achieved by adding
1660 before a buildworld).
1662 .Sh IPFW2 ENHANCEMENTS
1663 This Section lists the features that have been introduced in
1665 which were not present in
1667 We list them in order of the potential impact that they can
1668 have in writing your rulesets.
1669 You might want to consider using these features in order to
1670 write your rulesets in a more efficient way.
1671 .Bl -tag -width indent
1672 .It Handling of non-IPv4 packets
1674 will silently accept all non-IPv4 packets (which
1677 .Em net.link.ether.bridge_ipfw=1 Ns
1680 will filter all packets (including non-IPv4 ones) according to the ruleset.
1681 To achieve the same behaviour as
1683 you can use the following as the very first rule in your ruleset:
1685 .Dl "ipfw add 1 allow layer2 not mac-type ip"
1689 option might seem redundant, but it is necessary -- packets
1690 passed to the firewall from layer3 will not have a MAC header,
1693 pattern will always fail on them, and the
1695 operator will make this rule into a pass-all.
1698 does not supports address sets (those in the form
1699 .Ar addr/masklen{num,num,...}
1702 .It Port specifications
1704 only allows one port range when specifying TCP and UDP ports, and
1705 is limited to 10 entries instead of the 15 allowed by
1709 you can only specify ports when the rule is requesting
1715 you can put port specifications in rules matching all packets,
1716 and the match will be attempted only on those packets carrying
1717 protocols which include port identifiers.
1721 allowed the first port entry to be specified as
1725 can be an arbitrary 16-bit mask.
1726 This syntax is of questionable usefulness and it is not
1727 supported anymore in
1731 does not support Or-blocks.
1734 does not generate keepalives for stateful sessions.
1735 As a consequence, it might cause idle sessions to drop because
1736 the lifetime of the dynamic rules expires.
1739 does not implement sets of rules.
1740 .It MAC header filtering and Layer-2 firewalling.
1742 does not implement filtering on MAC header fields, nor is it
1743 invoked on packets from
1746 .Cm ether_output_frame().
1748 .Em net.link.ether.ipfw
1749 has no effect there.
1751 The following options are not supported in
1754 .Cm dst-ip, dst-port, layer2, mac, mac-type, src-ip, src-port.
1756 Additionally, the following options are not supported in
1761 .Cm ipid, iplen, ipprecedence, iptos, ipttl,
1762 .Cm ipversion, tcpack, tcpseq, tcpwin .
1763 .It Dummynet options
1764 The following option for
1766 pipes/queues is not supported:
1770 There are far too many possible uses of
1772 so this Section will only give a small set of examples.
1774 .Ss BASIC PACKET FILTERING
1775 This command adds an entry which denies all tcp packets from
1776 .Em cracker.evil.org
1777 to the telnet port of
1779 from being forwarded by the host:
1781 .Dl "ipfw add deny tcp from cracker.evil.org to wolf.tambov.su telnet"
1783 This one disallows any connection from the entire cracker's
1786 .Dl "ipfw add deny ip from 123.45.67.0/24 to my.host.org"
1788 A first and efficient way to limit access (not using dynamic rules)
1789 is the use of the following rules:
1791 .Dl "ipfw add allow tcp from any to any established"
1792 .Dl "ipfw add allow tcp from net1 portlist1 to net2 portlist2 setup"
1793 .Dl "ipfw add allow tcp from net3 portlist3 to net3 portlist3 setup"
1795 .Dl "ipfw add deny tcp from any to any"
1797 The first rule will be a quick match for normal TCP packets,
1798 but it will not match the initial SYN packet, which will be
1801 rules only for selected source/destination pairs.
1802 All other SYN packets will be rejected by the final
1806 If you administer one or more subnets, you can take advantage of the
1808 syntax to specify address sets and or-blocks and write extremely
1809 compact rulesets which selectively enable services to blocks
1810 of clients, as below:
1812 .Dl "goodguys=\*q{ 10.1.2.0/24{20,35,66,18} or 10.2.3.0/28{6,3,11} }\*q"
1813 .Dl "badguys=\*q10.1.2.0/24{8,38,60}\*q"
1815 .Dl "ipfw add allow ip from ${goodguys} to any"
1816 .Dl "ipfw add deny ip from ${badguys} to any"
1817 .Dl "... normal policies ..."
1821 syntax would require a separate rule for each IP in the above
1824 In order to protect a site from flood attacks involving fake
1825 TCP packets, it is safer to use dynamic rules:
1827 .Dl "ipfw add check-state"
1828 .Dl "ipfw add deny tcp from any to any established"
1829 .Dl "ipfw add allow tcp from my-net to any setup keep-state"
1831 This will let the firewall install dynamic rules only for
1832 those connection which start with a regular SYN packet coming
1833 from the inside of our network.
1834 Dynamic rules are checked when encountering the first
1841 rule should usually be placed near the beginning of the
1842 ruleset to minimize the amount of work scanning the ruleset.
1843 Your mileage may vary.
1845 To limit the number of connections a user can open
1846 you can use the following type of rules:
1848 .Dl "ipfw add allow tcp from my-net/24 to any setup limit src-addr 10"
1849 .Dl "ipfw add allow tcp from any to me setup limit src-addr 4"
1851 The former (assuming it runs on a gateway) will allow each host
1852 on a /24 network to open at most 10 TCP connections.
1853 The latter can be placed on a server to make sure that a single
1854 client does not use more than 4 simultaneous connections.
1857 stateful rules can be subject to denial-of-service attacks
1858 by a SYN-flood which opens a huge number of dynamic rules.
1859 The effects of such attacks can be partially limited by
1862 variables which control the operation of the firewall.
1864 Here is a good usage of the
1866 command to see accounting records and timestamp information:
1870 or in short form without timestamps:
1874 which is equivalent to:
1878 Next rule diverts all incoming packets from 192.168.2.0/24
1879 to divert port 5000:
1881 .Dl ipfw divert 5000 ip from 192.168.2.0/24 to any in
1884 The following rules show some of the applications of
1888 for simulations and the like.
1890 This rule drops random incoming packets with a probability
1893 .Dl "ipfw add prob 0.05 deny ip from any to any in"
1895 A similar effect can be achieved making use of dummynet pipes:
1897 .Dl "ipfw add pipe 10 ip from any to any"
1898 .Dl "ipfw pipe 10 config plr 0.05"
1900 We can use pipes to artificially limit bandwidth, e.g. on a
1901 machine acting as a router, if we want to limit traffic from
1902 local clients on 192.168.2.0/24 we do:
1904 .Dl "ipfw add pipe 1 ip from 192.168.2.0/24 to any out"
1905 .Dl "ipfw pipe 1 config bw 300Kbit/s queue 50KBytes"
1907 note that we use the
1909 modifier so that the rule is not used twice.
1910 Remember in fact that
1912 rules are checked both on incoming and outgoing packets.
1914 Should we want to simulate a bidirectional link with bandwidth
1915 limitations, the correct way is the following:
1917 .Dl "ipfw add pipe 1 ip from any to any out"
1918 .Dl "ipfw add pipe 2 ip from any to any in"
1919 .Dl "ipfw pipe 1 config bw 64Kbit/s queue 10Kbytes"
1920 .Dl "ipfw pipe 2 config bw 64Kbit/s queue 10Kbytes"
1922 The above can be very useful, e.g. if you want to see how
1923 your fancy Web page will look for a residential user who
1924 is connected only through a slow link.
1925 You should not use only one pipe for both directions, unless
1926 you want to simulate a half-duplex medium (e.g. AppleTalk,
1928 It is not necessary that both pipes have the same configuration,
1929 so we can also simulate asymmetric links.
1931 Should we want to verify network performance with the RED queue
1932 management algorithm:
1934 .Dl "ipfw add pipe 1 ip from any to any"
1935 .Dl "ipfw pipe 1 config bw 500Kbit/s queue 100 red 0.002/30/80/0.1"
1937 Another typical application of the traffic shaper is to
1938 introduce some delay in the communication.
1939 This can significantly affect applications which do a lot of Remote
1940 Procedure Calls, and where the round-trip-time of the
1941 connection often becomes a limiting factor much more than
1944 .Dl "ipfw add pipe 1 ip from any to any out"
1945 .Dl "ipfw add pipe 2 ip from any to any in"
1946 .Dl "ipfw pipe 1 config delay 250ms bw 1Mbit/s"
1947 .Dl "ipfw pipe 2 config delay 250ms bw 1Mbit/s"
1949 Per-flow queueing can be useful for a variety of purposes.
1950 A very simple one is counting traffic:
1952 .Dl "ipfw add pipe 1 tcp from any to any"
1953 .Dl "ipfw add pipe 1 udp from any to any"
1954 .Dl "ipfw add pipe 1 ip from any to any"
1955 .Dl "ipfw pipe 1 config mask all"
1957 The above set of rules will create queues (and collect
1958 statistics) for all traffic.
1959 Because the pipes have no limitations, the only effect is
1960 collecting statistics.
1961 Note that we need 3 rules, not just the last one, because
1964 tries to match IP packets it will not consider ports, so we
1965 would not see connections on separate ports as different
1968 A more sophisticated example is limiting the outbound traffic
1969 on a net with per-host limits, rather than per-network limits:
1971 .Dl "ipfw add pipe 1 ip from 192.168.2.0/24 to any out"
1972 .Dl "ipfw add pipe 2 ip from any to 192.168.2.0/24 in"
1973 .Dl "ipfw pipe 1 config mask src-ip 0x000000ff bw 200Kbit/s queue 20Kbytes"
1974 .Dl "ipfw pipe 2 config mask dst-ip 0x000000ff bw 200Kbit/s queue 20Kbytes"
1976 To add a set of rules atomically, e.g. set 18:
1978 .Dl "ipfw disable set 18"
1979 .Dl "ipfw add NN set 18 ... # repeat as needed"
1980 .Dl "ipfw enable set 18"
1982 To delete a set of rules atomically the command is simply:
1984 .Dl "ipfw delete set 18"
1986 To test a ruleset and disable it and regain control if something goes wrong:
1988 .Dl "ipfw disable set 18"
1989 .Dl "ipfw add NN set 18 ... # repeat as needed"
1990 .Dl "ipfw enable set 18 ; echo done; sleep 30 && ipfw disable set 18"
1992 Here if everything goes well, you press control-C before the "sleep"
1993 terminates, and your ruleset will be left active. Otherwise, e.g. if
1994 you cannot access your box, the ruleset will be disabled after
1995 the sleep terminates thus restoring the previous situation.
2012 The syntax has grown over the years and sometimes it might be confusing.
2013 Unfortunately, backward compatibility prevents cleaning up mistakes
2014 made in the definition of the syntax.
2018 Misconfiguring the firewall can put your computer in an unusable state,
2019 possibly shutting down network services and requiring console access to
2020 regain control of it.
2022 Incoming packet fragments diverted by
2026 are reassembled before delivery to the socket.
2027 The action used on those packet is the one from the
2028 rule which matches the first fragment of the packet.
2030 Packets that match a
2032 rule should not be immediately accepted, but should continue
2033 going through the rule list.
2034 This may be fixed in a later version.
2036 Packets diverted to userland, and then reinserted by a userland process
2039 will lose various packet attributes, including their source interface.
2040 If a packet is reinserted in this manner, later rules may be incorrectly
2041 applied, making the order of
2043 rules in the rule sequence very important.
2045 .An Ugen J. S. Antsilevich ,
2046 .An Poul-Henning Kamp ,
2052 API based upon code written by
2058 traffic shaper supported by Akamba Corp.
2062 utility first appeared in
2067 Stateful extensions were introduced in
2070 was introduced in Summer 2002.