9 .Nd User interface for firewall, traffic shaper, packet scheduler,
12 .Ss FIREWALL CONFIGURATION
21 .Op Ar rule | first-last ...
29 .Brq Cm delete | zero | resetlog
33 .Cm set Oo Cm disable Ar number ... Oc Op Cm enable Ar number ...
37 .Ar number Cm to Ar number
39 .Cm set swap Ar number number
45 .Brq Cm firewall | altq | one_pass | debug | verbose | dyn_keepalive
48 .Brq Cm firewall | altq | one_pass | debug | verbose | dyn_keepalive
51 .Oo Cm set Ar N Oc Cm table Ar name Cm create Ar create-options
53 .Oo Cm set Ar N Oc Cm table Ar name Cm destroy
55 .Oo Cm set Ar N Oc Cm table Ar name Cm modify Ar modify-options
57 .Oo Cm set Ar N Oc Cm table Ar name Cm swap Ar name
59 .Oo Cm set Ar N Oc Cm table Ar name Cm add Ar table-key Op Ar value
61 .Oo Cm set Ar N Oc Cm table Ar name Cm add Op Ar table-key Ar value ...
63 .Oo Cm set Ar N Oc Cm table Ar name Cm atomic add Op Ar table-key Ar value ...
65 .Oo Cm set Ar N Oc Cm table Ar name Cm delete Op Ar table-key ...
67 .Oo Cm set Ar N Oc Cm table Ar name Cm lookup Ar addr
69 .Oo Cm set Ar N Oc Cm table Ar name Cm lock
71 .Oo Cm set Ar N Oc Cm table Ar name Cm unlock
73 .Oo Cm set Ar N Oc Cm table
77 .Oo Cm set Ar N Oc Cm table
81 .Oo Cm set Ar N Oc Cm table
85 .Oo Cm set Ar N Oc Cm table
88 .Ss DUMMYNET CONFIGURATION (TRAFFIC SHAPER AND PACKET SCHEDULER)
90 .Brq Cm pipe | queue | sched
96 .Brq Cm pipe | queue | sched
97 .Brq Cm delete | list | show
116 .Ss INTERNAL DIAGNOSTICS
126 utility is the user interface for controlling the
130 traffic shaper/packet scheduler, and the
131 in-kernel NAT services.
133 A firewall configuration, or
137 numbered from 1 to 65535.
138 Packets are passed to the firewall
139 from a number of different places in the protocol stack
140 (depending on the source and destination of the packet,
141 it is possible for the firewall to be
142 invoked multiple times on the same packet).
143 The packet passed to the firewall is compared
144 against each of the rules in the
147 (multiple rules with the same number are permitted, in which case
148 they are processed in order of insertion).
149 When a match is found, the action corresponding to the
150 matching rule is performed.
152 Depending on the action and certain system settings, packets
153 can be reinjected into the firewall at some rule after the
154 matching one for further processing.
156 A ruleset always includes a
158 rule (numbered 65535) which cannot be modified or deleted,
159 and matches all packets.
160 The action associated with the
166 depending on how the kernel is configured.
168 If the ruleset includes one or more rules with the
173 the firewall will have a
175 behaviour, i.e., upon a match it will create
177 i.e., rules that match packets with the same 5-tuple
178 (protocol, source and destination addresses and ports)
179 as the packet which caused their creation.
180 Dynamic rules, which have a limited lifetime, are checked
181 at the first occurrence of a
186 rule, and are typically used to open the firewall on-demand to
187 legitimate traffic only.
189 .Sx STATEFUL FIREWALL
192 Sections below for more information on the stateful behaviour of
195 All rules (including dynamic ones) have a few associated counters:
196 a packet count, a byte count, a log count and a timestamp
197 indicating the time of the last match.
198 Counters can be displayed or reset with
202 Each rule belongs to one of 32 different
206 commands to atomically manipulate sets, such as enable,
207 disable, swap sets, move all rules in a set to another
208 one, delete all rules in a set.
209 These can be useful to
210 install temporary configurations, or to test them.
213 for more information on
216 Rules can be added with the
218 command; deleted individually or in groups with the
220 command, and globally (except those in set 31) with the
222 command; displayed, optionally with the content of the
228 Finally, counters can be reset with the
235 The following general options are available when invoking
237 .Bl -tag -width indent
239 Show counter values when listing rules.
242 command implies this option.
244 Only show the action and the comment, not the body of a rule.
248 When entering or showing rules, print them in compact form,
249 i.e., omitting the "ip from any to any" string
250 when this does not carry any additional information.
252 When listing, show dynamic rules in addition to static ones.
256 is specified, also show expired dynamic rules.
258 Do not ask for confirmation for commands that can cause problems
261 If there is no tty associated with the process, this is implied.
263 When listing a table (see the
265 section below for more information on lookup tables), format values
267 By default, values are shown as integers.
269 Only check syntax of the command strings, without actually passing
272 Try to resolve addresses and service names in output.
274 Be quiet when executing the
284 This is useful when updating rulesets by executing multiple
288 .Ql sh\ /etc/rc.firewall ) ,
289 or by processing a file with many
291 rules across a remote login session.
292 It also stops a table add or delete
293 from failing if the entry already exists or is not present.
295 The reason why this option may be important is that
296 for some of these actions,
298 may print a message; if the action results in blocking the
299 traffic to the remote client,
300 the remote login session will be closed
301 and the rest of the ruleset will not be processed.
302 Access to the console would then be required to recover.
304 When listing rules, show the
306 each rule belongs to.
307 If this flag is not specified, disabled rules will not be
310 When listing pipes, sort according to one of the four
311 counters (total or current packets or bytes).
313 When listing, show last match timestamp converted with ctime().
315 When listing, show last match timestamp as seconds from the epoch.
316 This form can be more convenient for postprocessing by scripts.
318 .Ss LIST OF RULES AND PREPROCESSING
319 To ease configuration, rules can be put into a file which is
322 as shown in the last synopsis line.
326 The file will be read line by line and applied as arguments to the
330 Optionally, a preprocessor can be specified using
334 is to be piped through.
335 Useful preprocessors include
341 does not start with a slash
343 as its first character, the usual
345 name search is performed.
346 Care should be taken with this in environments where not all
347 file systems are mounted (yet) by the time
349 is being run (e.g.\& when they are mounted over NFS).
352 has been specified, any additional arguments are passed on to the preprocessor
354 This allows for flexible configuration files (like conditionalizing
355 them on the local hostname) and the use of macros to centralize
356 frequently required arguments like IP addresses.
357 .Ss TRAFFIC SHAPER CONFIGURATION
363 commands are used to configure the traffic shaper and packet scheduler.
365 .Sx TRAFFIC SHAPER (DUMMYNET) CONFIGURATION
366 Section below for details.
368 If the world and the kernel get out of sync the
370 ABI may break, preventing you from being able to add any rules.
371 This can adversely affect the booting process.
376 to temporarily disable the firewall to regain access to the network,
377 allowing you to fix the problem.
379 A packet is checked against the active ruleset in multiple places
380 in the protocol stack, under control of several sysctl variables.
381 These places and variables are shown below, and it is important to
382 have this picture in mind in order to design a correct ruleset.
383 .Bd -literal -offset indent
386 +----------->-----------+
388 [ip(6)_input] [ip(6)_output] net.inet(6).ip(6).fw.enable=1
391 [ether_demux] [ether_output_frame] net.link.ether.ipfw=1
393 +-->--[bdg_forward]-->--+ net.link.bridge.ipfw=1
399 times the same packet goes through the firewall can
400 vary between 0 and 4 depending on packet source and
401 destination, and system configuration.
403 Note that as packets flow through the stack, headers can be
404 stripped or added to it, and so they may or may not be available
406 E.g., incoming packets will include the MAC header when
410 but the same packets will have the MAC header stripped off when
417 Also note that each packet is always checked against the complete ruleset,
418 irrespective of the place where the check occurs, or the source of the packet.
419 If a rule contains some match patterns or actions which are not valid
420 for the place of invocation (e.g.\& trying to match a MAC header within
424 the match pattern will not match, but a
426 operator in front of such patterns
430 match on those packets.
431 It is thus the responsibility of
432 the programmer, if necessary, to write a suitable ruleset to
433 differentiate among the possible places.
435 rules can be useful here, as an example:
436 .Bd -literal -offset indent
437 # packets from ether_demux or bdg_forward
438 ipfw add 10 skipto 1000 all from any to any layer2 in
439 # packets from ip_input
440 ipfw add 10 skipto 2000 all from any to any not layer2 in
441 # packets from ip_output
442 ipfw add 10 skipto 3000 all from any to any not layer2 out
443 # packets from ether_output_frame
444 ipfw add 10 skipto 4000 all from any to any layer2 out
447 (yes, at the moment there is no way to differentiate between
448 ether_demux and bdg_forward).
450 In general, each keyword or argument must be provided as
451 a separate command line argument, with no leading or trailing
453 Keywords are case-sensitive, whereas arguments may
454 or may not be case-sensitive depending on their nature
455 (e.g.\& uid's are, hostnames are not).
457 Some arguments (e.g., port or address lists) are comma-separated
459 In this case, spaces after commas ',' are allowed to make
460 the line more readable.
461 You can also put the entire
462 command (including flags) into a single argument.
463 E.g., the following forms are equivalent:
464 .Bd -literal -offset indent
465 ipfw -q add deny src-ip 10.0.0.0/24,127.0.0.1/8
466 ipfw -q add deny src-ip 10.0.0.0/24, 127.0.0.1/8
467 ipfw "-q add deny src-ip 10.0.0.0/24, 127.0.0.1/8"
470 The format of firewall rules is the following:
471 .Bd -ragged -offset indent
474 .Op Cm set Ar set_number
475 .Op Cm prob Ar match_probability
477 .Op Cm log Op Cm logamount Ar number
487 where the body of the rule specifies which information is used
488 for filtering packets, among the following:
490 .Bl -tag -width "Source and dest. addresses and ports" -offset XXX -compact
491 .It Layer-2 header fields
493 .It IPv4 and IPv6 Protocol
495 .It Source and dest. addresses and ports
499 .It Transmit and receive interface
501 .It Misc. IP header fields
502 Version, type of service, datagram length, identification,
503 fragment flag (non-zero IP offset),
506 .It IPv6 Extension headers
507 Fragmentation, Hop-by-Hop options,
508 Routing Headers, Source routing rthdr0, Mobile IPv6 rthdr2, IPSec options.
510 .It Misc. TCP header fields
511 TCP flags (SYN, FIN, ACK, RST, etc.),
512 sequence number, acknowledgment number,
520 When the packet can be associated with a local socket.
522 Whether a packet came from a divert socket (e.g.,
524 .It Fib annotation state
525 Whether a packet has been tagged for using a specific FIB (routing table)
526 in future forwarding decisions.
529 Note that some of the above information, e.g.\& source MAC or IP addresses and
530 TCP/UDP ports, can be easily spoofed, so filtering on those fields
531 alone might not guarantee the desired results.
532 .Bl -tag -width indent
534 Each rule is associated with a
536 in the range 1..65535, with the latter reserved for the
539 Rules are checked sequentially by rule number.
540 Multiple rules can have the same number, in which case they are
541 checked (and listed) according to the order in which they have
543 If a rule is entered without specifying a number, the kernel will
544 assign one in such a way that the rule becomes the last one
548 Automatic rule numbers are assigned by incrementing the last
549 non-default rule number by the value of the sysctl variable
550 .Ar net.inet.ip.fw.autoinc_step
551 which defaults to 100.
552 If this is not possible (e.g.\& because we would go beyond the
553 maximum allowed rule number), the number of the last
554 non-default value is used instead.
555 .It Cm set Ar set_number
556 Each rule is associated with a
559 Sets can be individually disabled and enabled, so this parameter
560 is of fundamental importance for atomic ruleset manipulation.
561 It can be also used to simplify deletion of groups of rules.
562 If a rule is entered without specifying a set number,
565 Set 31 is special in that it cannot be disabled,
566 and rules in set 31 are not deleted by the
568 command (but you can delete them with the
569 .Nm ipfw delete set 31
571 Set 31 is also used for the
574 .It Cm prob Ar match_probability
575 A match is only declared with the specified probability
576 (floating point number between 0 and 1).
577 This can be useful for a number of applications such as
578 random packet drop or
581 to simulate the effect of multiple paths leading to out-of-order
584 Note: this condition is checked before any other condition, including
585 ones such as keep-state or check-state which might have side effects.
586 .It Cm log Op Cm logamount Ar number
587 Packets matching a rule with the
589 keyword will be made available for logging in two ways:
590 if the sysctl variable
591 .Va net.inet.ip.fw.verbose
592 is set to 0 (default), one can use
597 This pseudo interface can be created after a boot
598 manually by using the following command:
599 .Bd -literal -offset indent
600 # ifconfig ipfw0 create
603 Or, automatically at boot time by adding the following
607 .Bd -literal -offset indent
611 There is no overhead if no
613 is attached to the pseudo interface.
616 .Va net.inet.ip.fw.verbose
617 is set to 1, packets will be logged to
621 facility up to a maximum of
626 is specified, the limit is taken from the sysctl variable
627 .Va net.inet.ip.fw.verbose_limit .
628 In both cases, a value of 0 means unlimited logging.
630 Once the limit is reached, logging can be re-enabled by
631 clearing the logging counter or the packet counter for that entry, see the
635 Note: logging is done after all other packet matching conditions
636 have been successfully verified, and before performing the final
637 action (accept, deny, etc.) on the packet.
639 When a packet matches a rule with the
641 keyword, the numeric tag for the given
643 in the range 1..65534 will be attached to the packet.
644 The tag acts as an internal marker (it is not sent out over
645 the wire) that can be used to identify these packets later on.
646 This can be used, for example, to provide trust between interfaces
647 and to start doing policy-based filtering.
648 A packet can have multiple tags at the same time.
649 Tags are "sticky", meaning once a tag is applied to a packet by a
650 matching rule it exists until explicit removal.
651 Tags are kept with the packet everywhere within the kernel, but are
652 lost when packet leaves the kernel, for example, on transmitting
653 packet out to the network or sending packet to a
657 To check for previously applied tags, use the
660 To delete previously applied tag, use the
664 Note: since tags are kept with the packet everywhere in kernelspace,
665 they can be set and unset anywhere in the kernel network subsystem
668 facility), not only by means of the
674 For example, there can be a specialized
676 node doing traffic analyzing and tagging for later inspecting
678 .It Cm untag Ar number
679 When a packet matches a rule with the
681 keyword, the tag with the number
683 is searched among the tags attached to this packet and,
684 if found, removed from it.
685 Other tags bound to packet, if present, are left untouched.
687 When a packet matches a rule with the
689 keyword, the ALTQ identifier for the given
694 Note that this ALTQ tag is only meaningful for packets going "out" of IPFW,
695 and not being rejected or going to divert sockets.
696 Note that if there is insufficient memory at the time the packet is
697 processed, it will not be tagged, so it is wise to make your ALTQ
698 "default" queue policy account for this.
701 rules match a single packet, only the first one adds the ALTQ classification
703 In doing so, traffic may be shaped by using
704 .Cm count Cm altq Ar queue
705 rules for classification early in the ruleset, then later applying
706 the filtering decision.
711 rules may come later and provide the actual filtering decisions in
712 addition to the fallback ALTQ tag.
716 to set up the queues before IPFW will be able to look them up by name,
717 and if the ALTQ disciplines are rearranged, the rules in containing the
718 queue identifiers in the kernel will likely have gone stale and need
720 Stale queue identifiers will probably result in misclassification.
722 All system ALTQ processing can be turned on or off via
727 .Cm disable Ar altq .
729 .Va net.inet.ip.fw.one_pass
730 is irrelevant to ALTQ traffic shaping, as the actual rule action is followed
731 always after adding an ALTQ tag.
734 A rule can be associated with one of the following actions, which
735 will be executed when the packet matches the body of the rule.
736 .Bl -tag -width indent
737 .It Cm allow | accept | pass | permit
738 Allow packets that match rule.
739 The search terminates.
741 Checks the packet against the dynamic ruleset.
742 If a match is found, execute the action associated with
743 the rule which generated this dynamic rule, otherwise
744 move to the next rule.
747 rules do not have a body.
750 rule is found, the dynamic ruleset is checked at the first
756 Update counters for all packets that match rule.
757 The search continues with the next rule.
759 Discard packets that match this rule.
760 The search terminates.
761 .It Cm divert Ar port
762 Divert packets that match this rule to the
766 The search terminates.
767 .It Cm fwd | forward Ar ipaddr | tablearg Ns Op , Ns Ar port
768 Change the next-hop on matching packets to
770 which can be an IP address or a host name.
771 For IPv4, the next hop can also be supplied by the last table
772 looked up for the packet by using the
774 keyword instead of an explicit address.
775 The search terminates if this rule matches.
779 is a local address, then matching packets will be forwarded to
781 (or the port number in the packet if one is not specified in the rule)
782 on the local machine.
786 is not a local address, then the port number
787 (if specified) is ignored, and the packet will be
788 forwarded to the remote address, using the route as found in
789 the local routing table for that IP.
793 rule will not match layer-2 packets (those received
794 on ether_input, ether_output, or bridged).
798 action does not change the contents of the packet at all.
799 In particular, the destination address remains unmodified, so
800 packets forwarded to another system will usually be rejected by that system
801 unless there is a matching rule on that system to capture them.
802 For packets forwarded locally,
803 the local address of the socket will be
804 set to the original destination address of the packet.
807 entry look rather weird but is intended for
808 use with transparent proxy servers.
809 .It Cm nat Ar nat_nr | tablearg
812 (for network address translation, address redirect, etc.):
814 .Sx NETWORK ADDRESS TRANSLATION (NAT)
815 Section for further information.
816 .It Cm pipe Ar pipe_nr
820 (for bandwidth limitation, delay, etc.).
822 .Sx TRAFFIC SHAPER (DUMMYNET) CONFIGURATION
823 Section for further information.
824 The search terminates; however, on exit from the pipe and if
828 .Va net.inet.ip.fw.one_pass
829 is not set, the packet is passed again to the firewall code
830 starting from the next rule.
831 .It Cm queue Ar queue_nr
835 (for bandwidth limitation using WF2Q+).
841 Discard packets that match this rule, and if the
842 packet is a TCP packet, try to send a TCP reset (RST) notice.
843 The search terminates.
845 Discard packets that match this rule, and if the
846 packet is a TCP packet, try to send a TCP reset (RST) notice.
847 The search terminates.
848 .It Cm skipto Ar number | tablearg
849 Skip all subsequent rules numbered less than
851 The search continues with the first rule numbered
854 It is possible to use the
856 keyword with a skipto for a
858 skipto. Skipto may work either in O(log(N)) or in O(1) depending
859 on amount of memory and/or sysctl variables.
862 section for more details.
863 .It Cm call Ar number | tablearg
864 The current rule number is saved in the internal stack and
865 ruleset processing continues with the first rule numbered
868 If later a rule with the
870 action is encountered, the processing returns to the first rule
873 rule plus one or higher
874 (the same behaviour as with packets returning from
879 This could be used to make somewhat like an assembly language
881 calls to rules with common checks for different interfaces, etc.
883 Rule with any number could be called, not just forward jumps as with
885 So, to prevent endless loops in case of mistakes, both
889 actions don't do any jumps and simply go to the next rule if memory
890 cannot be allocated or stack overflowed/underflowed.
892 Internally stack for rule numbers is implemented using
894 facility and currently has size of 16 entries.
895 As mbuf tags are lost when packet leaves the kernel,
897 should not be used in subroutines to avoid endless loops
898 and other undesired effects.
900 Takes rule number saved to internal stack by the last
902 action and returns ruleset processing to the first rule
903 with number greater than number of corresponding
906 See description of the
908 action for more details.
914 and thus are unconditional, but
916 command-line utility currently requires every action except
919 While it is sometimes useful to return only on some packets,
920 usually you want to print just
923 A workaround for this is to use new syntax and
926 .Bd -literal -offset indent
927 # Add a rule without actual body
928 ipfw add 2999 return via any
930 # List rules without "from any to any" part
934 This cosmetic annoyance may be fixed in future releases.
936 Send a copy of packets matching this rule to the
940 The search continues with the next rule.
941 .It Cm unreach Ar code
942 Discard packets that match this rule, and try to send an ICMP
943 unreachable notice with code
947 is a number from 0 to 255, or one of these aliases:
948 .Cm net , host , protocol , port ,
949 .Cm needfrag , srcfail , net-unknown , host-unknown ,
950 .Cm isolated , net-prohib , host-prohib , tosnet ,
951 .Cm toshost , filter-prohib , host-precedence
953 .Cm precedence-cutoff .
954 The search terminates.
955 .It Cm unreach6 Ar code
956 Discard packets that match this rule, and try to send an ICMPv6
957 unreachable notice with code
961 is a number from 0, 1, 3 or 4, or one of these aliases:
962 .Cm no-route, admin-prohib, address
965 The search terminates.
966 .It Cm netgraph Ar cookie
967 Divert packet into netgraph with given
969 The search terminates.
970 If packet is later returned from netgraph it is either
971 accepted or continues with the next rule, depending on
972 .Va net.inet.ip.fw.one_pass
974 .It Cm ngtee Ar cookie
975 A copy of packet is diverted into netgraph, original
976 packet continues with the next rule.
979 for more information on
984 .It Cm setfib Ar fibnum | tablearg
985 The packet is tagged so as to use the FIB (routing table)
987 in any subsequent forwarding decisions.
988 In the current implementation, this is limited to the values 0 through 15, see
990 Processing continues at the next rule.
991 It is possible to use the
994 If the tablearg value is not within the compiled range of fibs,
995 the packet's fib is set to 0.
996 .It Cm setdscp Ar DSCP | number | tablearg
997 Set specified DiffServ codepoint for an IPv4/IPv6 packet.
998 Processing continues at the next rule.
999 Supported values are:
1045 Additionally, DSCP value can be specified by number (0..64).
1046 It is also possible to use the
1048 keyword with setdscp.
1049 If the tablearg value is not within the 0..64 range, lower 6 bits of supplied
1052 Queue and reassemble IP fragments.
1053 If the packet is not fragmented, counters are updated and
1054 processing continues with the next rule.
1055 If the packet is the last logical fragment, the packet is reassembled and, if
1056 .Va net.inet.ip.fw.one_pass
1057 is set to 0, processing continues with the next rule.
1058 Otherwise, the packet is allowed to pass and the search terminates.
1059 If the packet is a fragment in the middle of a logical group of fragments,
1061 processing stops immediately.
1063 Fragment handling can be tuned via
1064 .Va net.inet.ip.maxfragpackets
1066 .Va net.inet.ip.maxfragsperpacket
1067 which limit, respectively, the maximum number of processable
1068 fragments (default: 800) and
1069 the maximum number of fragments per packet (default: 16).
1071 NOTA BENE: since fragments do not contain port numbers,
1072 they should be avoided with the
1075 Alternatively, direction-based (like
1079 ) and source-based (like
1081 ) match patterns can be used to select fragments.
1083 Usually a simple rule like:
1084 .Bd -literal -offset indent
1085 # reassemble incoming fragments
1086 ipfw add reass all from any to any in
1089 is all you need at the beginning of your ruleset.
1092 The body of a rule contains zero or more patterns (such as
1093 specific source and destination addresses or ports,
1094 protocol options, incoming or outgoing interfaces, etc.)
1095 that the packet must match in order to be recognised.
1096 In general, the patterns are connected by (implicit)
1098 operators -- i.e., all must match in order for the
1100 Individual patterns can be prefixed by the
1102 operator to reverse the result of the match, as in
1104 .Dl "ipfw add 100 allow ip from not 1.2.3.4 to any"
1106 Additionally, sets of alternative match patterns
1108 can be constructed by putting the patterns in
1109 lists enclosed between parentheses ( ) or braces { }, and
1112 operator as follows:
1114 .Dl "ipfw add 100 allow ip from { x or not y or z } to any"
1116 Only one level of parentheses is allowed.
1117 Beware that most shells have special meanings for parentheses
1118 or braces, so it is advisable to put a backslash \\ in front of them
1119 to prevent such interpretations.
1121 The body of a rule must in general include a source and destination
1125 can be used in various places to specify that the content of
1126 a required field is irrelevant.
1128 The rule body has the following format:
1129 .Bd -ragged -offset indent
1130 .Op Ar proto Cm from Ar src Cm to Ar dst
1134 The first part (proto from src to dst) is for backward
1135 compatibility with earlier versions of
1139 any match pattern (including MAC headers, IP protocols,
1140 addresses and ports) can be specified in the
1144 Rule fields have the following meaning:
1145 .Bl -tag -width indent
1146 .It Ar proto : protocol | Cm { Ar protocol Cm or ... }
1147 .It Ar protocol : Oo Cm not Oc Ar protocol-name | protocol-number
1148 An IP protocol specified by number or name
1149 (for a complete list see
1150 .Pa /etc/protocols ) ,
1151 or one of the following keywords:
1152 .Bl -tag -width indent
1154 Matches IPv4 packets.
1156 Matches IPv6 packets.
1165 option will be treated as inner protocol.
1173 .Cm { Ar protocol Cm or ... }
1176 is provided for convenience only but its use is deprecated.
1177 .It Ar src No and Ar dst : Bro Cm addr | Cm { Ar addr Cm or ... } Brc Op Oo Cm not Oc Ar ports
1178 An address (or a list, see below)
1179 optionally followed by
1185 with multiple addresses) is provided for convenience only and
1186 its use is discouraged.
1187 .It Ar addr : Oo Cm not Oc Bro
1188 .Cm any | me | me6 |
1189 .Cm table Ns Pq Ar name Ns Op , Ns Ar value
1190 .Ar | addr-list | addr-set
1192 .Bl -tag -width indent
1194 matches any IP address.
1196 matches any IP address configured on an interface in the system.
1198 matches any IPv6 address configured on an interface in the system.
1199 The address list is evaluated at the time the packet is
1201 .It Cm table Ns Pq Ar name Ns Op , Ns Ar value
1202 Matches any IPv4 or IPv6 address for which an entry exists in the lookup table
1204 If an optional 32-bit unsigned
1206 is also specified, an entry will match only if it has this value.
1209 section below for more information on lookup tables.
1211 .It Ar addr-list : ip-addr Ns Op Ns , Ns Ar addr-list
1213 A host or subnet address specified in one of the following ways:
1214 .Bl -tag -width indent
1215 .It Ar numeric-ip | hostname
1216 Matches a single IPv4 address, specified as dotted-quad or a hostname.
1217 Hostnames are resolved at the time the rule is added to the firewall list.
1218 .It Ar addr Ns / Ns Ar masklen
1219 Matches all addresses with base
1221 (specified as an IP address, a network number, or a hostname)
1225 As an example, 1.2.3.4/25 or 1.2.3.0/25 will match
1226 all IP numbers from 1.2.3.0 to 1.2.3.127 .
1227 .It Ar addr Ns : Ns Ar mask
1228 Matches all addresses with base
1230 (specified as an IP address, a network number, or a hostname)
1233 specified as a dotted quad.
1234 As an example, 1.2.3.4:255.0.255.0 or 1.0.3.0:255.0.255.0 will match
1236 This form is advised only for non-contiguous
1238 It is better to resort to the
1239 .Ar addr Ns / Ns Ar masklen
1240 format for contiguous masks, which is more compact and less
1243 .It Ar addr-set : addr Ns Oo Ns / Ns Ar masklen Oc Ns Cm { Ns Ar list Ns Cm }
1244 .It Ar list : Bro Ar num | num-num Brc Ns Op Ns , Ns Ar list
1245 Matches all addresses with base address
1247 (specified as an IP address, a network number, or a hostname)
1248 and whose last byte is in the list between braces { } .
1249 Note that there must be no spaces between braces and
1250 numbers (spaces after commas are allowed).
1251 Elements of the list can be specified as single entries
1255 field is used to limit the size of the set of addresses,
1256 and can have any value between 24 and 32.
1258 it will be assumed as 24.
1260 This format is particularly useful to handle sparse address sets
1261 within a single rule.
1262 Because the matching occurs using a
1263 bitmask, it takes constant time and dramatically reduces
1264 the complexity of rulesets.
1266 As an example, an address specified as 1.2.3.4/24{128,35-55,89}
1267 or 1.2.3.0/24{128,35-55,89}
1268 will match the following IP addresses:
1270 1.2.3.128, 1.2.3.35 to 1.2.3.55, 1.2.3.89 .
1271 .It Ar addr6-list : ip6-addr Ns Op Ns , Ns Ar addr6-list
1273 A host or subnet specified one of the following ways:
1274 .Bl -tag -width indent
1275 .It Ar numeric-ip | hostname
1276 Matches a single IPv6 address as allowed by
1279 Hostnames are resolved at the time the rule is added to the firewall
1281 .It Ar addr Ns / Ns Ar masklen
1282 Matches all IPv6 addresses with base
1284 (specified as allowed by
1292 No support for sets of IPv6 addresses is provided because IPv6 addresses
1293 are typically random past the initial prefix.
1294 .It Ar ports : Bro Ar port | port Ns \&- Ns Ar port Ns Brc Ns Op , Ns Ar ports
1295 For protocols which support port numbers (such as TCP and UDP), optional
1297 may be specified as one or more ports or port ranges, separated
1298 by commas but no spaces, and an optional
1303 notation specifies a range of ports (including boundaries).
1307 may be used instead of numeric port values.
1308 The length of the port list is limited to 30 ports or ranges,
1309 though one can specify larger ranges by using an
1313 section of the rule.
1317 can be used to escape the dash
1319 character in a service name (from a shell, the backslash must be
1320 typed twice to avoid the shell itself interpreting it as an escape
1323 .Dl "ipfw add count tcp from any ftp\e\e-data-ftp to any"
1325 Fragmented packets which have a non-zero offset (i.e., not the first
1326 fragment) will never match a rule which has one or more port
1330 option for details on matching fragmented packets.
1332 .Ss RULE OPTIONS (MATCH PATTERNS)
1333 Additional match patterns can be used within
1335 Zero or more of these so-called
1337 can be present in a rule, optionally prefixed by the
1339 operand, and possibly grouped into
1342 The following match patterns can be used (listed in alphabetical order):
1343 .Bl -tag -width indent
1344 .It Cm // this is a comment.
1345 Inserts the specified text as a comment in the rule.
1346 Everything following // is considered as a comment and stored in the rule.
1347 You can have comment-only rules, which are listed as having a
1349 action followed by the comment.
1354 Matches only packets generated by a divert socket.
1355 .It Cm diverted-loopback
1356 Matches only packets coming from a divert socket back into the IP stack
1358 .It Cm diverted-output
1359 Matches only packets going from a divert socket back outward to the IP
1360 stack output for delivery.
1361 .It Cm dst-ip Ar ip-address
1362 Matches IPv4 packets whose destination IP is one of the address(es)
1363 specified as argument.
1364 .It Bro Cm dst-ip6 | dst-ipv6 Brc Ar ip6-address
1365 Matches IPv6 packets whose destination IP is one of the address(es)
1366 specified as argument.
1367 .It Cm dst-port Ar ports
1368 Matches IP packets whose destination port is one of the port(s)
1369 specified as argument.
1371 Matches TCP packets that have the RST or ACK bits set.
1372 .It Cm ext6hdr Ar header
1373 Matches IPv6 packets containing the extended header given by
1375 Supported headers are:
1381 any type of Routing Header
1383 Source routing Routing Header Type 0
1385 Mobile IPv6 Routing Header Type 2
1389 IPSec authentication headers
1391 and IPsec encapsulated security payload headers
1393 .It Cm fib Ar fibnum
1394 Matches a packet that has been tagged to use
1395 the given FIB (routing table) number.
1396 .It Cm flow Ar table Ns Pq Ar name Ns Op , Ns Ar value
1397 Search for the flow entry in lookup table
1399 If not found, the match fails.
1400 Otherwise, the match succeeds and
1402 is set to the value extracted from the table.
1404 This option can be useful to quickly dispatch traffic based on
1405 certain packet fields.
1408 section below for more information on lookup tables.
1409 .It Cm flow-id Ar labels
1410 Matches IPv6 packets containing any of the flow labels given in
1413 is a comma separated list of numeric flow labels.
1415 Matches packets that are fragments and not the first
1416 fragment of an IP datagram.
1417 Note that these packets will not have
1418 the next protocol header (e.g.\& TCP, UDP) so options that look into
1419 these headers cannot match.
1421 Matches all TCP or UDP packets sent by or received for a
1425 may be specified by name or number.
1426 .It Cm jail Ar prisonID
1427 Matches all TCP or UDP packets sent by or received for the
1428 jail whos prison ID is
1430 .It Cm icmptypes Ar types
1431 Matches ICMP packets whose ICMP type is in the list
1433 The list may be specified as any combination of
1434 individual types (numeric) separated by commas.
1435 .Em Ranges are not allowed .
1436 The supported ICMP types are:
1440 destination unreachable
1448 router advertisement
1452 time-to-live exceeded
1464 address mask request
1466 and address mask reply
1468 .It Cm icmp6types Ar types
1469 Matches ICMP6 packets whose ICMP6 type is in the list of
1471 The list may be specified as any combination of
1472 individual types (numeric) separated by commas.
1473 .Em Ranges are not allowed .
1475 Matches incoming or outgoing packets, respectively.
1479 are mutually exclusive (in fact,
1483 .It Cm ipid Ar id-list
1484 Matches IPv4 packets whose
1486 field has value included in
1488 which is either a single value or a list of values or ranges
1489 specified in the same way as
1491 .It Cm iplen Ar len-list
1492 Matches IP packets whose total length, including header and data, is
1495 which is either a single value or a list of values or ranges
1496 specified in the same way as
1498 .It Cm ipoptions Ar spec
1499 Matches packets whose IPv4 header contains the comma separated list of
1500 options specified in
1502 The supported IP options are:
1505 (strict source route),
1507 (loose source route),
1509 (record packet route) and
1512 The absence of a particular option may be denoted
1515 .It Cm ipprecedence Ar precedence
1516 Matches IPv4 packets whose precedence field is equal to
1519 Matches packets that have IPSEC history associated with them
1520 (i.e., the packet comes encapsulated in IPSEC, the kernel
1521 has IPSEC support and IPSEC_FILTERTUNNEL option, and can correctly
1524 Note that specifying
1526 is different from specifying
1528 as the latter will only look at the specific IP protocol field,
1529 irrespective of IPSEC kernel support and the validity of the IPSEC data.
1531 Further note that this flag is silently ignored in kernels without
1533 It does not affect rule processing when given and the
1534 rules are handled as if with no
1537 .It Cm iptos Ar spec
1538 Matches IPv4 packets whose
1540 field contains the comma separated list of
1541 service types specified in
1543 The supported IP types of service are:
1546 .Pq Dv IPTOS_LOWDELAY ,
1548 .Pq Dv IPTOS_THROUGHPUT ,
1550 .Pq Dv IPTOS_RELIABILITY ,
1552 .Pq Dv IPTOS_MINCOST ,
1554 .Pq Dv IPTOS_ECN_CE .
1555 The absence of a particular type may be denoted
1558 .It Cm dscp spec Ns Op , Ns Ar spec
1559 Matches IPv4/IPv6 packets whose
1561 field value is contained in
1564 Multiple values can be specified via
1565 the comma separated list.
1566 Value can be one of keywords used in
1568 action or exact number.
1569 .It Cm ipttl Ar ttl-list
1570 Matches IPv4 packets whose time to live is included in
1572 which is either a single value or a list of values or ranges
1573 specified in the same way as
1575 .It Cm ipversion Ar ver
1576 Matches IP packets whose IP version field is
1579 Upon a match, the firewall will create a dynamic rule, whose
1580 default behaviour is to match bidirectional traffic between
1581 source and destination IP/port using the same protocol.
1582 The rule has a limited lifetime (controlled by a set of
1584 variables), and the lifetime is refreshed every time a matching
1587 Matches only layer2 packets, i.e., those passed to
1589 from ether_demux() and ether_output_frame().
1590 .It Cm limit Bro Cm src-addr | src-port | dst-addr | dst-port Brc Ar N
1591 The firewall will only allow
1593 connections with the same
1594 set of parameters as specified in the rule.
1596 of source and destination addresses and ports can be
1599 only IPv4 flows are supported.
1600 .It Cm lookup Bro Cm dst-ip | dst-port | src-ip | src-port | uid | jail Brc Ar name
1601 Search an entry in lookup table
1603 that matches the field specified as argument.
1604 If not found, the match fails.
1605 Otherwise, the match succeeds and
1607 is set to the value extracted from the table.
1609 This option can be useful to quickly dispatch traffic based on
1610 certain packet fields.
1613 section below for more information on lookup tables.
1614 .It Cm { MAC | mac } Ar dst-mac src-mac
1615 Match packets with a given
1619 addresses, specified as the
1621 keyword (matching any MAC address), or six groups of hex digits
1622 separated by colons,
1623 and optionally followed by a mask indicating the significant bits.
1624 The mask may be specified using either of the following methods:
1625 .Bl -enum -width indent
1629 followed by the number of significant bits.
1630 For example, an address with 33 significant bits could be specified as:
1632 .Dl "MAC 10:20:30:40:50:60/33 any"
1636 followed by a bitmask specified as six groups of hex digits separated
1638 For example, an address in which the last 16 bits are significant could
1641 .Dl "MAC 10:20:30:40:50:60&00:00:00:00:ff:ff any"
1643 Note that the ampersand character has a special meaning in many shells
1644 and should generally be escaped.
1646 Note that the order of MAC addresses (destination first,
1648 the same as on the wire, but the opposite of the one used for
1650 .It Cm mac-type Ar mac-type
1651 Matches packets whose Ethernet Type field
1652 corresponds to one of those specified as argument.
1654 is specified in the same way as
1656 (i.e., one or more comma-separated single values or ranges).
1657 You can use symbolic names for known values such as
1658 .Em vlan , ipv4, ipv6 .
1659 Values can be entered as decimal or hexadecimal (if prefixed by 0x),
1660 and they are always printed as hexadecimal (unless the
1662 option is used, in which case symbolic resolution will be attempted).
1663 .It Cm proto Ar protocol
1664 Matches packets with the corresponding IP protocol.
1665 .It Cm recv | xmit | via Brq Ar ifX | Ar if Ns Cm * | Ar table Ns Po Ar name Ns Oo , Ns Ar value Oc Pc | Ar ipno | Ar any
1666 Matches packets received, transmitted or going through,
1667 respectively, the interface specified by exact name
1671 by IP address, or through some interface.
1674 may be used to match interface by its kernel ifindex.
1677 section below for more information on lookup tables.
1681 keyword causes the interface to always be checked.
1688 then only the receive or transmit interface (respectively)
1690 By specifying both, it is possible to match packets based on
1691 both receive and transmit interface, e.g.:
1693 .Dl "ipfw add deny ip from any to any out recv ed0 xmit ed1"
1697 interface can be tested on either incoming or outgoing packets,
1700 interface can only be tested on outgoing packets.
1705 is invalid) whenever
1709 A packet might not have a receive or transmit interface: packets
1710 originating from the local host have no receive interface,
1711 while packets destined for the local host have no transmit
1714 Matches TCP packets that have the SYN bit set but no ACK bit.
1715 This is the short form of
1716 .Dq Li tcpflags\ syn,!ack .
1718 Matches packets that are associated to a local socket and
1719 for which the SO_USER_COOKIE socket option has been set
1720 to a non-zero value.
1721 As a side effect, the value of the
1722 option is made available as
1724 value, which in turn can be used as
1729 .It Cm src-ip Ar ip-address
1730 Matches IPv4 packets whose source IP is one of the address(es)
1731 specified as an argument.
1732 .It Cm src-ip6 Ar ip6-address
1733 Matches IPv6 packets whose source IP is one of the address(es)
1734 specified as an argument.
1735 .It Cm src-port Ar ports
1736 Matches IP packets whose source port is one of the port(s)
1737 specified as argument.
1738 .It Cm tagged Ar tag-list
1739 Matches packets whose tags are included in
1741 which is either a single value or a list of values or ranges
1742 specified in the same way as
1744 Tags can be applied to the packet using
1746 rule action parameter (see it's description for details on tags).
1747 .It Cm tcpack Ar ack
1749 Match if the TCP header acknowledgment number field is set to
1751 .It Cm tcpdatalen Ar tcpdatalen-list
1752 Matches TCP packets whose length of TCP data is
1753 .Ar tcpdatalen-list ,
1754 which is either a single value or a list of values or ranges
1755 specified in the same way as
1757 .It Cm tcpflags Ar spec
1759 Match if the TCP header contains the comma separated list of
1762 The supported TCP flags are:
1771 The absence of a particular flag may be denoted
1774 A rule which contains a
1776 specification can never match a fragmented packet which has
1780 option for details on matching fragmented packets.
1781 .It Cm tcpseq Ar seq
1783 Match if the TCP header sequence number field is set to
1785 .It Cm tcpwin Ar tcpwin-list
1786 Matches TCP packets whose header window field is set to
1788 which is either a single value or a list of values or ranges
1789 specified in the same way as
1791 .It Cm tcpoptions Ar spec
1793 Match if the TCP header contains the comma separated list of
1794 options specified in
1796 The supported TCP options are:
1799 (maximum segment size),
1801 (tcp window advertisement),
1805 (rfc1323 timestamp) and
1807 (rfc1644 t/tcp connection count).
1808 The absence of a particular option may be denoted
1812 Match all TCP or UDP packets sent by or received for a
1816 may be matched by name or identification number.
1818 For incoming packets,
1819 a routing table lookup is done on the packet's source address.
1820 If the interface on which the packet entered the system matches the
1821 outgoing interface for the route,
1823 If the interfaces do not match up,
1824 the packet does not match.
1825 All outgoing packets or packets with no incoming interface match.
1827 The name and functionality of the option is intentionally similar to
1828 the Cisco IOS command:
1830 .Dl ip verify unicast reverse-path
1832 This option can be used to make anti-spoofing rules to reject all
1833 packets with source addresses not from this interface.
1837 For incoming packets,
1838 a routing table lookup is done on the packet's source address.
1839 If a route to the source address exists, but not the default route
1840 or a blackhole/reject route, the packet matches.
1841 Otherwise, the packet does not match.
1842 All outgoing packets match.
1844 The name and functionality of the option is intentionally similar to
1845 the Cisco IOS command:
1847 .Dl ip verify unicast source reachable-via any
1849 This option can be used to make anti-spoofing rules to reject all
1850 packets whose source address is unreachable.
1852 For incoming packets, the packet's source address is checked if it
1853 belongs to a directly connected network.
1854 If the network is directly connected, then the interface the packet
1855 came on in is compared to the interface the network is connected to.
1856 When incoming interface and directly connected interface are not the
1857 same, the packet does not match.
1858 Otherwise, the packet does match.
1859 All outgoing packets match.
1861 This option can be used to make anti-spoofing rules to reject all
1862 packets that pretend to be from a directly connected network but do
1863 not come in through that interface.
1864 This option is similar to but more restricted than
1866 because it engages only on packets with source addresses of directly
1867 connected networks instead of all source addresses.
1870 Lookup tables are useful to handle large sparse sets of
1871 addresses or other search keys (e.g., ports, jail IDs, interface names).
1872 In the rest of this section we will use the term ``key''.
1873 Table name needs to match the following spec:
1875 Tables with the same name can be created in different
1877 However, rule links to the tables in
1880 This behavior can be controlled by
1881 .Va net.inet.ip.fw.tables_sets
1885 section for more information.
1886 There may be up to 65535 different lookup tables.
1888 The following table types are supported:
1889 .Bl -tag -width indent
1890 .It Ar table-type : Ar addr | iface | number | flow
1891 .It Ar table-key : Ar addr Ns Oo / Ns Ar masklen Oc | iface-name | number | flow-spec
1892 .It Ar flow-spec : Ar flow-field Ns Op , Ns Ar flow-spec
1893 .It Ar flow-field : src-ip | proto | src-port | dst-ip | dst-port
1895 matches IPv4 or IPv6 address.
1896 Each entry is represented by an
1897 .Ar addr Ns Op / Ns Ar masklen
1898 and will match all addresses with base
1900 (specified as an IPv4/IPv6 address, or a hostname) and mask width of
1905 is not specified, it defaults to 32 for IPv4 and 128 for IPv6.
1906 When looking up an IP address in a table, the most specific
1909 matches interface names.
1910 Each entry is represented by string treated as interface name.
1911 Wildcards are not supported.
1913 maches protocol ports, uids/gids or jail IDs.
1914 Each entry is represented by 32-bit unsigned integer.
1915 Ranges are not supported.
1917 Matches packet fields specified by
1919 type suboptions with table entries.
1922 Tables require explicit creation via
1926 The following creation options are supported:
1927 .Bl -tag -width indent
1928 .It Ar create-options : Ar create-option | create-options
1929 .It Ar create-option : Cm type Ar table-type | Cm valtype Ar value-mask | Cm algo Ar algo-desc |
1930 .Cm limit Ar number | Cm locked
1936 Table algorithm to use (see below).
1938 Maximum number of items that may be inserted into table.
1940 Restrict any table modifications.
1943 Some of these options may be modified later via
1946 The following options can be changed:
1947 .Bl -tag -width indent
1948 .It Ar modify-options : Ar modify-option | modify-options
1949 .It Ar modify-option : Cm limit Ar number
1951 Alter maximum number of items that may be inserted into table.
1954 Additionally, table can be locked or unlocked using
1962 can be swapped with each other using
1965 Swap may fail if tables limits are set and data exchange
1966 would result in limits hit.
1967 Operation is performed atomically.
1969 One or more entries can be added to a table at once using
1972 Addition of all items are performed atomically.
1973 By default, error in addition of one entry does not influence
1974 addition of other entries. However, non-zero error code is returned
1978 keyword may be specified before
1980 to indicate all-or-none add request.
1982 One or more entries can be removed from a table at once using
1985 By default, error in removal of one entry does not influence
1986 removing of other entries. However, non-zero error code is returned
1989 It may be possible to check what entry will be found on particular
1995 This functionality is optional and may be unsupported in some algorithms.
1997 The following operations can be performed on
2002 .Bl -tag -width indent
2006 Removes all entries.
2008 Shows generic table information.
2010 Shows generic table information and algo-specific data.
2013 The following lookup algorithms are supported:
2014 .Bl -tag -width indent
2015 .It Ar algo-desc : algo-name | "algo-name algo-data"
2016 .It Ar algo-name: Ar addr:radix | addr:hash | iface:array | number:array | flow:hash
2018 Separate Radix trees for IPv4 and IPv6, the same way as the routing table (see
2024 Separate auto-growing hashes for IPv4 and IPv6.
2025 Accepts entries with the same mask length specified initially via
2026 .Cm "addr:hash masks=/v4,/v6"
2027 algorithm creation options.
2028 Assume /32 and /128 masks by default.
2029 Search removes host bits (according to mask) from supplied address and checks
2030 resulting key in appropriate hash.
2031 Mostly optimized for /64 and byte-ranged IPv6 masks.
2033 Array storing sorted indexes for entries which are presented in the system.
2034 Optimized for very fast lookup.
2036 Array storing sorted u32 numbers.
2038 Auto-growing hash storing flow entries.
2039 Search calculates hash on required packet fields and searches for matching
2040 entries in selected bucket.
2045 feature provides the ability to use a value, looked up in the table, as
2046 the argument for a rule action, action parameter or rule option.
2047 This can significantly reduce number of rules in some configurations.
2048 If two tables are used in a rule, the result of the second (destination)
2051 Each record may hold one or more values according to
2053 This mask is set on table creation via
2056 The following value types are supported:
2057 .Bl -tag -width indent
2058 .It Ar value-mask : Ar value-type Ns Op , Ns Ar value-mask
2059 .It Ar value-type : Ar skipto | pipe | fib | nat | dscp | tag | divert |
2060 .Ar netgraph | limit | ipv4
2062 rule number to jump to.
2066 fib number to match/set.
2068 nat number to jump to.
2070 dscp value to match/set.
2072 tag number to match/set.
2074 port number to divert traffic to.
2076 hook number to move packet to.
2078 maximum number of connections.
2080 IPv4 nexthop to fwd packets to.
2085 argument can be used with the following actions:
2086 .Cm nat, pipe , queue, divert, tee, netgraph, ngtee, fwd, skipto, setfib,
2094 action, the user should be aware that the code will walk the ruleset
2095 up to a rule equal to, or past, the given number.
2099 Section for example usage of tables and the tablearg keyword.
2101 Each rule or table belongs to one of 32 different
2104 Set 31 is reserved for the default rule.
2106 By default, rules or tables are put in set 0, unless you use the
2108 attribute when adding a new rule or table.
2109 Sets can be individually and atomically enabled or disabled,
2110 so this mechanism permits an easy way to store multiple configurations
2111 of the firewall and quickly (and atomically) switch between them.
2113 By default, tables from set 0 are referenced when adding rule with
2114 table opcodes regardless of rule set.
2115 This behavior can be changed by setting
2116 .Va net.inet.ip.fw.tables_set
2118 Rule's set will then be used for table references.
2120 The command to enable/disable sets is
2121 .Bd -ragged -offset indent
2123 .Cm set Oo Cm disable Ar number ... Oc Op Cm enable Ar number ...
2130 sections can be specified.
2131 Command execution is atomic on all the sets specified in the command.
2132 By default, all sets are enabled.
2134 When you disable a set, its rules behave as if they do not exist
2135 in the firewall configuration, with only one exception:
2136 .Bd -ragged -offset indent
2137 dynamic rules created from a rule before it had been disabled
2138 will still be active until they expire.
2140 dynamic rules you have to explicitly delete the parent rule
2141 which generated them.
2144 The set number of rules can be changed with the command
2145 .Bd -ragged -offset indent
2148 .Brq Cm rule Ar rule-number | old-set
2152 Also, you can atomically swap two rulesets with the command
2153 .Bd -ragged -offset indent
2155 .Cm set swap Ar first-set second-set
2160 Section on some possible uses of sets of rules.
2161 .Sh STATEFUL FIREWALL
2162 Stateful operation is a way for the firewall to dynamically
2163 create rules for specific flows when packets that
2164 match a given pattern are detected.
2165 Support for stateful
2166 operation comes through the
2167 .Cm check-state , keep-state
2173 Dynamic rules are created when a packet matches a
2177 rule, causing the creation of a
2179 rule which will match all and only packets with
2183 .Em src-ip/src-port dst-ip/dst-port
2188 are used here only to denote the initial match addresses, but they
2189 are completely equivalent afterwards).
2190 Dynamic rules will be checked at the first
2191 .Cm check-state, keep-state
2194 occurrence, and the action performed upon a match will be the same
2195 as in the parent rule.
2197 Note that no additional attributes other than protocol and IP addresses
2198 and ports are checked on dynamic rules.
2200 The typical use of dynamic rules is to keep a closed firewall configuration,
2201 but let the first TCP SYN packet from the inside network install a
2202 dynamic rule for the flow so that packets belonging to that session
2203 will be allowed through the firewall:
2205 .Dl "ipfw add check-state"
2206 .Dl "ipfw add allow tcp from my-subnet to any setup keep-state"
2207 .Dl "ipfw add deny tcp from any to any"
2209 A similar approach can be used for UDP, where an UDP packet coming
2210 from the inside will install a dynamic rule to let the response through
2213 .Dl "ipfw add check-state"
2214 .Dl "ipfw add allow udp from my-subnet to any keep-state"
2215 .Dl "ipfw add deny udp from any to any"
2217 Dynamic rules expire after some time, which depends on the status
2218 of the flow and the setting of some
2222 .Sx SYSCTL VARIABLES
2224 For TCP sessions, dynamic rules can be instructed to periodically
2225 send keepalive packets to refresh the state of the rule when it is
2230 for more examples on how to use dynamic rules.
2231 .Sh TRAFFIC SHAPER (DUMMYNET) CONFIGURATION
2233 is also the user interface for the
2235 traffic shaper, packet scheduler and network emulator, a subsystem that
2236 can artificially queue, delay or drop packets
2237 emulating the behaviour of certain network links
2238 or queueing systems.
2241 operates by first using the firewall to select packets
2242 using any match pattern that can be used in
2245 Matching packets are then passed to either of two
2246 different objects, which implement the traffic regulation:
2247 .Bl -hang -offset XXXX
2253 with given bandwidth and propagation delay,
2254 driven by a FIFO scheduler and a single queue with programmable
2255 queue size and packet loss rate.
2256 Packets are appended to the queue as they come out from
2258 and then transferred in FIFO order to the link at the desired rate.
2262 is an abstraction used to implement packet scheduling
2263 using one of several packet scheduling algorithms.
2266 are first grouped into flows according to a mask on the 5-tuple.
2267 Flows are then passed to the scheduler associated to the
2269 and each flow uses scheduling parameters (weight and others)
2270 as configured in the
2273 A scheduler in turn is connected to an emulated link,
2274 and arbitrates the link's bandwidth among backlogged flows according to
2275 weights and to the features of the scheduling algorithm in use.
2280 can be used to set hard limits to the bandwidth that a flow can use, whereas
2282 can be used to determine how different flows share the available bandwidth.
2284 A graphical representation of the binding of queues,
2285 flows, schedulers and links is below.
2286 .Bd -literal -offset indent
2287 (flow_mask|sched_mask) sched_mask
2288 +---------+ weight Wx +-------------+
2289 | |->-[flow]-->--| |-+
2290 -->--| QUEUE x | ... | | |
2291 | |->-[flow]-->--| SCHEDuler N | |
2293 ... | +--[LINK N]-->--
2294 +---------+ weight Wy | | +--[LINK N]-->--
2295 | |->-[flow]-->--| | |
2296 -->--| QUEUE y | ... | | |
2297 | |->-[flow]-->--| | |
2298 +---------+ +-------------+ |
2301 It is important to understand the role of the SCHED_MASK
2302 and FLOW_MASK, which are configured through the commands
2303 .Dl "ipfw sched N config mask SCHED_MASK ..."
2305 .Dl "ipfw queue X config mask FLOW_MASK ..." .
2307 The SCHED_MASK is used to assign flows to one or more
2308 scheduler instances, one for each
2309 value of the packet's 5-tuple after applying SCHED_MASK.
2310 As an example, using ``src-ip 0xffffff00'' creates one instance
2311 for each /24 destination subnet.
2313 The FLOW_MASK, together with the SCHED_MASK, is used to split
2315 As an example, using
2316 ``src-ip 0x000000ff''
2317 together with the previous SCHED_MASK makes a flow for
2318 each individual source address.
2319 In turn, flows for each /24
2320 subnet will be sent to the same scheduler instance.
2322 The above diagram holds even for the
2324 case, with the only restriction that a
2326 only supports a SCHED_MASK, and forces the use of a FIFO
2327 scheduler (these are for backward compatibility reasons;
2328 in fact, internally, a
2330 pipe is implemented exactly as above).
2332 There are two modes of
2340 mode tries to emulate a real link: the
2342 scheduler ensures that the packet will not leave the pipe faster than it
2343 would on the real link with a given bandwidth.
2346 mode allows certain packets to bypass the
2348 scheduler (if packet flow does not exceed pipe's bandwidth).
2349 This is the reason why the
2351 mode requires less CPU cycles per packet (on average) and packet latency
2352 can be significantly lower in comparison to a real link with the same
2358 mode can be enabled by setting the
2359 .Va net.inet.ip.dummynet.io_fast
2361 variable to a non-zero value.
2363 .Ss PIPE, QUEUE AND SCHEDULER CONFIGURATION
2369 configuration commands are the following:
2370 .Bd -ragged -offset indent
2371 .Cm pipe Ar number Cm config Ar pipe-configuration
2373 .Cm queue Ar number Cm config Ar queue-configuration
2375 .Cm sched Ar number Cm config Ar sched-configuration
2378 The following parameters can be configured for a pipe:
2380 .Bl -tag -width indent -compact
2381 .It Cm bw Ar bandwidth | device
2382 Bandwidth, measured in
2385 .Brq Cm bit/s | Byte/s .
2388 A value of 0 (default) means unlimited bandwidth.
2389 The unit must immediately follow the number, as in
2391 .Dl "ipfw pipe 1 config bw 300Kbit/s"
2393 If a device name is specified instead of a numeric value, as in
2395 .Dl "ipfw pipe 1 config bw tun0"
2397 then the transmit clock is supplied by the specified device.
2398 At the moment only the
2400 device supports this
2401 functionality, for use in conjunction with
2404 .It Cm delay Ar ms-delay
2405 Propagation delay, measured in milliseconds.
2406 The value is rounded to the next multiple of the clock tick
2407 (typically 10ms, but it is a good practice to run kernels
2409 .Dq "options HZ=1000"
2411 the granularity to 1ms or less).
2412 The default value is 0, meaning no delay.
2414 .It Cm burst Ar size
2415 If the data to be sent exceeds the pipe's bandwidth limit
2416 (and the pipe was previously idle), up to
2418 bytes of data are allowed to bypass the
2420 scheduler, and will be sent as fast as the physical link allows.
2421 Any additional data will be transmitted at the rate specified
2425 The burst size depends on how long the pipe has been idle;
2426 the effective burst size is calculated as follows:
2433 .It Cm profile Ar filename
2434 A file specifying the additional overhead incurred in the transmission
2435 of a packet on the link.
2437 Some link types introduce extra delays in the transmission
2438 of a packet, e.g., because of MAC level framing, contention on
2439 the use of the channel, MAC level retransmissions and so on.
2440 From our point of view, the channel is effectively unavailable
2441 for this extra time, which is constant or variable depending
2443 Additionally, packets may be dropped after this
2444 time (e.g., on a wireless link after too many retransmissions).
2445 We can model the additional delay with an empirical curve
2446 that represents its distribution.
2447 .Bd -literal -offset indent
2448 cumulative probability
2458 +-------*------------------->
2461 The empirical curve may have both vertical and horizontal lines.
2462 Vertical lines represent constant delay for a range of
2464 Horizontal lines correspond to a discontinuity in the delay
2465 distribution: the pipe will use the largest delay for a
2468 The file format is the following, with whitespace acting as
2469 a separator and '#' indicating the beginning a comment:
2470 .Bl -tag -width indent
2471 .It Cm name Ar identifier
2472 optional name (listed by "ipfw pipe show")
2473 to identify the delay distribution;
2475 the bandwidth used for the pipe.
2476 If not specified here, it must be present
2477 explicitly as a configuration parameter for the pipe;
2478 .It Cm loss-level Ar L
2479 the probability above which packets are lost.
2480 (0.0 <= L <= 1.0, default 1.0 i.e., no loss);
2482 the number of samples used in the internal
2483 representation of the curve (2..1024; default 100);
2484 .It Cm "delay prob" | "prob delay"
2485 One of these two lines is mandatory and defines
2486 the format of the following lines with data points.
2488 2 or more lines representing points in the curve,
2489 with either delay or probability first, according
2490 to the chosen format.
2491 The unit for delay is milliseconds.
2492 Data points do not need to be sorted.
2493 Also, the number of actual lines can be different
2494 from the value of the "samples" parameter:
2496 utility will sort and interpolate
2497 the curve as needed.
2500 Example of a profile file:
2501 .Bd -literal -offset indent
2506 0 200 # minimum overhead is 200ms
2512 #configuration file end
2516 The following parameters can be configured for a queue:
2518 .Bl -tag -width indent -compact
2519 .It Cm pipe Ar pipe_nr
2520 Connects a queue to the specified pipe.
2521 Multiple queues (with the same or different weights) can be connected to
2522 the same pipe, which specifies the aggregate rate for the set of queues.
2524 .It Cm weight Ar weight
2525 Specifies the weight to be used for flows matching this queue.
2526 The weight must be in the range 1..100, and defaults to 1.
2529 The following case-insensitive parameters can be configured for a
2532 .Bl -tag -width indent -compact
2533 .It Cm type Ar {fifo | wf2q+ | rr | qfq}
2534 specifies the scheduling algorithm to use.
2535 .Bl -tag -width indent -compact
2537 is just a FIFO scheduler (which means that all packets
2538 are stored in the same queue as they arrive to the scheduler).
2539 FIFO has O(1) per-packet time complexity, with very low
2540 constants (estimate 60-80ns on a 2GHz desktop machine)
2541 but gives no service guarantees.
2543 implements the WF2Q+ algorithm, which is a Weighted Fair Queueing
2544 algorithm which permits flows to share bandwidth according to
2546 Note that weights are not priorities; even a flow
2547 with a minuscule weight will never starve.
2548 WF2Q+ has O(log N) per-packet processing cost, where N is the number
2549 of flows, and is the default algorithm used by previous versions
2552 implements the Deficit Round Robin algorithm, which has O(1) processing
2553 costs (roughly, 100-150ns per packet)
2554 and permits bandwidth allocation according to weights, but
2555 with poor service guarantees.
2557 implements the QFQ algorithm, which is a very fast variant of
2558 WF2Q+, with similar service guarantees and O(1) processing
2559 costs (roughly, 200-250ns per packet).
2563 In addition to the type, all parameters allowed for a pipe can also
2564 be specified for a scheduler.
2566 Finally, the following parameters can be configured for both
2569 .Bl -tag -width XXXX -compact
2570 .It Cm buckets Ar hash-table-size
2571 Specifies the size of the hash table used for storing the
2573 Default value is 64 controlled by the
2576 .Va net.inet.ip.dummynet.hash_size ,
2577 allowed range is 16 to 65536.
2579 .It Cm mask Ar mask-specifier
2580 Packets sent to a given pipe or queue by an
2582 rule can be further classified into multiple flows, each of which is then
2586 A flow identifier is constructed by masking the IP addresses,
2587 ports and protocol types as specified with the
2589 options in the configuration of the pipe or queue.
2590 For each different flow identifier, a new pipe or queue is created
2591 with the same parameters as the original object, and matching packets
2596 are used, each flow will get the same bandwidth as defined by the pipe,
2599 are used, each flow will share the parent's pipe bandwidth evenly
2600 with other flows generated by the same queue (note that other queues
2601 with different weights might be connected to the same pipe).
2603 Available mask specifiers are a combination of one or more of the following:
2605 .Cm dst-ip Ar mask ,
2606 .Cm dst-ip6 Ar mask ,
2607 .Cm src-ip Ar mask ,
2608 .Cm src-ip6 Ar mask ,
2609 .Cm dst-port Ar mask ,
2610 .Cm src-port Ar mask ,
2611 .Cm flow-id Ar mask ,
2616 where the latter means all bits in all fields are significant.
2619 When a packet is dropped by a
2621 queue or pipe, the error
2622 is normally reported to the caller routine in the kernel, in the
2623 same way as it happens when a device queue fills up.
2625 option reports the packet as successfully delivered, which can be
2626 needed for some experimental setups where you want to simulate
2627 loss or congestion at a remote router.
2629 .It Cm plr Ar packet-loss-rate
2632 .Ar packet-loss-rate
2633 is a floating-point number between 0 and 1, with 0 meaning no
2634 loss, 1 meaning 100% loss.
2635 The loss rate is internally represented on 31 bits.
2637 .It Cm queue Brq Ar slots | size Ns Cm Kbytes
2642 Default value is 50 slots, which
2643 is the typical queue size for Ethernet devices.
2644 Note that for slow speed links you should keep the queue
2645 size short or your traffic might be affected by a significant
2647 E.g., 50 max-sized ethernet packets (1500 bytes) mean 600Kbit
2648 or 20s of queue on a 30Kbit/s pipe.
2649 Even worse effects can result if you get packets from an
2650 interface with a much larger MTU, e.g.\& the loopback interface
2651 with its 16KB packets.
2655 .Em net.inet.ip.dummynet.pipe_byte_limit
2657 .Em net.inet.ip.dummynet.pipe_slot_limit
2658 control the maximum lengths that can be specified.
2660 .It Cm red | gred Ar w_q Ns / Ns Ar min_th Ns / Ns Ar max_th Ns / Ns Ar max_p
2662 Make use of the RED (Random Early Detection) queue management algorithm.
2667 point numbers between 0 and 1 (inclusive), while
2671 are integer numbers specifying thresholds for queue management
2672 (thresholds are computed in bytes if the queue has been defined
2673 in bytes, in slots otherwise).
2674 The two parameters can also be of the same value if needed. The
2676 also supports the gentle RED variant (gred) and ECN (Explicit Congestion
2677 Notification) as optional. Three
2679 variables can be used to control the RED behaviour:
2680 .Bl -tag -width indent
2681 .It Va net.inet.ip.dummynet.red_lookup_depth
2682 specifies the accuracy in computing the average queue
2683 when the link is idle (defaults to 256, must be greater than zero)
2684 .It Va net.inet.ip.dummynet.red_avg_pkt_size
2685 specifies the expected average packet size (defaults to 512, must be
2687 .It Va net.inet.ip.dummynet.red_max_pkt_size
2688 specifies the expected maximum packet size, only used when queue
2689 thresholds are in bytes (defaults to 1500, must be greater than zero).
2693 When used with IPv6 data,
2695 currently has several limitations.
2696 Information necessary to route link-local packets to an
2697 interface is not available after processing by
2699 so those packets are dropped in the output path.
2700 Care should be taken to ensure that link-local packets are not passed to
2703 Here are some important points to consider when designing your
2707 Remember that you filter both packets going
2711 Most connections need packets going in both directions.
2713 Remember to test very carefully.
2714 It is a good idea to be near the console when doing this.
2715 If you cannot be near the console,
2716 use an auto-recovery script such as the one in
2717 .Pa /usr/share/examples/ipfw/change_rules.sh .
2719 Do not forget the loopback interface.
2724 There are circumstances where fragmented datagrams are unconditionally
2726 TCP packets are dropped if they do not contain at least 20 bytes of
2727 TCP header, UDP packets are dropped if they do not contain a full 8
2728 byte UDP header, and ICMP packets are dropped if they do not contain
2729 4 bytes of ICMP header, enough to specify the ICMP type, code, and
2731 These packets are simply logged as
2733 since there may not be enough good data in the packet to produce a
2734 meaningful log entry.
2736 Another type of packet is unconditionally dropped, a TCP packet with a
2737 fragment offset of one.
2738 This is a valid packet, but it only has one use, to try
2739 to circumvent firewalls.
2740 When logging is enabled, these packets are
2741 reported as being dropped by rule -1.
2743 If you are logged in over a network, loading the
2747 is probably not as straightforward as you would think.
2748 The following command line is recommended:
2749 .Bd -literal -offset indent
2751 ipfw add 32000 allow ip from any to any
2754 Along the same lines, doing an
2755 .Bd -literal -offset indent
2759 in similar surroundings is also a bad idea.
2763 filter list may not be modified if the system security level
2764 is set to 3 or higher
2767 for information on system security levels).
2769 .Sh PACKET DIVERSION
2772 socket bound to the specified port will receive all packets
2773 diverted to that port.
2774 If no socket is bound to the destination port, or if the divert module is
2775 not loaded, or if the kernel was not compiled with divert socket support,
2776 the packets are dropped.
2777 .Sh NETWORK ADDRESS TRANSLATION (NAT)
2779 support in-kernel NAT using the kernel version of
2782 The nat configuration command is the following:
2783 .Bd -ragged -offset indent
2788 .Ar nat-configuration
2792 The following parameters can be configured:
2793 .Bl -tag -width indent
2794 .It Cm ip Ar ip_address
2795 Define an ip address to use for aliasing.
2797 Use ip address of NIC for aliasing, dynamically changing
2798 it if NIC's ip address changes.
2800 Enable logging on this nat instance.
2802 Deny any incoming connection from outside world.
2804 Try to leave the alias port numbers unchanged from
2805 the actual local port numbers.
2807 Traffic on the local network not originating from an
2808 unregistered address spaces will be ignored.
2810 Reset table of the packet aliasing engine on address change.
2812 Reverse the way libalias handles aliasing.
2814 Obey transparent proxy rules only, packet aliasing is not performed.
2816 Skip instance in case of global state lookup (see below).
2819 Some specials value can be supplied instead of
2821 .Bl -tag -width indent
2823 Looks up translation state in all configured nat instances.
2824 If an entry is found, packet is aliased according to that entry.
2825 If no entry was found in any of the instances, packet is passed unchanged,
2826 and no new entry will be created.
2828 .Sx MULTIPLE INSTANCES
2831 for more information.
2833 Uses argument supplied in lookup table.
2836 section below for more information on lookup tables.
2839 To let the packet continue after being (de)aliased, set the sysctl variable
2840 .Va net.inet.ip.fw.one_pass
2842 For more information about aliasing modes, refer to
2846 for some examples about nat usage.
2847 .Ss REDIRECT AND LSNAT SUPPORT IN IPFW
2848 Redirect and LSNAT support follow closely the syntax used in
2852 for some examples on how to do redirect and lsnat.
2853 .Ss SCTP NAT SUPPORT
2854 SCTP nat can be configured in a similar manner to TCP through the
2857 The main difference is that
2859 does not do port translation.
2860 Since the local and global side ports will be the same,
2861 there is no need to specify both.
2862 Ports are redirected as follows:
2863 .Bd -ragged -offset indent
2869 .Cm redirect_port sctp
2870 .Ar ip_address [,addr_list] {[port | port-port] [,ports]}
2876 configuration can be done in real-time through the
2879 All may be changed dynamically, though the hash_table size will only
2884 .Sx SYSCTL VARIABLES
2887 Tunables can be set in
2893 before ipfw module gets loaded.
2894 .Bl -tag -width indent
2895 .It Va net.inet.ip.fw.default_to_accept: No 0
2896 Defines ipfw last rule behavior.
2897 This value overrides
2898 .Cd "options IPFW_DEFAULT_TO_(ACCEPT|DENY)"
2899 from kernel configuration file.
2900 .It Va net.inet.ip.fw.tables_max: No 128
2901 Defines number of tables available in ipfw.
2902 Number cannot exceed 65534.
2904 .Sh SYSCTL VARIABLES
2907 variables controls the behaviour of the firewall and
2909 .Pq Nm dummynet , bridge , sctp nat .
2910 These are shown below together with their default value
2911 (but always check with the
2913 command what value is actually in use) and meaning:
2914 .Bl -tag -width indent
2915 .It Va net.inet.ip.alias.sctp.accept_global_ootb_addip: No 0
2918 responds to receipt of global OOTB ASCONF-AddIP:
2919 .Bl -tag -width indent
2921 No response (unless a partially matching association exists -
2922 ports and vtags match but global address does not)
2925 will accept and process all OOTB global AddIP messages.
2928 Option 1 should never be selected as this forms a security risk.
2930 establish multiple fake associations by sending AddIP messages.
2931 .It Va net.inet.ip.alias.sctp.chunk_proc_limit: No 5
2932 Defines the maximum number of chunks in an SCTP packet that will be
2934 packet that matches an existing association.
2935 This value is enforced to be greater or equal than
2936 .Cm net.inet.ip.alias.sctp.initialising_chunk_proc_limit .
2938 a DoS risk yet setting too low a value may result in
2939 important control chunks in
2940 the packet not being located and parsed.
2941 .It Va net.inet.ip.alias.sctp.error_on_ootb: No 1
2944 responds to any Out-of-the-Blue (OOTB) packets with ErrorM packets.
2945 An OOTB packet is a packet that arrives with no existing association
2948 and is not an INIT or ASCONF-AddIP packet:
2949 .Bl -tag -width indent
2951 ErrorM is never sent in response to OOTB packets.
2953 ErrorM is only sent to OOTB packets received on the local side.
2955 ErrorM is sent to the local side and on the global side ONLY if there is a
2956 partial match (ports and vtags match but the source global IP does not).
2957 This value is only useful if the
2959 is tracking global IP addresses.
2961 ErrorM is sent in response to all OOTB packets on both
2962 the local and global side
2966 At the moment the default is 0, since the ErrorM packet is not yet
2967 supported by most SCTP stacks.
2968 When it is supported, and if not tracking
2969 global addresses, we recommend setting this value to 1 to allow
2970 multi-homed local hosts to function with the
2972 To track global addresses, we recommend setting this value to 2 to
2973 allow global hosts to be informed when they need to (re)send an
2975 Value 3 should never be chosen (except for debugging) as the
2977 will respond to all OOTB global packets (a DoS risk).
2978 .It Va net.inet.ip.alias.sctp.hashtable_size: No 2003
2979 Size of hash tables used for
2981 lookups (100 < prime_number > 1000001).
2984 size for any future created
2986 instance and therefore must be set prior to creating a
2989 The table sizes may be changed to suit specific needs.
2990 If there will be few
2991 concurrent associations, and memory is scarce, you may make these smaller.
2992 If there will be many thousands (or millions) of concurrent associations, you
2993 should make these larger.
2994 A prime number is best for the table size.
2996 update function will adjust your input value to the next highest prime number.
2997 .It Va net.inet.ip.alias.sctp.holddown_time: No 0
2998 Hold association in table for this many seconds after receiving a
3000 This allows endpoints to correct shutdown gracefully if a
3001 shutdown_complete is lost and retransmissions are required.
3002 .It Va net.inet.ip.alias.sctp.init_timer: No 15
3003 Timeout value while waiting for (INIT-ACK|AddIP-ACK).
3004 This value cannot be 0.
3005 .It Va net.inet.ip.alias.sctp.initialising_chunk_proc_limit: No 2
3006 Defines the maximum number of chunks in an SCTP packet that will be parsed when
3007 no existing association exists that matches that packet.
3009 will only be an INIT or ASCONF-AddIP packet.
3010 A higher value may become a DoS
3011 risk as malformed packets can consume processing resources.
3012 .It Va net.inet.ip.alias.sctp.param_proc_limit: No 25
3013 Defines the maximum number of parameters within a chunk that will be
3016 As for other similar sysctl variables, larger values pose a DoS risk.
3017 .It Va net.inet.ip.alias.sctp.log_level: No 0
3018 Level of detail in the system log messages (0 \- minimal, 1 \- event,
3019 2 \- info, 3 \- detail, 4 \- debug, 5 \- max debug).
3021 option in high loss environments.
3022 .It Va net.inet.ip.alias.sctp.shutdown_time: No 15
3023 Timeout value while waiting for SHUTDOWN-COMPLETE.
3024 This value cannot be 0.
3025 .It Va net.inet.ip.alias.sctp.track_global_addresses: No 0
3026 Enables/disables global IP address tracking within the
3029 upper limit on the number of addresses tracked for each association:
3030 .Bl -tag -width indent
3032 Global tracking is disabled
3034 Enables tracking, the maximum number of addresses tracked for each
3035 association is limited to this value
3038 This variable is fully dynamic, the new value will be adopted for all newly
3039 arriving associations, existing associations are treated
3040 as they were previously.
3041 Global tracking will decrease the number of collisions within the
3044 of increased processing load, memory usage, complexity, and possible
3047 problems in complex networks with multiple
3049 We recommend not tracking
3050 global IP addresses, this will still result in a fully functional
3052 .It Va net.inet.ip.alias.sctp.up_timer: No 300
3053 Timeout value to keep an association up with no traffic.
3054 This value cannot be 0.
3055 .It Va net.inet.ip.dummynet.expire : No 1
3056 Lazily delete dynamic pipes/queue once they have no pending traffic.
3057 You can disable this by setting the variable to 0, in which case
3058 the pipes/queues will only be deleted when the threshold is reached.
3059 .It Va net.inet.ip.dummynet.hash_size : No 64
3060 Default size of the hash table used for dynamic pipes/queues.
3061 This value is used when no
3063 option is specified when configuring a pipe/queue.
3064 .It Va net.inet.ip.dummynet.io_fast : No 0
3065 If set to a non-zero value,
3070 operation (see above) is enabled.
3071 .It Va net.inet.ip.dummynet.io_pkt
3072 Number of packets passed to
3074 .It Va net.inet.ip.dummynet.io_pkt_drop
3075 Number of packets dropped by
3077 .It Va net.inet.ip.dummynet.io_pkt_fast
3078 Number of packets bypassed by the
3081 .It Va net.inet.ip.dummynet.max_chain_len : No 16
3082 Target value for the maximum number of pipes/queues in a hash bucket.
3084 .Cm max_chain_len*hash_size
3085 is used to determine the threshold over which empty pipes/queues
3086 will be expired even when
3087 .Cm net.inet.ip.dummynet.expire=0 .
3088 .It Va net.inet.ip.dummynet.red_lookup_depth : No 256
3089 .It Va net.inet.ip.dummynet.red_avg_pkt_size : No 512
3090 .It Va net.inet.ip.dummynet.red_max_pkt_size : No 1500
3091 Parameters used in the computations of the drop probability
3092 for the RED algorithm.
3093 .It Va net.inet.ip.dummynet.pipe_byte_limit : No 1048576
3094 .It Va net.inet.ip.dummynet.pipe_slot_limit : No 100
3095 The maximum queue size that can be specified in bytes or packets.
3096 These limits prevent accidental exhaustion of resources such as mbufs.
3097 If you raise these limits,
3098 you should make sure the system is configured so that sufficient resources
3100 .It Va net.inet.ip.fw.autoinc_step : No 100
3101 Delta between rule numbers when auto-generating them.
3102 The value must be in the range 1..1000.
3103 .It Va net.inet.ip.fw.curr_dyn_buckets : Va net.inet.ip.fw.dyn_buckets
3104 The current number of buckets in the hash table for dynamic rules
3106 .It Va net.inet.ip.fw.debug : No 1
3107 Controls debugging messages produced by
3109 .It Va net.inet.ip.fw.default_rule : No 65535
3110 The default rule number (read-only).
3112 .Nm , the default rule is the last one, so its number
3113 can also serve as the highest number allowed for a rule.
3114 .It Va net.inet.ip.fw.dyn_buckets : No 256
3115 The number of buckets in the hash table for dynamic rules.
3116 Must be a power of 2, up to 65536.
3117 It only takes effect when all dynamic rules have expired, so you
3118 are advised to use a
3120 command to make sure that the hash table is resized.
3121 .It Va net.inet.ip.fw.dyn_count : No 3
3122 Current number of dynamic rules
3124 .It Va net.inet.ip.fw.dyn_keepalive : No 1
3125 Enables generation of keepalive packets for
3127 rules on TCP sessions.
3128 A keepalive is generated to both
3129 sides of the connection every 5 seconds for the last 20
3130 seconds of the lifetime of the rule.
3131 .It Va net.inet.ip.fw.dyn_max : No 8192
3132 Maximum number of dynamic rules.
3133 When you hit this limit, no more dynamic rules can be
3134 installed until old ones expire.
3135 .It Va net.inet.ip.fw.dyn_ack_lifetime : No 300
3136 .It Va net.inet.ip.fw.dyn_syn_lifetime : No 20
3137 .It Va net.inet.ip.fw.dyn_fin_lifetime : No 1
3138 .It Va net.inet.ip.fw.dyn_rst_lifetime : No 1
3139 .It Va net.inet.ip.fw.dyn_udp_lifetime : No 5
3140 .It Va net.inet.ip.fw.dyn_short_lifetime : No 30
3141 These variables control the lifetime, in seconds, of dynamic
3143 Upon the initial SYN exchange the lifetime is kept short,
3144 then increased after both SYN have been seen, then decreased
3145 again during the final FIN exchange or when a RST is received.
3147 .Em dyn_fin_lifetime
3149 .Em dyn_rst_lifetime
3150 must be strictly lower than 5 seconds, the period of
3151 repetition of keepalives.
3152 The firewall enforces that.
3153 .It Va net.inet.ip.fw.dyn_keep_states: No 0
3154 Keep dynamic states on rule/set deletion.
3155 States are relinked to default rule (65535).
3156 This can be handly for ruleset reload.
3157 Turned off by default.
3158 .It Va net.inet.ip.fw.enable : No 1
3159 Enables the firewall.
3160 Setting this variable to 0 lets you run your machine without
3161 firewall even if compiled in.
3162 .It Va net.inet6.ip6.fw.enable : No 1
3163 provides the same functionality as above for the IPv6 case.
3164 .It Va net.inet.ip.fw.one_pass : No 1
3165 When set, the packet exiting from the
3169 node is not passed though the firewall again.
3170 Otherwise, after an action, the packet is
3171 reinjected into the firewall at the next rule.
3172 .It Va net.inet.ip.fw.tables_max : No 128
3173 Maximum number of tables.
3174 .It Va net.inet.ip.fw.verbose : No 1
3175 Enables verbose messages.
3176 .It Va net.inet.ip.fw.verbose_limit : No 0
3177 Limits the number of messages produced by a verbose firewall.
3178 .It Va net.inet6.ip6.fw.deny_unknown_exthdrs : No 1
3179 If enabled packets with unknown IPv6 Extension Headers will be denied.
3180 .It Va net.link.ether.ipfw : No 0
3181 Controls whether layer-2 packets are passed to
3184 .It Va net.link.bridge.ipfw : No 0
3185 Controls whether bridged packets are passed to
3189 .Sh INTERNAL DIAGNOSTICS
3190 There are some commands that may be useful to understand current state
3191 of certain subsystems inside kernel module.
3192 These commands provide debugging output which may change without notice.
3194 Currently the following commands are available as
3197 .Bl -tag -width indent
3199 Lists all interface which are currently tracked by
3201 with their in-kernel status.
3203 List all table lookup algorithms currently available.
3206 There are far too many possible uses of
3208 so this Section will only give a small set of examples.
3210 .Ss BASIC PACKET FILTERING
3211 This command adds an entry which denies all tcp packets from
3212 .Em cracker.evil.org
3213 to the telnet port of
3215 from being forwarded by the host:
3217 .Dl "ipfw add deny tcp from cracker.evil.org to wolf.tambov.su telnet"
3219 This one disallows any connection from the entire cracker's
3222 .Dl "ipfw add deny ip from 123.45.67.0/24 to my.host.org"
3224 A first and efficient way to limit access (not using dynamic rules)
3225 is the use of the following rules:
3227 .Dl "ipfw add allow tcp from any to any established"
3228 .Dl "ipfw add allow tcp from net1 portlist1 to net2 portlist2 setup"
3229 .Dl "ipfw add allow tcp from net3 portlist3 to net3 portlist3 setup"
3231 .Dl "ipfw add deny tcp from any to any"
3233 The first rule will be a quick match for normal TCP packets,
3234 but it will not match the initial SYN packet, which will be
3237 rules only for selected source/destination pairs.
3238 All other SYN packets will be rejected by the final
3242 If you administer one or more subnets, you can take advantage
3243 of the address sets and or-blocks and write extremely
3244 compact rulesets which selectively enable services to blocks
3245 of clients, as below:
3247 .Dl "goodguys=\*q{ 10.1.2.0/24{20,35,66,18} or 10.2.3.0/28{6,3,11} }\*q"
3248 .Dl "badguys=\*q10.1.2.0/24{8,38,60}\*q"
3250 .Dl "ipfw add allow ip from ${goodguys} to any"
3251 .Dl "ipfw add deny ip from ${badguys} to any"
3252 .Dl "... normal policies ..."
3256 option could be used to do automated anti-spoofing by adding the
3257 following to the top of a ruleset:
3259 .Dl "ipfw add deny ip from any to any not verrevpath in"
3261 This rule drops all incoming packets that appear to be coming to the
3262 system on the wrong interface.
3263 For example, a packet with a source
3264 address belonging to a host on a protected internal network would be
3265 dropped if it tried to enter the system from an external interface.
3269 option could be used to do similar but more restricted anti-spoofing
3270 by adding the following to the top of a ruleset:
3272 .Dl "ipfw add deny ip from any to any not antispoof in"
3274 This rule drops all incoming packets that appear to be coming from another
3275 directly connected system but on the wrong interface.
3276 For example, a packet with a source address of
3277 .Li 192.168.0.0/24 ,
3286 option could be used to (re)mark user traffic,
3287 by adding the following to the appropriate place in ruleset:
3289 .Dl "ipfw add setdscp be ip from any to any dscp af11,af21"
3291 In order to protect a site from flood attacks involving fake
3292 TCP packets, it is safer to use dynamic rules:
3294 .Dl "ipfw add check-state"
3295 .Dl "ipfw add deny tcp from any to any established"
3296 .Dl "ipfw add allow tcp from my-net to any setup keep-state"
3298 This will let the firewall install dynamic rules only for
3299 those connection which start with a regular SYN packet coming
3300 from the inside of our network.
3301 Dynamic rules are checked when encountering the first
3310 rule should usually be placed near the beginning of the
3311 ruleset to minimize the amount of work scanning the ruleset.
3312 Your mileage may vary.
3314 To limit the number of connections a user can open
3315 you can use the following type of rules:
3317 .Dl "ipfw add allow tcp from my-net/24 to any setup limit src-addr 10"
3318 .Dl "ipfw add allow tcp from any to me setup limit src-addr 4"
3320 The former (assuming it runs on a gateway) will allow each host
3321 on a /24 network to open at most 10 TCP connections.
3322 The latter can be placed on a server to make sure that a single
3323 client does not use more than 4 simultaneous connections.
3326 stateful rules can be subject to denial-of-service attacks
3327 by a SYN-flood which opens a huge number of dynamic rules.
3328 The effects of such attacks can be partially limited by
3331 variables which control the operation of the firewall.
3333 Here is a good usage of the
3335 command to see accounting records and timestamp information:
3339 or in short form without timestamps:
3343 which is equivalent to:
3347 Next rule diverts all incoming packets from 192.168.2.0/24
3348 to divert port 5000:
3350 .Dl ipfw divert 5000 ip from 192.168.2.0/24 to any in
3352 The following rules show some of the applications of
3356 for simulations and the like.
3358 This rule drops random incoming packets with a probability
3361 .Dl "ipfw add prob 0.05 deny ip from any to any in"
3363 A similar effect can be achieved making use of
3367 .Dl "ipfw add pipe 10 ip from any to any"
3368 .Dl "ipfw pipe 10 config plr 0.05"
3370 We can use pipes to artificially limit bandwidth, e.g.\& on a
3371 machine acting as a router, if we want to limit traffic from
3372 local clients on 192.168.2.0/24 we do:
3374 .Dl "ipfw add pipe 1 ip from 192.168.2.0/24 to any out"
3375 .Dl "ipfw pipe 1 config bw 300Kbit/s queue 50KBytes"
3377 note that we use the
3379 modifier so that the rule is not used twice.
3380 Remember in fact that
3382 rules are checked both on incoming and outgoing packets.
3384 Should we want to simulate a bidirectional link with bandwidth
3385 limitations, the correct way is the following:
3387 .Dl "ipfw add pipe 1 ip from any to any out"
3388 .Dl "ipfw add pipe 2 ip from any to any in"
3389 .Dl "ipfw pipe 1 config bw 64Kbit/s queue 10Kbytes"
3390 .Dl "ipfw pipe 2 config bw 64Kbit/s queue 10Kbytes"
3392 The above can be very useful, e.g.\& if you want to see how
3393 your fancy Web page will look for a residential user who
3394 is connected only through a slow link.
3395 You should not use only one pipe for both directions, unless
3396 you want to simulate a half-duplex medium (e.g.\& AppleTalk,
3398 It is not necessary that both pipes have the same configuration,
3399 so we can also simulate asymmetric links.
3401 Should we want to verify network performance with the RED queue
3402 management algorithm:
3404 .Dl "ipfw add pipe 1 ip from any to any"
3405 .Dl "ipfw pipe 1 config bw 500Kbit/s queue 100 red 0.002/30/80/0.1"
3407 Another typical application of the traffic shaper is to
3408 introduce some delay in the communication.
3409 This can significantly affect applications which do a lot of Remote
3410 Procedure Calls, and where the round-trip-time of the
3411 connection often becomes a limiting factor much more than
3414 .Dl "ipfw add pipe 1 ip from any to any out"
3415 .Dl "ipfw add pipe 2 ip from any to any in"
3416 .Dl "ipfw pipe 1 config delay 250ms bw 1Mbit/s"
3417 .Dl "ipfw pipe 2 config delay 250ms bw 1Mbit/s"
3419 Per-flow queueing can be useful for a variety of purposes.
3420 A very simple one is counting traffic:
3422 .Dl "ipfw add pipe 1 tcp from any to any"
3423 .Dl "ipfw add pipe 1 udp from any to any"
3424 .Dl "ipfw add pipe 1 ip from any to any"
3425 .Dl "ipfw pipe 1 config mask all"
3427 The above set of rules will create queues (and collect
3428 statistics) for all traffic.
3429 Because the pipes have no limitations, the only effect is
3430 collecting statistics.
3431 Note that we need 3 rules, not just the last one, because
3434 tries to match IP packets it will not consider ports, so we
3435 would not see connections on separate ports as different
3438 A more sophisticated example is limiting the outbound traffic
3439 on a net with per-host limits, rather than per-network limits:
3441 .Dl "ipfw add pipe 1 ip from 192.168.2.0/24 to any out"
3442 .Dl "ipfw add pipe 2 ip from any to 192.168.2.0/24 in"
3443 .Dl "ipfw pipe 1 config mask src-ip 0x000000ff bw 200Kbit/s queue 20Kbytes"
3444 .Dl "ipfw pipe 2 config mask dst-ip 0x000000ff bw 200Kbit/s queue 20Kbytes"
3446 In the following example, we need to create several traffic bandwidth
3447 classes and we need different hosts/networks to fall into different classes.
3448 We create one pipe for each class and configure them accordingly.
3449 Then we create a single table and fill it with IP subnets and addresses.
3450 For each subnet/host we set the argument equal to the number of the pipe
3452 Then we classify traffic using a single rule:
3454 .Dl "ipfw pipe 1 config bw 1000Kbyte/s"
3455 .Dl "ipfw pipe 4 config bw 4000Kbyte/s"
3457 .Dl "ipfw table T1 create type addr"
3458 .Dl "ipfw table T1 add 192.168.2.0/24 1"
3459 .Dl "ipfw table T1 add 192.168.0.0/27 4"
3460 .Dl "ipfw table T1 add 192.168.0.2 1"
3462 .Dl "ipfw add pipe tablearg ip from 'table(T1)' to any"
3466 action, the table entries may include hostnames and IP addresses.
3468 .Dl "ipfw table T2 create type addr ftype ip"
3469 .Dl "ipfw table T2 add 192.168.2.0/24 10.23.2.1"
3470 .Dl "ipfw table T21 add 192.168.0.0/27 router1.dmz"
3472 .Dl "ipfw add 100 fwd tablearg ip from any to table(1)"
3474 In the following example per-interface firewall is created:
3476 .Dl "ipfw table IN create type iface valtype skipto,fib"
3477 .Dl "ipfw table IN add vlan20 12000,12"
3478 .Dl "ipfw table IN add vlan30 13000,13"
3479 .Dl "ipfw table OUT create type iface valtype skipto"
3480 .Dl "ipfw table OUT add vlan20 22000"
3481 .Dl "ipfw table OUT add vlan30 23000"
3483 .Dl "ipfw add 100 ipfw setfib tablearg ip from any to any recv 'table(IN)' in"
3484 .Dl "ipfw add 200 ipfw skipto tablearg ip from any to any recv 'table(IN)' in"
3485 .Dl "ipfw add 300 ipfw skipto tablearg ip from any to any xmit 'table(OUT)' out"
3487 The following example illustrate usage of flow tables:
3489 .Dl "ipfw table fl create type flow:flow:src-ip,proto,dst-ip,dst-port"
3490 .Dl "ipfw table fl add 2a02:6b8:77::88,tcp,2a02:6b8:77::99,80 11"
3491 .Dl "ipfw table fl add 10.0.0.1,udp,10.0.0.2,53 12"
3493 .Dl "ipfw add 100 allow ip from any to any flow 'table(fl,11)' recv ix0"
3495 To add a set of rules atomically, e.g.\& set 18:
3497 .Dl "ipfw set disable 18"
3498 .Dl "ipfw add NN set 18 ... # repeat as needed"
3499 .Dl "ipfw set enable 18"
3501 To delete a set of rules atomically the command is simply:
3503 .Dl "ipfw delete set 18"
3505 To test a ruleset and disable it and regain control if something goes wrong:
3507 .Dl "ipfw set disable 18"
3508 .Dl "ipfw add NN set 18 ... # repeat as needed"
3509 .Dl "ipfw set enable 18; echo done; sleep 30 && ipfw set disable 18"
3511 Here if everything goes well, you press control-C before the "sleep"
3512 terminates, and your ruleset will be left active.
3513 Otherwise, e.g.\& if
3514 you cannot access your box, the ruleset will be disabled after
3515 the sleep terminates thus restoring the previous situation.
3517 To show rules of the specific set:
3519 .Dl "ipfw set 18 show"
3521 To show rules of the disabled set:
3523 .Dl "ipfw -S set 18 show"
3525 To clear a specific rule counters of the specific set:
3527 .Dl "ipfw set 18 zero NN"
3529 To delete a specific rule of the specific set:
3531 .Dl "ipfw set 18 delete NN"
3532 .Ss NAT, REDIRECT AND LSNAT
3533 First redirect all the traffic to nat instance 123:
3535 .Dl "ipfw add nat 123 all from any to any"
3537 Then to configure nat instance 123 to alias all the outgoing traffic with ip
3538 192.168.0.123, blocking all incoming connections, trying to keep
3539 same ports on both sides, clearing aliasing table on address change
3540 and keeping a log of traffic/link statistics:
3542 .Dl "ipfw nat 123 config ip 192.168.0.123 log deny_in reset same_ports"
3544 Or to change address of instance 123, aliasing table will be cleared (see
3547 .Dl "ipfw nat 123 config ip 10.0.0.1"
3549 To see configuration of nat instance 123:
3551 .Dl "ipfw nat 123 show config"
3553 To show logs of all the instances in range 111-999:
3555 .Dl "ipfw nat 111-999 show"
3557 To see configurations of all instances:
3559 .Dl "ipfw nat show config"
3561 Or a redirect rule with mixed modes could looks like:
3563 .Dl "ipfw nat 123 config redirect_addr 10.0.0.1 10.0.0.66"
3564 .Dl " redirect_port tcp 192.168.0.1:80 500"
3565 .Dl " redirect_proto udp 192.168.1.43 192.168.1.1"
3566 .Dl " redirect_addr 192.168.0.10,192.168.0.11"
3567 .Dl " 10.0.0.100 # LSNAT"
3568 .Dl " redirect_port tcp 192.168.0.1:80,192.168.0.10:22"
3571 or it could be split in:
3573 .Dl "ipfw nat 1 config redirect_addr 10.0.0.1 10.0.0.66"
3574 .Dl "ipfw nat 2 config redirect_port tcp 192.168.0.1:80 500"
3575 .Dl "ipfw nat 3 config redirect_proto udp 192.168.1.43 192.168.1.1"
3576 .Dl "ipfw nat 4 config redirect_addr 192.168.0.10,192.168.0.11,192.168.0.12"
3578 .Dl "ipfw nat 5 config redirect_port tcp"
3579 .Dl " 192.168.0.1:80,192.168.0.10:22,192.168.0.20:25 500"
3600 utility first appeared in
3605 Stateful extensions were introduced in
3608 was introduced in Summer 2002.
3610 .An Ugen J. S. Antsilevich ,
3611 .An Poul-Henning Kamp ,
3617 API based upon code written by
3621 Dummynet has been introduced by Luigi Rizzo in 1997-1998.
3623 Some early work (1999-2000) on the
3625 traffic shaper supported by Akamba Corp.
3627 The ipfw core (ipfw2) has been completely redesigned and
3628 reimplemented by Luigi Rizzo in summer 2002.
3631 options have been added by various developer over the years.
3634 In-kernel NAT support written by
3635 .An Paolo Pisati Aq Mt piso@FreeBSD.org
3636 as part of a Summer of Code 2005 project.
3640 support has been developed by
3641 .An The Centre for Advanced Internet Architectures (CAIA) Aq http://www.caia.swin.edu.au .
3642 The primary developers and maintainers are David Hayes and Jason But.
3643 For further information visit:
3644 .Aq http://www.caia.swin.edu.au/urp/SONATA
3646 Delay profiles have been developed by Alessandro Cerri and
3647 Luigi Rizzo, supported by the
3648 European Commission within Projects Onelab and Onelab2.
3650 The syntax has grown over the years and sometimes it might be confusing.
3651 Unfortunately, backward compatibility prevents cleaning up mistakes
3652 made in the definition of the syntax.
3656 Misconfiguring the firewall can put your computer in an unusable state,
3657 possibly shutting down network services and requiring console access to
3658 regain control of it.
3660 Incoming packet fragments diverted by
3662 are reassembled before delivery to the socket.
3663 The action used on those packet is the one from the
3664 rule which matches the first fragment of the packet.
3666 Packets diverted to userland, and then reinserted by a userland process
3667 may lose various packet attributes.
3668 The packet source interface name
3669 will be preserved if it is shorter than 8 bytes and the userland process
3670 saves and reuses the sockaddr_in
3673 otherwise, it may be lost.
3674 If a packet is reinserted in this manner, later rules may be incorrectly
3675 applied, making the order of
3677 rules in the rule sequence very important.
3679 Dummynet drops all packets with IPv6 link-local addresses.
3685 may not behave as expected.
3686 In particular, incoming SYN packets may
3687 have no uid or gid associated with them since they do not yet belong
3688 to a TCP connection, and the uid/gid associated with a packet may not
3689 be as expected if the associated process calls
3691 or similar system calls.
3693 Rule syntax is subject to the command line environment and some patterns
3694 may need to be escaped with the backslash character
3695 or quoted appropriately.
3697 Due to the architecture of
3699 ipfw nat is not compatible with the TCP segmentation offloading (TSO).
3700 Thus, to reliably nat your network traffic, please disable TSO
3704 ICMP error messages are not implicitly matched by dynamic rules
3705 for the respective conversations.
3706 To avoid failures of network error detection and path MTU discovery,
3707 ICMP error messages may need to be allowed explicitly through static
3714 actions may lead to confusing behaviour if ruleset has mistakes,
3715 and/or interaction with other subsystems (netgraph, dummynet, etc.) is used.
3716 One possible case for this is packet leaving
3718 in subroutine on the input pass, while later on output encountering unpaired
3721 As the call stack is kept intact after input pass, packet will suddenly
3722 return to the rule number used on input pass, not on output one.
3723 Order of processing should be checked carefully to avoid such mistakes.