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
57 .Oo Cm set Ar N Oc Cm table Ar name Cm modify Ar modify-options
59 .Oo Cm set Ar N Oc Cm table Ar name Cm swap Ar name
61 .Oo Cm set Ar N Oc Cm table Ar name Cm add Ar table-key Op Ar value
63 .Oo Cm set Ar N Oc Cm table Ar name Cm add Op Ar table-key Ar value ...
65 .Oo Cm set Ar N Oc Cm table Ar name Cm atomic add Op Ar table-key Ar value ...
67 .Oo Cm set Ar N Oc Cm table Ar name Cm delete Op Ar table-key ...
69 .Oo Cm set Ar N Oc Cm table Ar name Cm lookup Ar addr
71 .Oo Cm set Ar N Oc Cm table Ar name Cm lock
73 .Oo Cm set Ar N Oc Cm table Ar name Cm unlock
75 .Oo Cm set Ar N Oc Cm table
79 .Oo Cm set Ar N Oc Cm table
83 .Oo Cm set Ar N Oc Cm table
87 .Oo Cm set Ar N Oc Cm table
90 .Ss DUMMYNET CONFIGURATION (TRAFFIC SHAPER AND PACKET SCHEDULER)
92 .Brq Cm pipe | queue | sched
98 .Brq Cm pipe | queue | sched
99 .Brq Cm delete | list | show
108 .Ss STATEFUL IPv6/IPv4 NETWORK ADDRESS AND PROTOCOL TRANSLATION
110 .Oo Cm set Ar N Oc Cm nat64lsn Ar name Cm create Ar create-options
112 .Oo Cm set Ar N Oc Cm nat64lsn Ar name Cm config Ar config-options
114 .Oo Cm set Ar N Oc Cm nat64lsn
119 .Oo Cm set Ar N Oc Cm nat64lsn
123 .Oo Cm set Ar N Oc Cm nat64lsn Ar name Cm stats Op Cm reset
124 .Ss STATELESS IPv6/IPv4 NETWORK ADDRESS AND PROTOCOL TRANSLATION
126 .Oo Cm set Ar N Oc Cm nat64stl Ar name Cm create Ar create-options
128 .Oo Cm set Ar N Oc Cm nat64stl Ar name Cm config Ar config-options
130 .Oo Cm set Ar N Oc Cm nat64stl
134 .Oo Cm set Ar N Oc Cm nat64stl
138 .Oo Cm set Ar N Oc Cm nat64stl Ar name Cm stats Op Cm reset
139 .Ss IPv6-to-IPv6 NETWORK PREFIX TRANSLATION
141 .Oo Cm set Ar N Oc Cm nptv6 Ar name Cm create Ar create-options
143 .Oo Cm set Ar N Oc Cm nptv6
147 .Oo Cm set Ar N Oc Cm nptv6
151 .Oo Cm set Ar N Oc Cm nptv6 Ar name Cm stats Op Cm reset
152 .Ss INTERNAL DIAGNOSTICS
159 .Ss LIST OF RULES AND PREPROCESSING
172 utility is the user interface for controlling the
176 traffic shaper/packet scheduler, and the
177 in-kernel NAT services.
179 A firewall configuration, or
183 numbered from 1 to 65535.
184 Packets are passed to the firewall
185 from a number of different places in the protocol stack
186 (depending on the source and destination of the packet,
187 it is possible for the firewall to be
188 invoked multiple times on the same packet).
189 The packet passed to the firewall is compared
190 against each of the rules in the
193 (multiple rules with the same number are permitted, in which case
194 they are processed in order of insertion).
195 When a match is found, the action corresponding to the
196 matching rule is performed.
198 Depending on the action and certain system settings, packets
199 can be reinjected into the firewall at some rule after the
200 matching one for further processing.
202 A ruleset always includes a
204 rule (numbered 65535) which cannot be modified or deleted,
205 and matches all packets.
206 The action associated with the
212 depending on how the kernel is configured.
214 If the ruleset includes one or more rules with the
221 the firewall will have a
223 behaviour, i.e., upon a match it will create
225 i.e., rules that match packets with the same 5-tuple
226 (protocol, source and destination addresses and ports)
227 as the packet which caused their creation.
228 Dynamic rules, which have a limited lifetime, are checked
229 at the first occurrence of a
234 rule, and are typically used to open the firewall on-demand to
235 legitimate traffic only.
242 for all packets (not only these matched by the rule) but
249 .Sx STATEFUL FIREWALL
252 Sections below for more information on the stateful behaviour of
255 All rules (including dynamic ones) have a few associated counters:
256 a packet count, a byte count, a log count and a timestamp
257 indicating the time of the last match.
258 Counters can be displayed or reset with
262 Each rule belongs to one of 32 different
266 commands to atomically manipulate sets, such as enable,
267 disable, swap sets, move all rules in a set to another
268 one, delete all rules in a set.
269 These can be useful to
270 install temporary configurations, or to test them.
273 for more information on
276 Rules can be added with the
278 command; deleted individually or in groups with the
280 command, and globally (except those in set 31) with the
282 command; displayed, optionally with the content of the
288 Finally, counters can be reset with the
295 The following general options are available when invoking
297 .Bl -tag -width indent
299 Show counter values when listing rules.
302 command implies this option.
304 Only show the action and the comment, not the body of a rule.
308 When entering or showing rules, print them in compact form,
309 i.e., omitting the "ip from any to any" string
310 when this does not carry any additional information.
312 When listing, show dynamic rules in addition to static ones.
316 is specified, also show expired dynamic rules.
318 Run without prompting for confirmation for commands that can cause problems if misused,
321 If there is no tty associated with the process, this is implied.
324 command with this flag ignores possible errors,
325 i.e., nonexistent rule number.
326 And for batched commands execution continues with the next command.
328 When listing a table (see the
330 section below for more information on lookup tables), format values
332 By default, values are shown as integers.
334 Only check syntax of the command strings, without actually passing
337 Try to resolve addresses and service names in output.
339 Be quiet when executing the
349 This is useful when updating rulesets by executing multiple
353 .Ql sh\ /etc/rc.firewall ) ,
354 or by processing a file with many
356 rules across a remote login session.
357 It also stops a table add or delete
358 from failing if the entry already exists or is not present.
360 The reason why this option may be important is that
361 for some of these actions,
363 may print a message; if the action results in blocking the
364 traffic to the remote client,
365 the remote login session will be closed
366 and the rest of the ruleset will not be processed.
367 Access to the console would then be required to recover.
369 When listing rules, show the
371 each rule belongs to.
372 If this flag is not specified, disabled rules will not be
375 When listing pipes, sort according to one of the four
376 counters (total or current packets or bytes).
378 When listing, show last match timestamp converted with ctime().
380 When listing, show last match timestamp as seconds from the epoch.
381 This form can be more convenient for postprocessing by scripts.
383 .Ss LIST OF RULES AND PREPROCESSING
384 To ease configuration, rules can be put into a file which is
387 as shown in the last synopsis line.
391 The file will be read line by line and applied as arguments to the
395 Optionally, a preprocessor can be specified using
399 is to be piped through.
400 Useful preprocessors include
406 does not start with a slash
408 as its first character, the usual
410 name search is performed.
411 Care should be taken with this in environments where not all
412 file systems are mounted (yet) by the time
414 is being run (e.g.\& when they are mounted over NFS).
417 has been specified, any additional arguments are passed on to the preprocessor
419 This allows for flexible configuration files (like conditionalizing
420 them on the local hostname) and the use of macros to centralize
421 frequently required arguments like IP addresses.
422 .Ss TRAFFIC SHAPER CONFIGURATION
428 commands are used to configure the traffic shaper and packet scheduler.
430 .Sx TRAFFIC SHAPER (DUMMYNET) CONFIGURATION
431 Section below for details.
433 If the world and the kernel get out of sync the
435 ABI may break, preventing you from being able to add any rules.
436 This can adversely affect the booting process.
441 to temporarily disable the firewall to regain access to the network,
442 allowing you to fix the problem.
444 A packet is checked against the active ruleset in multiple places
445 in the protocol stack, under control of several sysctl variables.
446 These places and variables are shown below, and it is important to
447 have this picture in mind in order to design a correct ruleset.
448 .Bd -literal -offset indent
451 +----------->-----------+
453 [ip(6)_input] [ip(6)_output] net.inet(6).ip(6).fw.enable=1
456 [ether_demux] [ether_output_frame] net.link.ether.ipfw=1
458 +-->--[bdg_forward]-->--+ net.link.bridge.ipfw=1
464 times the same packet goes through the firewall can
465 vary between 0 and 4 depending on packet source and
466 destination, and system configuration.
468 Note that as packets flow through the stack, headers can be
469 stripped or added to it, and so they may or may not be available
471 E.g., incoming packets will include the MAC header when
475 but the same packets will have the MAC header stripped off when
482 Also note that each packet is always checked against the complete ruleset,
483 irrespective of the place where the check occurs, or the source of the packet.
484 If a rule contains some match patterns or actions which are not valid
485 for the place of invocation (e.g.\& trying to match a MAC header within
489 the match pattern will not match, but a
491 operator in front of such patterns
495 match on those packets.
496 It is thus the responsibility of
497 the programmer, if necessary, to write a suitable ruleset to
498 differentiate among the possible places.
500 rules can be useful here, as an example:
501 .Bd -literal -offset indent
502 # packets from ether_demux or bdg_forward
503 ipfw add 10 skipto 1000 all from any to any layer2 in
504 # packets from ip_input
505 ipfw add 10 skipto 2000 all from any to any not layer2 in
506 # packets from ip_output
507 ipfw add 10 skipto 3000 all from any to any not layer2 out
508 # packets from ether_output_frame
509 ipfw add 10 skipto 4000 all from any to any layer2 out
512 (yes, at the moment there is no way to differentiate between
513 ether_demux and bdg_forward).
515 Also note that only actions
524 frames and all other actions act as if they were
527 Full set of actions is supported for IP packets without
532 action does not divert
536 In general, each keyword or argument must be provided as
537 a separate command line argument, with no leading or trailing
539 Keywords are case-sensitive, whereas arguments may
540 or may not be case-sensitive depending on their nature
541 (e.g.\& uid's are, hostnames are not).
543 Some arguments (e.g., port or address lists) are comma-separated
545 In this case, spaces after commas ',' are allowed to make
546 the line more readable.
547 You can also put the entire
548 command (including flags) into a single argument.
549 E.g., the following forms are equivalent:
550 .Bd -literal -offset indent
551 ipfw -q add deny src-ip 10.0.0.0/24,127.0.0.1/8
552 ipfw -q add deny src-ip 10.0.0.0/24, 127.0.0.1/8
553 ipfw "-q add deny src-ip 10.0.0.0/24, 127.0.0.1/8"
556 The format of firewall rules is the following:
557 .Bd -ragged -offset indent
560 .Op Cm set Ar set_number
561 .Op Cm prob Ar match_probability
563 .Op Cm log Op Cm logamount Ar number
573 where the body of the rule specifies which information is used
574 for filtering packets, among the following:
576 .Bl -tag -width "Source and dest. addresses and ports" -offset XXX -compact
577 .It Layer-2 header fields
579 .It IPv4 and IPv6 Protocol
580 SCTP, TCP, UDP, ICMP, etc.
581 .It Source and dest. addresses and ports
585 .It Transmit and receive interface
587 .It Misc. IP header fields
588 Version, type of service, datagram length, identification,
589 fragment flag (non-zero IP offset),
592 .It IPv6 Extension headers
593 Fragmentation, Hop-by-Hop options,
594 Routing Headers, Source routing rthdr0, Mobile IPv6 rthdr2, IPSec options.
596 .It Misc. TCP header fields
597 TCP flags (SYN, FIN, ACK, RST, etc.),
598 sequence number, acknowledgment number,
606 When the packet can be associated with a local socket.
608 Whether a packet came from a divert socket (e.g.,
610 .It Fib annotation state
611 Whether a packet has been tagged for using a specific FIB (routing table)
612 in future forwarding decisions.
615 Note that some of the above information, e.g.\& source MAC or IP addresses and
616 TCP/UDP ports, can be easily spoofed, so filtering on those fields
617 alone might not guarantee the desired results.
618 .Bl -tag -width indent
620 Each rule is associated with a
622 in the range 1..65535, with the latter reserved for the
625 Rules are checked sequentially by rule number.
626 Multiple rules can have the same number, in which case they are
627 checked (and listed) according to the order in which they have
629 If a rule is entered without specifying a number, the kernel will
630 assign one in such a way that the rule becomes the last one
634 Automatic rule numbers are assigned by incrementing the last
635 non-default rule number by the value of the sysctl variable
636 .Ar net.inet.ip.fw.autoinc_step
637 which defaults to 100.
638 If this is not possible (e.g.\& because we would go beyond the
639 maximum allowed rule number), the number of the last
640 non-default value is used instead.
641 .It Cm set Ar set_number
642 Each rule is associated with a
645 Sets can be individually disabled and enabled, so this parameter
646 is of fundamental importance for atomic ruleset manipulation.
647 It can be also used to simplify deletion of groups of rules.
648 If a rule is entered without specifying a set number,
651 Set 31 is special in that it cannot be disabled,
652 and rules in set 31 are not deleted by the
654 command (but you can delete them with the
655 .Nm ipfw delete set 31
657 Set 31 is also used for the
660 .It Cm prob Ar match_probability
661 A match is only declared with the specified probability
662 (floating point number between 0 and 1).
663 This can be useful for a number of applications such as
664 random packet drop or
667 to simulate the effect of multiple paths leading to out-of-order
670 Note: this condition is checked before any other condition, including
677 .It Cm log Op Cm logamount Ar number
678 Packets matching a rule with the
680 keyword will be made available for logging in two ways:
681 if the sysctl variable
682 .Va net.inet.ip.fw.verbose
683 is set to 0 (default), one can use
688 This pseudo interface can be created after a boot
689 manually by using the following command:
690 .Bd -literal -offset indent
691 # ifconfig ipfw0 create
694 Or, automatically at boot time by adding the following
698 .Bd -literal -offset indent
702 There is no overhead if no
704 is attached to the pseudo interface.
707 .Va net.inet.ip.fw.verbose
708 is set to 1, packets will be logged to
712 facility up to a maximum of
717 is specified, the limit is taken from the sysctl variable
718 .Va net.inet.ip.fw.verbose_limit .
719 In both cases, a value of 0 means unlimited logging.
721 Once the limit is reached, logging can be re-enabled by
722 clearing the logging counter or the packet counter for that entry, see the
726 Note: logging is done after all other packet matching conditions
727 have been successfully verified, and before performing the final
728 action (accept, deny, etc.) on the packet.
730 When a packet matches a rule with the
732 keyword, the numeric tag for the given
734 in the range 1..65534 will be attached to the packet.
735 The tag acts as an internal marker (it is not sent out over
736 the wire) that can be used to identify these packets later on.
737 This can be used, for example, to provide trust between interfaces
738 and to start doing policy-based filtering.
739 A packet can have multiple tags at the same time.
740 Tags are "sticky", meaning once a tag is applied to a packet by a
741 matching rule it exists until explicit removal.
742 Tags are kept with the packet everywhere within the kernel, but are
743 lost when packet leaves the kernel, for example, on transmitting
744 packet out to the network or sending packet to a
748 To check for previously applied tags, use the
751 To delete previously applied tag, use the
755 Note: since tags are kept with the packet everywhere in kernelspace,
756 they can be set and unset anywhere in the kernel network subsystem
759 facility), not only by means of the
765 For example, there can be a specialized
767 node doing traffic analyzing and tagging for later inspecting
769 .It Cm untag Ar number
770 When a packet matches a rule with the
772 keyword, the tag with the number
774 is searched among the tags attached to this packet and,
775 if found, removed from it.
776 Other tags bound to packet, if present, are left untouched.
778 When a packet matches a rule with the
780 keyword, the ALTQ identifier for the given
785 Note that this ALTQ tag is only meaningful for packets going "out" of IPFW,
786 and not being rejected or going to divert sockets.
787 Note that if there is insufficient memory at the time the packet is
788 processed, it will not be tagged, so it is wise to make your ALTQ
789 "default" queue policy account for this.
792 rules match a single packet, only the first one adds the ALTQ classification
794 In doing so, traffic may be shaped by using
795 .Cm count Cm altq Ar queue
796 rules for classification early in the ruleset, then later applying
797 the filtering decision.
802 rules may come later and provide the actual filtering decisions in
803 addition to the fallback ALTQ tag.
807 to set up the queues before IPFW will be able to look them up by name,
808 and if the ALTQ disciplines are rearranged, the rules in containing the
809 queue identifiers in the kernel will likely have gone stale and need
811 Stale queue identifiers will probably result in misclassification.
813 All system ALTQ processing can be turned on or off via
818 .Cm disable Ar altq .
820 .Va net.inet.ip.fw.one_pass
821 is irrelevant to ALTQ traffic shaping, as the actual rule action is followed
822 always after adding an ALTQ tag.
825 A rule can be associated with one of the following actions, which
826 will be executed when the packet matches the body of the rule.
827 .Bl -tag -width indent
828 .It Cm allow | accept | pass | permit
829 Allow packets that match rule.
830 The search terminates.
831 .It Cm check-state Op Ar :flowname | Cm :any
832 Checks the packet against the dynamic ruleset.
833 If a match is found, execute the action associated with
834 the rule which generated this dynamic rule, otherwise
835 move to the next rule.
838 rules do not have a body.
841 rule is found, the dynamic ruleset is checked at the first
848 is symbolic name assigned to dynamic rule by
853 can be used to ignore states flowname when matching.
856 keyword is special name used for compatibility with old rulesets.
858 Update counters for all packets that match rule.
859 The search continues with the next rule.
861 Discard packets that match this rule.
862 The search terminates.
863 .It Cm divert Ar port
864 Divert packets that match this rule to the
868 The search terminates.
869 .It Cm fwd | forward Ar ipaddr | tablearg Ns Op , Ns Ar port
870 Change the next-hop on matching packets to
872 which can be an IP address or a host name.
873 The next hop can also be supplied by the last table
874 looked up for the packet by using the
876 keyword instead of an explicit address.
877 The search terminates if this rule matches.
881 is a local address, then matching packets will be forwarded to
883 (or the port number in the packet if one is not specified in the rule)
884 on the local machine.
888 is not a local address, then the port number
889 (if specified) is ignored, and the packet will be
890 forwarded to the remote address, using the route as found in
891 the local routing table for that IP.
895 rule will not match layer-2 packets (those received
896 on ether_input, ether_output, or bridged).
900 action does not change the contents of the packet at all.
901 In particular, the destination address remains unmodified, so
902 packets forwarded to another system will usually be rejected by that system
903 unless there is a matching rule on that system to capture them.
904 For packets forwarded locally,
905 the local address of the socket will be
906 set to the original destination address of the packet.
909 entry look rather weird but is intended for
910 use with transparent proxy servers.
911 .It Cm nat Ar nat_nr | tablearg
914 (for network address translation, address redirect, etc.):
916 .Sx NETWORK ADDRESS TRANSLATION (NAT)
917 Section for further information.
918 .It Cm nat64lsn Ar name
919 Pass packet to a stateful NAT64 instance (for IPv6/IPv4 network address and
920 protocol translation): see the
921 .Sx IPv6/IPv4 NETWORK ADDRESS AND PROTOCOL TRANSLATION
922 Section for further information.
923 .It Cm nat64stl Ar name
924 Pass packet to a stateless NAT64 instance (for IPv6/IPv4 network address and
925 protocol translation): see the
926 .Sx IPv6/IPv4 NETWORK ADDRESS AND PROTOCOL TRANSLATION
927 Section for further information.
929 Pass packet to a NPTv6 instance (for IPv6-to-IPv6 network prefix translation):
931 .Sx IPv6-to-IPv6 NETWORK PREFIX TRANSLATION (NPTv6)
932 Section for further information.
933 .It Cm pipe Ar pipe_nr
937 (for bandwidth limitation, delay, etc.).
939 .Sx TRAFFIC SHAPER (DUMMYNET) CONFIGURATION
940 Section for further information.
941 The search terminates; however, on exit from the pipe and if
945 .Va net.inet.ip.fw.one_pass
946 is not set, the packet is passed again to the firewall code
947 starting from the next rule.
948 .It Cm queue Ar queue_nr
952 (for bandwidth limitation using WF2Q+).
958 Discard packets that match this rule, and if the
959 packet is a TCP packet, try to send a TCP reset (RST) notice.
960 The search terminates.
962 Discard packets that match this rule, and if the
963 packet is a TCP packet, try to send a TCP reset (RST) notice.
964 The search terminates.
965 .It Cm skipto Ar number | tablearg
966 Skip all subsequent rules numbered less than
968 The search continues with the first rule numbered
971 It is possible to use the
973 keyword with a skipto for a
975 skipto. Skipto may work either in O(log(N)) or in O(1) depending
976 on amount of memory and/or sysctl variables.
979 section for more details.
980 .It Cm call Ar number | tablearg
981 The current rule number is saved in the internal stack and
982 ruleset processing continues with the first rule numbered
985 If later a rule with the
987 action is encountered, the processing returns to the first rule
990 rule plus one or higher
991 (the same behaviour as with packets returning from
996 This could be used to make somewhat like an assembly language
998 calls to rules with common checks for different interfaces, etc.
1000 Rule with any number could be called, not just forward jumps as with
1002 So, to prevent endless loops in case of mistakes, both
1006 actions don't do any jumps and simply go to the next rule if memory
1007 cannot be allocated or stack overflowed/underflowed.
1009 Internally stack for rule numbers is implemented using
1011 facility and currently has size of 16 entries.
1012 As mbuf tags are lost when packet leaves the kernel,
1014 should not be used in subroutines to avoid endless loops
1015 and other undesired effects.
1017 Takes rule number saved to internal stack by the last
1019 action and returns ruleset processing to the first rule
1020 with number greater than number of corresponding
1023 See description of the
1025 action for more details.
1031 and thus are unconditional, but
1033 command-line utility currently requires every action except
1036 While it is sometimes useful to return only on some packets,
1037 usually you want to print just
1040 A workaround for this is to use new syntax and
1043 .Bd -literal -offset indent
1044 # Add a rule without actual body
1045 ipfw add 2999 return via any
1047 # List rules without "from any to any" part
1051 This cosmetic annoyance may be fixed in future releases.
1053 Send a copy of packets matching this rule to the
1055 socket bound to port
1057 The search continues with the next rule.
1058 .It Cm unreach Ar code
1059 Discard packets that match this rule, and try to send an ICMP
1060 unreachable notice with code
1064 is a number from 0 to 255, or one of these aliases:
1065 .Cm net , host , protocol , port ,
1066 .Cm needfrag , srcfail , net-unknown , host-unknown ,
1067 .Cm isolated , net-prohib , host-prohib , tosnet ,
1068 .Cm toshost , filter-prohib , host-precedence
1070 .Cm precedence-cutoff .
1071 The search terminates.
1072 .It Cm unreach6 Ar code
1073 Discard packets that match this rule, and try to send an ICMPv6
1074 unreachable notice with code
1078 is a number from 0, 1, 3 or 4, or one of these aliases:
1079 .Cm no-route, admin-prohib, address
1082 The search terminates.
1083 .It Cm netgraph Ar cookie
1084 Divert packet into netgraph with given
1086 The search terminates.
1087 If packet is later returned from netgraph it is either
1088 accepted or continues with the next rule, depending on
1089 .Va net.inet.ip.fw.one_pass
1091 .It Cm ngtee Ar cookie
1092 A copy of packet is diverted into netgraph, original
1093 packet continues with the next rule.
1096 for more information on
1101 .It Cm setfib Ar fibnum | tablearg
1102 The packet is tagged so as to use the FIB (routing table)
1104 in any subsequent forwarding decisions.
1105 In the current implementation, this is limited to the values 0 through 15, see
1107 Processing continues at the next rule.
1108 It is possible to use the
1110 keyword with setfib.
1111 If the tablearg value is not within the compiled range of fibs,
1112 the packet's fib is set to 0.
1113 .It Cm setdscp Ar DSCP | number | tablearg
1114 Set specified DiffServ codepoint for an IPv4/IPv6 packet.
1115 Processing continues at the next rule.
1116 Supported values are:
1162 Additionally, DSCP value can be specified by number (0..64).
1163 It is also possible to use the
1165 keyword with setdscp.
1166 If the tablearg value is not within the 0..64 range, lower 6 bits of supplied
1168 .It Cm tcp-setmss Ar mss
1169 Set the Maximum Segment Size (MSS) in the TCP segment to value
1173 should be loaded or kernel should have
1174 .Cm options IPFIREWALL_PMOD
1175 to be able use this action.
1176 This command does not change a packet if original MSS value is lower than
1178 Both TCP over IPv4 and over IPv6 are supported.
1179 Regardless of matched a packet or not by the
1181 rule, the search continues with the next rule.
1183 Queue and reassemble IPv4 fragments.
1184 If the packet is not fragmented, counters are updated and
1185 processing continues with the next rule.
1186 If the packet is the last logical fragment, the packet is reassembled and, if
1187 .Va net.inet.ip.fw.one_pass
1188 is set to 0, processing continues with the next rule.
1189 Otherwise, the packet is allowed to pass and the search terminates.
1190 If the packet is a fragment in the middle of a logical group of fragments,
1192 processing stops immediately.
1194 Fragment handling can be tuned via
1195 .Va net.inet.ip.maxfragpackets
1197 .Va net.inet.ip.maxfragsperpacket
1198 which limit, respectively, the maximum number of processable
1199 fragments (default: 800) and
1200 the maximum number of fragments per packet (default: 16).
1202 NOTA BENE: since fragments do not contain port numbers,
1203 they should be avoided with the
1206 Alternatively, direction-based (like
1210 ) and source-based (like
1212 ) match patterns can be used to select fragments.
1214 Usually a simple rule like:
1215 .Bd -literal -offset indent
1216 # reassemble incoming fragments
1217 ipfw add reass all from any to any in
1220 is all you need at the beginning of your ruleset.
1222 Discard packets that match this rule, and if the packet is an SCTP packet,
1223 try to send an SCTP packet containing an ABORT chunk.
1224 The search terminates.
1226 Discard packets that match this rule, and if the packet is an SCTP packet,
1227 try to send an SCTP packet containing an ABORT chunk.
1228 The search terminates.
1231 The body of a rule contains zero or more patterns (such as
1232 specific source and destination addresses or ports,
1233 protocol options, incoming or outgoing interfaces, etc.)
1234 that the packet must match in order to be recognised.
1235 In general, the patterns are connected by (implicit)
1237 operators -- i.e., all must match in order for the
1239 Individual patterns can be prefixed by the
1241 operator to reverse the result of the match, as in
1243 .Dl "ipfw add 100 allow ip from not 1.2.3.4 to any"
1245 Additionally, sets of alternative match patterns
1247 can be constructed by putting the patterns in
1248 lists enclosed between parentheses ( ) or braces { }, and
1251 operator as follows:
1253 .Dl "ipfw add 100 allow ip from { x or not y or z } to any"
1255 Only one level of parentheses is allowed.
1256 Beware that most shells have special meanings for parentheses
1257 or braces, so it is advisable to put a backslash \\ in front of them
1258 to prevent such interpretations.
1260 The body of a rule must in general include a source and destination
1264 can be used in various places to specify that the content of
1265 a required field is irrelevant.
1267 The rule body has the following format:
1268 .Bd -ragged -offset indent
1269 .Op Ar proto Cm from Ar src Cm to Ar dst
1273 The first part (proto from src to dst) is for backward
1274 compatibility with earlier versions of
1278 any match pattern (including MAC headers, IP protocols,
1279 addresses and ports) can be specified in the
1283 Rule fields have the following meaning:
1284 .Bl -tag -width indent
1285 .It Ar proto : protocol | Cm { Ar protocol Cm or ... }
1286 .It Ar protocol : Oo Cm not Oc Ar protocol-name | protocol-number
1287 An IP protocol specified by number or name
1288 (for a complete list see
1289 .Pa /etc/protocols ) ,
1290 or one of the following keywords:
1291 .Bl -tag -width indent
1293 Matches IPv4 packets.
1295 Matches IPv6 packets.
1304 option will be treated as inner protocol.
1312 .Cm { Ar protocol Cm or ... }
1315 is provided for convenience only but its use is deprecated.
1316 .It Ar src No and Ar dst : Bro Cm addr | Cm { Ar addr Cm or ... } Brc Op Oo Cm not Oc Ar ports
1317 An address (or a list, see below)
1318 optionally followed by
1324 with multiple addresses) is provided for convenience only and
1325 its use is discouraged.
1326 .It Ar addr : Oo Cm not Oc Bro
1327 .Cm any | me | me6 |
1328 .Cm table Ns Pq Ar name Ns Op , Ns Ar value
1329 .Ar | addr-list | addr-set
1331 .Bl -tag -width indent
1333 matches any IP address.
1335 matches any IP address configured on an interface in the system.
1337 matches any IPv6 address configured on an interface in the system.
1338 The address list is evaluated at the time the packet is
1340 .It Cm table Ns Pq Ar name Ns Op , Ns Ar value
1341 Matches any IPv4 or IPv6 address for which an entry exists in the lookup table
1343 If an optional 32-bit unsigned
1345 is also specified, an entry will match only if it has this value.
1348 section below for more information on lookup tables.
1350 .It Ar addr-list : ip-addr Ns Op Ns , Ns Ar addr-list
1352 A host or subnet address specified in one of the following ways:
1353 .Bl -tag -width indent
1354 .It Ar numeric-ip | hostname
1355 Matches a single IPv4 address, specified as dotted-quad or a hostname.
1356 Hostnames are resolved at the time the rule is added to the firewall list.
1357 .It Ar addr Ns / Ns Ar masklen
1358 Matches all addresses with base
1360 (specified as an IP address, a network number, or a hostname)
1364 As an example, 1.2.3.4/25 or 1.2.3.0/25 will match
1365 all IP numbers from 1.2.3.0 to 1.2.3.127 .
1366 .It Ar addr Ns : Ns Ar mask
1367 Matches all addresses with base
1369 (specified as an IP address, a network number, or a hostname)
1372 specified as a dotted quad.
1373 As an example, 1.2.3.4:255.0.255.0 or 1.0.3.0:255.0.255.0 will match
1375 This form is advised only for non-contiguous
1377 It is better to resort to the
1378 .Ar addr Ns / Ns Ar masklen
1379 format for contiguous masks, which is more compact and less
1382 .It Ar addr-set : addr Ns Oo Ns / Ns Ar masklen Oc Ns Cm { Ns Ar list Ns Cm }
1383 .It Ar list : Bro Ar num | num-num Brc Ns Op Ns , Ns Ar list
1384 Matches all addresses with base address
1386 (specified as an IP address, a network number, or a hostname)
1387 and whose last byte is in the list between braces { } .
1388 Note that there must be no spaces between braces and
1389 numbers (spaces after commas are allowed).
1390 Elements of the list can be specified as single entries
1394 field is used to limit the size of the set of addresses,
1395 and can have any value between 24 and 32.
1397 it will be assumed as 24.
1399 This format is particularly useful to handle sparse address sets
1400 within a single rule.
1401 Because the matching occurs using a
1402 bitmask, it takes constant time and dramatically reduces
1403 the complexity of rulesets.
1405 As an example, an address specified as 1.2.3.4/24{128,35-55,89}
1406 or 1.2.3.0/24{128,35-55,89}
1407 will match the following IP addresses:
1409 1.2.3.128, 1.2.3.35 to 1.2.3.55, 1.2.3.89 .
1410 .It Ar addr6-list : ip6-addr Ns Op Ns , Ns Ar addr6-list
1412 A host or subnet specified one of the following ways:
1413 .Bl -tag -width indent
1414 .It Ar numeric-ip | hostname
1415 Matches a single IPv6 address as allowed by
1418 Hostnames are resolved at the time the rule is added to the firewall
1420 .It Ar addr Ns / Ns Ar masklen
1421 Matches all IPv6 addresses with base
1423 (specified as allowed by
1429 .It Ar addr Ns / Ns Ar mask
1430 Matches all IPv6 addresses with base
1432 (specified as allowed by
1437 specified as allowed by
1439 As an example, fe::640:0:0/ffff::ffff:ffff:0:0 will match
1441 This form is advised only for non-contiguous
1443 It is better to resort to the
1444 .Ar addr Ns / Ns Ar masklen
1445 format for contiguous masks, which is more compact and less
1449 No support for sets of IPv6 addresses is provided because IPv6 addresses
1450 are typically random past the initial prefix.
1451 .It Ar ports : Bro Ar port | port Ns \&- Ns Ar port Ns Brc Ns Op , Ns Ar ports
1452 For protocols which support port numbers (such as SCTP, TCP and UDP), optional
1454 may be specified as one or more ports or port ranges, separated
1455 by commas but no spaces, and an optional
1460 notation specifies a range of ports (including boundaries).
1464 may be used instead of numeric port values.
1465 The length of the port list is limited to 30 ports or ranges,
1466 though one can specify larger ranges by using an
1470 section of the rule.
1474 can be used to escape the dash
1476 character in a service name (from a shell, the backslash must be
1477 typed twice to avoid the shell itself interpreting it as an escape
1480 .Dl "ipfw add count tcp from any ftp\e\e-data-ftp to any"
1482 Fragmented packets which have a non-zero offset (i.e., not the first
1483 fragment) will never match a rule which has one or more port
1487 option for details on matching fragmented packets.
1489 .Ss RULE OPTIONS (MATCH PATTERNS)
1490 Additional match patterns can be used within
1492 Zero or more of these so-called
1494 can be present in a rule, optionally prefixed by the
1496 operand, and possibly grouped into
1499 The following match patterns can be used (listed in alphabetical order):
1500 .Bl -tag -width indent
1501 .It Cm // this is a comment.
1502 Inserts the specified text as a comment in the rule.
1503 Everything following // is considered as a comment and stored in the rule.
1504 You can have comment-only rules, which are listed as having a
1506 action followed by the comment.
1510 .It Cm defer-immediate-action | defer-action
1511 A rule with this option will not perform normal action
1512 upon a match. This option is intended to be used with
1516 as the dynamic rule, created but ignored on match, will work
1521 .Cm defer-immediate-action
1522 create a dynamic rule and continue with the next rule without actually
1523 performing the action part of this rule. When the rule is later activated
1524 via the state table, the action is performed as usual.
1526 Matches only packets generated by a divert socket.
1527 .It Cm diverted-loopback
1528 Matches only packets coming from a divert socket back into the IP stack
1530 .It Cm diverted-output
1531 Matches only packets going from a divert socket back outward to the IP
1532 stack output for delivery.
1533 .It Cm dst-ip Ar ip-address
1534 Matches IPv4 packets whose destination IP is one of the address(es)
1535 specified as argument.
1536 .It Bro Cm dst-ip6 | dst-ipv6 Brc Ar ip6-address
1537 Matches IPv6 packets whose destination IP is one of the address(es)
1538 specified as argument.
1539 .It Cm dst-port Ar ports
1540 Matches IP packets whose destination port is one of the port(s)
1541 specified as argument.
1543 Matches TCP packets that have the RST or ACK bits set.
1544 .It Cm ext6hdr Ar header
1545 Matches IPv6 packets containing the extended header given by
1547 Supported headers are:
1553 any type of Routing Header
1555 Source routing Routing Header Type 0
1557 Mobile IPv6 Routing Header Type 2
1561 IPSec authentication headers
1563 and IPsec encapsulated security payload headers
1565 .It Cm fib Ar fibnum
1566 Matches a packet that has been tagged to use
1567 the given FIB (routing table) number.
1568 .It Cm flow Ar table Ns Pq Ar name Ns Op , Ns Ar value
1569 Search for the flow entry in lookup table
1571 If not found, the match fails.
1572 Otherwise, the match succeeds and
1574 is set to the value extracted from the table.
1576 This option can be useful to quickly dispatch traffic based on
1577 certain packet fields.
1580 section below for more information on lookup tables.
1581 .It Cm flow-id Ar labels
1582 Matches IPv6 packets containing any of the flow labels given in
1585 is a comma separated list of numeric flow labels.
1587 Matches packets that are fragments and not the first
1588 fragment of an IP datagram.
1589 Note that these packets will not have
1590 the next protocol header (e.g.\& TCP, UDP) so options that look into
1591 these headers cannot match.
1593 Matches all TCP or UDP packets sent by or received for a
1597 may be specified by name or number.
1599 Matches all TCP or UDP packets sent by or received for the
1600 jail whose ID or name is
1602 .It Cm icmptypes Ar types
1603 Matches ICMP packets whose ICMP type is in the list
1605 The list may be specified as any combination of
1606 individual types (numeric) separated by commas.
1607 .Em Ranges are not allowed .
1608 The supported ICMP types are:
1612 destination unreachable
1620 router advertisement
1624 time-to-live exceeded
1636 address mask request
1638 and address mask reply
1640 .It Cm icmp6types Ar types
1641 Matches ICMP6 packets whose ICMP6 type is in the list of
1643 The list may be specified as any combination of
1644 individual types (numeric) separated by commas.
1645 .Em Ranges are not allowed .
1647 Matches incoming or outgoing packets, respectively.
1651 are mutually exclusive (in fact,
1655 .It Cm ipid Ar id-list
1656 Matches IPv4 packets whose
1658 field has value included in
1660 which is either a single value or a list of values or ranges
1661 specified in the same way as
1663 .It Cm iplen Ar len-list
1664 Matches IP packets whose total length, including header and data, is
1667 which is either a single value or a list of values or ranges
1668 specified in the same way as
1670 .It Cm ipoptions Ar spec
1671 Matches packets whose IPv4 header contains the comma separated list of
1672 options specified in
1674 The supported IP options are:
1677 (strict source route),
1679 (loose source route),
1681 (record packet route) and
1684 The absence of a particular option may be denoted
1687 .It Cm ipprecedence Ar precedence
1688 Matches IPv4 packets whose precedence field is equal to
1691 Matches packets that have IPSEC history associated with them
1692 (i.e., the packet comes encapsulated in IPSEC, the kernel
1693 has IPSEC support, and can correctly decapsulate it).
1695 Note that specifying
1697 is different from specifying
1699 as the latter will only look at the specific IP protocol field,
1700 irrespective of IPSEC kernel support and the validity of the IPSEC data.
1702 Further note that this flag is silently ignored in kernels without
1704 It does not affect rule processing when given and the
1705 rules are handled as if with no
1708 .It Cm iptos Ar spec
1709 Matches IPv4 packets whose
1711 field contains the comma separated list of
1712 service types specified in
1714 The supported IP types of service are:
1717 .Pq Dv IPTOS_LOWDELAY ,
1719 .Pq Dv IPTOS_THROUGHPUT ,
1721 .Pq Dv IPTOS_RELIABILITY ,
1723 .Pq Dv IPTOS_MINCOST ,
1725 .Pq Dv IPTOS_ECN_CE .
1726 The absence of a particular type may be denoted
1729 .It Cm dscp spec Ns Op , Ns Ar spec
1730 Matches IPv4/IPv6 packets whose
1732 field value is contained in
1735 Multiple values can be specified via
1736 the comma separated list.
1737 Value can be one of keywords used in
1739 action or exact number.
1740 .It Cm ipttl Ar ttl-list
1741 Matches IPv4 packets whose time to live is included in
1743 which is either a single value or a list of values or ranges
1744 specified in the same way as
1746 .It Cm ipversion Ar ver
1747 Matches IP packets whose IP version field is
1749 .It Cm keep-state Op Ar :flowname
1750 Upon a match, the firewall will create a dynamic rule, whose
1751 default behaviour is to match bidirectional traffic between
1752 source and destination IP/port using the same protocol.
1753 The rule has a limited lifetime (controlled by a set of
1755 variables), and the lifetime is refreshed every time a matching
1759 is used to assign additional to addresses, ports and protocol parameter
1760 to dynamic rule. It can be used for more accurate matching by
1765 keyword is special name used for compatibility with old rulesets.
1767 Matches only layer2 packets, i.e., those passed to
1769 from ether_demux() and ether_output_frame().
1770 .It Cm limit Bro Cm src-addr | src-port | dst-addr | dst-port Brc Ar N Op Ar :flowname
1771 The firewall will only allow
1773 connections with the same
1774 set of parameters as specified in the rule.
1776 of source and destination addresses and ports can be
1778 .It Cm lookup Bro Cm dst-ip | dst-port | src-ip | src-port | uid | jail Brc Ar name
1779 Search an entry in lookup table
1781 that matches the field specified as argument.
1782 If not found, the match fails.
1783 Otherwise, the match succeeds and
1785 is set to the value extracted from the table.
1787 This option can be useful to quickly dispatch traffic based on
1788 certain packet fields.
1791 section below for more information on lookup tables.
1792 .It Cm { MAC | mac } Ar dst-mac src-mac
1793 Match packets with a given
1797 addresses, specified as the
1799 keyword (matching any MAC address), or six groups of hex digits
1800 separated by colons,
1801 and optionally followed by a mask indicating the significant bits.
1802 The mask may be specified using either of the following methods:
1803 .Bl -enum -width indent
1807 followed by the number of significant bits.
1808 For example, an address with 33 significant bits could be specified as:
1810 .Dl "MAC 10:20:30:40:50:60/33 any"
1814 followed by a bitmask specified as six groups of hex digits separated
1816 For example, an address in which the last 16 bits are significant could
1819 .Dl "MAC 10:20:30:40:50:60&00:00:00:00:ff:ff any"
1821 Note that the ampersand character has a special meaning in many shells
1822 and should generally be escaped.
1824 Note that the order of MAC addresses (destination first,
1826 the same as on the wire, but the opposite of the one used for
1828 .It Cm mac-type Ar mac-type
1829 Matches packets whose Ethernet Type field
1830 corresponds to one of those specified as argument.
1832 is specified in the same way as
1834 (i.e., one or more comma-separated single values or ranges).
1835 You can use symbolic names for known values such as
1836 .Em vlan , ipv4, ipv6 .
1837 Values can be entered as decimal or hexadecimal (if prefixed by 0x),
1838 and they are always printed as hexadecimal (unless the
1840 option is used, in which case symbolic resolution will be attempted).
1841 .It Cm proto Ar protocol
1842 Matches packets with the corresponding IP protocol.
1844 Upon a match, the firewall will create a dynamic rule as if
1847 However, this option doesn't imply an implicit
1851 .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
1852 Matches packets received, transmitted or going through,
1853 respectively, the interface specified by exact name
1857 by IP address, or through some interface.
1860 may be used to match interface by its kernel ifindex.
1863 section below for more information on lookup tables.
1867 keyword causes the interface to always be checked.
1874 then only the receive or transmit interface (respectively)
1876 By specifying both, it is possible to match packets based on
1877 both receive and transmit interface, e.g.:
1879 .Dl "ipfw add deny ip from any to any out recv ed0 xmit ed1"
1883 interface can be tested on either incoming or outgoing packets,
1886 interface can only be tested on outgoing packets.
1891 is invalid) whenever
1895 A packet might not have a receive or transmit interface: packets
1896 originating from the local host have no receive interface,
1897 while packets destined for the local host have no transmit
1899 .It Cm set-limit Bro Cm src-addr | src-port | dst-addr | dst-port Brc Ar N
1902 but does not have an implicit
1906 Matches TCP packets that have the SYN bit set but no ACK bit.
1907 This is the short form of
1908 .Dq Li tcpflags\ syn,!ack .
1910 Matches packets that are associated to a local socket and
1911 for which the SO_USER_COOKIE socket option has been set
1912 to a non-zero value.
1913 As a side effect, the value of the
1914 option is made available as
1916 value, which in turn can be used as
1921 .It Cm src-ip Ar ip-address
1922 Matches IPv4 packets whose source IP is one of the address(es)
1923 specified as an argument.
1924 .It Cm src-ip6 Ar ip6-address
1925 Matches IPv6 packets whose source IP is one of the address(es)
1926 specified as an argument.
1927 .It Cm src-port Ar ports
1928 Matches IP packets whose source port is one of the port(s)
1929 specified as argument.
1930 .It Cm tagged Ar tag-list
1931 Matches packets whose tags are included in
1933 which is either a single value or a list of values or ranges
1934 specified in the same way as
1936 Tags can be applied to the packet using
1938 rule action parameter (see it's description for details on tags).
1939 .It Cm tcpack Ar ack
1941 Match if the TCP header acknowledgment number field is set to
1943 .It Cm tcpdatalen Ar tcpdatalen-list
1944 Matches TCP packets whose length of TCP data is
1945 .Ar tcpdatalen-list ,
1946 which is either a single value or a list of values or ranges
1947 specified in the same way as
1949 .It Cm tcpflags Ar spec
1951 Match if the TCP header contains the comma separated list of
1954 The supported TCP flags are:
1963 The absence of a particular flag may be denoted
1966 A rule which contains a
1968 specification can never match a fragmented packet which has
1972 option for details on matching fragmented packets.
1973 .It Cm tcpseq Ar seq
1975 Match if the TCP header sequence number field is set to
1977 .It Cm tcpwin Ar tcpwin-list
1978 Matches TCP packets whose header window field is set to
1980 which is either a single value or a list of values or ranges
1981 specified in the same way as
1983 .It Cm tcpoptions Ar spec
1985 Match if the TCP header contains the comma separated list of
1986 options specified in
1988 The supported TCP options are:
1991 (maximum segment size),
1993 (tcp window advertisement),
1997 (rfc1323 timestamp) and
1999 (rfc1644 t/tcp connection count).
2000 The absence of a particular option may be denoted
2004 Match all TCP or UDP packets sent by or received for a
2008 may be matched by name or identification number.
2010 For incoming packets,
2011 a routing table lookup is done on the packet's source address.
2012 If the interface on which the packet entered the system matches the
2013 outgoing interface for the route,
2015 If the interfaces do not match up,
2016 the packet does not match.
2017 All outgoing packets or packets with no incoming interface match.
2019 The name and functionality of the option is intentionally similar to
2020 the Cisco IOS command:
2022 .Dl ip verify unicast reverse-path
2024 This option can be used to make anti-spoofing rules to reject all
2025 packets with source addresses not from this interface.
2029 For incoming packets,
2030 a routing table lookup is done on the packet's source address.
2031 If a route to the source address exists, but not the default route
2032 or a blackhole/reject route, the packet matches.
2033 Otherwise, the packet does not match.
2034 All outgoing packets match.
2036 The name and functionality of the option is intentionally similar to
2037 the Cisco IOS command:
2039 .Dl ip verify unicast source reachable-via any
2041 This option can be used to make anti-spoofing rules to reject all
2042 packets whose source address is unreachable.
2044 For incoming packets, the packet's source address is checked if it
2045 belongs to a directly connected network.
2046 If the network is directly connected, then the interface the packet
2047 came on in is compared to the interface the network is connected to.
2048 When incoming interface and directly connected interface are not the
2049 same, the packet does not match.
2050 Otherwise, the packet does match.
2051 All outgoing packets match.
2053 This option can be used to make anti-spoofing rules to reject all
2054 packets that pretend to be from a directly connected network but do
2055 not come in through that interface.
2056 This option is similar to but more restricted than
2058 because it engages only on packets with source addresses of directly
2059 connected networks instead of all source addresses.
2062 Lookup tables are useful to handle large sparse sets of
2063 addresses or other search keys (e.g., ports, jail IDs, interface names).
2064 In the rest of this section we will use the term ``key''.
2065 Table name needs to match the following spec:
2067 Tables with the same name can be created in different
2069 However, rule links to the tables in
2072 This behavior can be controlled by
2073 .Va net.inet.ip.fw.tables_sets
2077 section for more information.
2078 There may be up to 65535 different lookup tables.
2080 The following table types are supported:
2081 .Bl -tag -width indent
2082 .It Ar table-type : Ar addr | iface | number | flow
2083 .It Ar table-key : Ar addr Ns Oo / Ns Ar masklen Oc | iface-name | number | flow-spec
2084 .It Ar flow-spec : Ar flow-field Ns Op , Ns Ar flow-spec
2085 .It Ar flow-field : src-ip | proto | src-port | dst-ip | dst-port
2087 matches IPv4 or IPv6 address.
2088 Each entry is represented by an
2089 .Ar addr Ns Op / Ns Ar masklen
2090 and will match all addresses with base
2092 (specified as an IPv4/IPv6 address, or a hostname) and mask width of
2097 is not specified, it defaults to 32 for IPv4 and 128 for IPv6.
2098 When looking up an IP address in a table, the most specific
2101 matches interface names.
2102 Each entry is represented by string treated as interface name.
2103 Wildcards are not supported.
2105 maches protocol ports, uids/gids or jail IDs.
2106 Each entry is represented by 32-bit unsigned integer.
2107 Ranges are not supported.
2109 Matches packet fields specified by
2111 type suboptions with table entries.
2114 Tables require explicit creation via
2118 The following creation options are supported:
2119 .Bl -tag -width indent
2120 .It Ar create-options : Ar create-option | create-options
2121 .It Ar create-option : Cm type Ar table-type | Cm valtype Ar value-mask | Cm algo Ar algo-desc |
2122 .Cm limit Ar number | Cm locked
2128 Table algorithm to use (see below).
2130 Maximum number of items that may be inserted into table.
2132 Restrict any table modifications.
2135 Some of these options may be modified later via
2138 The following options can be changed:
2139 .Bl -tag -width indent
2140 .It Ar modify-options : Ar modify-option | modify-options
2141 .It Ar modify-option : Cm limit Ar number
2143 Alter maximum number of items that may be inserted into table.
2146 Additionally, table can be locked or unlocked using
2154 can be swapped with each other using
2157 Swap may fail if tables limits are set and data exchange
2158 would result in limits hit.
2159 Operation is performed atomically.
2161 One or more entries can be added to a table at once using
2164 Addition of all items are performed atomically.
2165 By default, error in addition of one entry does not influence
2166 addition of other entries. However, non-zero error code is returned
2170 keyword may be specified before
2172 to indicate all-or-none add request.
2174 One or more entries can be removed from a table at once using
2177 By default, error in removal of one entry does not influence
2178 removing of other entries. However, non-zero error code is returned
2181 It may be possible to check what entry will be found on particular
2187 This functionality is optional and may be unsupported in some algorithms.
2189 The following operations can be performed on
2194 .Bl -tag -width indent
2198 Removes all entries.
2200 Shows generic table information.
2202 Shows generic table information and algo-specific data.
2205 The following lookup algorithms are supported:
2206 .Bl -tag -width indent
2207 .It Ar algo-desc : algo-name | "algo-name algo-data"
2208 .It Ar algo-name: Ar addr:radix | addr:hash | iface:array | number:array | flow:hash
2210 Separate Radix trees for IPv4 and IPv6, the same way as the routing table (see
2216 Separate auto-growing hashes for IPv4 and IPv6.
2217 Accepts entries with the same mask length specified initially via
2218 .Cm "addr:hash masks=/v4,/v6"
2219 algorithm creation options.
2220 Assume /32 and /128 masks by default.
2221 Search removes host bits (according to mask) from supplied address and checks
2222 resulting key in appropriate hash.
2223 Mostly optimized for /64 and byte-ranged IPv6 masks.
2225 Array storing sorted indexes for entries which are presented in the system.
2226 Optimized for very fast lookup.
2228 Array storing sorted u32 numbers.
2230 Auto-growing hash storing flow entries.
2231 Search calculates hash on required packet fields and searches for matching
2232 entries in selected bucket.
2237 feature provides the ability to use a value, looked up in the table, as
2238 the argument for a rule action, action parameter or rule option.
2239 This can significantly reduce number of rules in some configurations.
2240 If two tables are used in a rule, the result of the second (destination)
2243 Each record may hold one or more values according to
2245 This mask is set on table creation via
2248 The following value types are supported:
2249 .Bl -tag -width indent
2250 .It Ar value-mask : Ar value-type Ns Op , Ns Ar value-mask
2251 .It Ar value-type : Ar skipto | pipe | fib | nat | dscp | tag | divert |
2252 .Ar netgraph | limit | ipv4
2254 rule number to jump to.
2258 fib number to match/set.
2260 nat number to jump to.
2262 dscp value to match/set.
2264 tag number to match/set.
2266 port number to divert traffic to.
2268 hook number to move packet to.
2270 maximum number of connections.
2272 IPv4 nexthop to fwd packets to.
2274 IPv6 nexthop to fwd packets to.
2279 argument can be used with the following actions:
2280 .Cm nat, pipe , queue, divert, tee, netgraph, ngtee, fwd, skipto, setfib,
2288 action, the user should be aware that the code will walk the ruleset
2289 up to a rule equal to, or past, the given number.
2293 Section for example usage of tables and the tablearg keyword.
2295 Each rule or table belongs to one of 32 different
2298 Set 31 is reserved for the default rule.
2300 By default, rules or tables are put in set 0, unless you use the
2302 attribute when adding a new rule or table.
2303 Sets can be individually and atomically enabled or disabled,
2304 so this mechanism permits an easy way to store multiple configurations
2305 of the firewall and quickly (and atomically) switch between them.
2307 By default, tables from set 0 are referenced when adding rule with
2308 table opcodes regardless of rule set.
2309 This behavior can be changed by setting
2310 .Va net.inet.ip.fw.tables_sets
2312 Rule's set will then be used for table references.
2314 The command to enable/disable sets is
2315 .Bd -ragged -offset indent
2317 .Cm set Oo Cm disable Ar number ... Oc Op Cm enable Ar number ...
2324 sections can be specified.
2325 Command execution is atomic on all the sets specified in the command.
2326 By default, all sets are enabled.
2328 When you disable a set, its rules behave as if they do not exist
2329 in the firewall configuration, with only one exception:
2330 .Bd -ragged -offset indent
2331 dynamic rules created from a rule before it had been disabled
2332 will still be active until they expire.
2334 dynamic rules you have to explicitly delete the parent rule
2335 which generated them.
2338 The set number of rules can be changed with the command
2339 .Bd -ragged -offset indent
2342 .Brq Cm rule Ar rule-number | old-set
2346 Also, you can atomically swap two rulesets with the command
2347 .Bd -ragged -offset indent
2349 .Cm set swap Ar first-set second-set
2354 Section on some possible uses of sets of rules.
2355 .Sh STATEFUL FIREWALL
2356 Stateful operation is a way for the firewall to dynamically
2357 create rules for specific flows when packets that
2358 match a given pattern are detected.
2359 Support for stateful
2360 operation comes through the
2361 .Cm check-state , keep-state , record-state , limit
2367 Dynamic rules are created when a packet matches a
2373 rule, causing the creation of a
2375 rule which will match all and only packets with
2379 .Em src-ip/src-port dst-ip/dst-port
2384 are used here only to denote the initial match addresses, but they
2385 are completely equivalent afterwards).
2391 This name is used in matching together with addresses, ports and protocol.
2392 Dynamic rules will be checked at the first
2393 .Cm check-state, keep-state
2396 occurrence, and the action performed upon a match will be the same
2397 as in the parent rule.
2399 Note that no additional attributes other than protocol and IP addresses
2400 and ports and :flowname are checked on dynamic rules.
2402 The typical use of dynamic rules is to keep a closed firewall configuration,
2403 but let the first TCP SYN packet from the inside network install a
2404 dynamic rule for the flow so that packets belonging to that session
2405 will be allowed through the firewall:
2407 .Dl "ipfw add check-state :OUTBOUND"
2408 .Dl "ipfw add allow tcp from my-subnet to any setup keep-state :OUTBOUND"
2409 .Dl "ipfw add deny tcp from any to any"
2411 A similar approach can be used for UDP, where an UDP packet coming
2412 from the inside will install a dynamic rule to let the response through
2415 .Dl "ipfw add check-state :OUTBOUND"
2416 .Dl "ipfw add allow udp from my-subnet to any keep-state :OUTBOUND"
2417 .Dl "ipfw add deny udp from any to any"
2419 Dynamic rules expire after some time, which depends on the status
2420 of the flow and the setting of some
2424 .Sx SYSCTL VARIABLES
2426 For TCP sessions, dynamic rules can be instructed to periodically
2427 send keepalive packets to refresh the state of the rule when it is
2432 for more examples on how to use dynamic rules.
2433 .Sh TRAFFIC SHAPER (DUMMYNET) CONFIGURATION
2435 is also the user interface for the
2437 traffic shaper, packet scheduler and network emulator, a subsystem that
2438 can artificially queue, delay or drop packets
2439 emulating the behaviour of certain network links
2440 or queueing systems.
2443 operates by first using the firewall to select packets
2444 using any match pattern that can be used in
2447 Matching packets are then passed to either of two
2448 different objects, which implement the traffic regulation:
2449 .Bl -hang -offset XXXX
2455 with given bandwidth and propagation delay,
2456 driven by a FIFO scheduler and a single queue with programmable
2457 queue size and packet loss rate.
2458 Packets are appended to the queue as they come out from
2460 and then transferred in FIFO order to the link at the desired rate.
2464 is an abstraction used to implement packet scheduling
2465 using one of several packet scheduling algorithms.
2468 are first grouped into flows according to a mask on the 5-tuple.
2469 Flows are then passed to the scheduler associated to the
2471 and each flow uses scheduling parameters (weight and others)
2472 as configured in the
2475 A scheduler in turn is connected to an emulated link,
2476 and arbitrates the link's bandwidth among backlogged flows according to
2477 weights and to the features of the scheduling algorithm in use.
2482 can be used to set hard limits to the bandwidth that a flow can use, whereas
2484 can be used to determine how different flows share the available bandwidth.
2486 A graphical representation of the binding of queues,
2487 flows, schedulers and links is below.
2488 .Bd -literal -offset indent
2489 (flow_mask|sched_mask) sched_mask
2490 +---------+ weight Wx +-------------+
2491 | |->-[flow]-->--| |-+
2492 -->--| QUEUE x | ... | | |
2493 | |->-[flow]-->--| SCHEDuler N | |
2495 ... | +--[LINK N]-->--
2496 +---------+ weight Wy | | +--[LINK N]-->--
2497 | |->-[flow]-->--| | |
2498 -->--| QUEUE y | ... | | |
2499 | |->-[flow]-->--| | |
2500 +---------+ +-------------+ |
2503 It is important to understand the role of the SCHED_MASK
2504 and FLOW_MASK, which are configured through the commands
2505 .Dl "ipfw sched N config mask SCHED_MASK ..."
2507 .Dl "ipfw queue X config mask FLOW_MASK ..." .
2509 The SCHED_MASK is used to assign flows to one or more
2510 scheduler instances, one for each
2511 value of the packet's 5-tuple after applying SCHED_MASK.
2512 As an example, using ``src-ip 0xffffff00'' creates one instance
2513 for each /24 destination subnet.
2515 The FLOW_MASK, together with the SCHED_MASK, is used to split
2517 As an example, using
2518 ``src-ip 0x000000ff''
2519 together with the previous SCHED_MASK makes a flow for
2520 each individual source address.
2521 In turn, flows for each /24
2522 subnet will be sent to the same scheduler instance.
2524 The above diagram holds even for the
2526 case, with the only restriction that a
2528 only supports a SCHED_MASK, and forces the use of a FIFO
2529 scheduler (these are for backward compatibility reasons;
2530 in fact, internally, a
2532 pipe is implemented exactly as above).
2534 There are two modes of
2542 mode tries to emulate a real link: the
2544 scheduler ensures that the packet will not leave the pipe faster than it
2545 would on the real link with a given bandwidth.
2548 mode allows certain packets to bypass the
2550 scheduler (if packet flow does not exceed pipe's bandwidth).
2551 This is the reason why the
2553 mode requires less CPU cycles per packet (on average) and packet latency
2554 can be significantly lower in comparison to a real link with the same
2560 mode can be enabled by setting the
2561 .Va net.inet.ip.dummynet.io_fast
2563 variable to a non-zero value.
2565 .Ss PIPE, QUEUE AND SCHEDULER CONFIGURATION
2571 configuration commands are the following:
2572 .Bd -ragged -offset indent
2573 .Cm pipe Ar number Cm config Ar pipe-configuration
2575 .Cm queue Ar number Cm config Ar queue-configuration
2577 .Cm sched Ar number Cm config Ar sched-configuration
2580 The following parameters can be configured for a pipe:
2582 .Bl -tag -width indent -compact
2583 .It Cm bw Ar bandwidth | device
2584 Bandwidth, measured in
2587 .Brq Cm bit/s | Byte/s .
2590 A value of 0 (default) means unlimited bandwidth.
2591 The unit must immediately follow the number, as in
2593 .Dl "ipfw pipe 1 config bw 300Kbit/s"
2595 If a device name is specified instead of a numeric value, as in
2597 .Dl "ipfw pipe 1 config bw tun0"
2599 then the transmit clock is supplied by the specified device.
2600 At the moment only the
2602 device supports this
2603 functionality, for use in conjunction with
2606 .It Cm delay Ar ms-delay
2607 Propagation delay, measured in milliseconds.
2608 The value is rounded to the next multiple of the clock tick
2609 (typically 10ms, but it is a good practice to run kernels
2611 .Dq "options HZ=1000"
2613 the granularity to 1ms or less).
2614 The default value is 0, meaning no delay.
2616 .It Cm burst Ar size
2617 If the data to be sent exceeds the pipe's bandwidth limit
2618 (and the pipe was previously idle), up to
2620 bytes of data are allowed to bypass the
2622 scheduler, and will be sent as fast as the physical link allows.
2623 Any additional data will be transmitted at the rate specified
2627 The burst size depends on how long the pipe has been idle;
2628 the effective burst size is calculated as follows:
2635 .It Cm profile Ar filename
2636 A file specifying the additional overhead incurred in the transmission
2637 of a packet on the link.
2639 Some link types introduce extra delays in the transmission
2640 of a packet, e.g., because of MAC level framing, contention on
2641 the use of the channel, MAC level retransmissions and so on.
2642 From our point of view, the channel is effectively unavailable
2643 for this extra time, which is constant or variable depending
2645 Additionally, packets may be dropped after this
2646 time (e.g., on a wireless link after too many retransmissions).
2647 We can model the additional delay with an empirical curve
2648 that represents its distribution.
2649 .Bd -literal -offset indent
2650 cumulative probability
2660 +-------*------------------->
2663 The empirical curve may have both vertical and horizontal lines.
2664 Vertical lines represent constant delay for a range of
2666 Horizontal lines correspond to a discontinuity in the delay
2667 distribution: the pipe will use the largest delay for a
2670 The file format is the following, with whitespace acting as
2671 a separator and '#' indicating the beginning a comment:
2672 .Bl -tag -width indent
2673 .It Cm name Ar identifier
2674 optional name (listed by "ipfw pipe show")
2675 to identify the delay distribution;
2677 the bandwidth used for the pipe.
2678 If not specified here, it must be present
2679 explicitly as a configuration parameter for the pipe;
2680 .It Cm loss-level Ar L
2681 the probability above which packets are lost.
2682 (0.0 <= L <= 1.0, default 1.0 i.e., no loss);
2684 the number of samples used in the internal
2685 representation of the curve (2..1024; default 100);
2686 .It Cm "delay prob" | "prob delay"
2687 One of these two lines is mandatory and defines
2688 the format of the following lines with data points.
2690 2 or more lines representing points in the curve,
2691 with either delay or probability first, according
2692 to the chosen format.
2693 The unit for delay is milliseconds.
2694 Data points do not need to be sorted.
2695 Also, the number of actual lines can be different
2696 from the value of the "samples" parameter:
2698 utility will sort and interpolate
2699 the curve as needed.
2702 Example of a profile file:
2703 .Bd -literal -offset indent
2708 0 200 # minimum overhead is 200ms
2714 #configuration file end
2718 The following parameters can be configured for a queue:
2720 .Bl -tag -width indent -compact
2721 .It Cm pipe Ar pipe_nr
2722 Connects a queue to the specified pipe.
2723 Multiple queues (with the same or different weights) can be connected to
2724 the same pipe, which specifies the aggregate rate for the set of queues.
2726 .It Cm weight Ar weight
2727 Specifies the weight to be used for flows matching this queue.
2728 The weight must be in the range 1..100, and defaults to 1.
2731 The following case-insensitive parameters can be configured for a
2734 .Bl -tag -width indent -compact
2735 .It Cm type Ar {fifo | wf2q+ | rr | qfq | fq_codel | fq_pie}
2736 specifies the scheduling algorithm to use.
2737 .Bl -tag -width indent -compact
2739 is just a FIFO scheduler (which means that all packets
2740 are stored in the same queue as they arrive to the scheduler).
2741 FIFO has O(1) per-packet time complexity, with very low
2742 constants (estimate 60-80ns on a 2GHz desktop machine)
2743 but gives no service guarantees.
2745 implements the WF2Q+ algorithm, which is a Weighted Fair Queueing
2746 algorithm which permits flows to share bandwidth according to
2748 Note that weights are not priorities; even a flow
2749 with a minuscule weight will never starve.
2750 WF2Q+ has O(log N) per-packet processing cost, where N is the number
2751 of flows, and is the default algorithm used by previous versions
2754 implements the Deficit Round Robin algorithm, which has O(1) processing
2755 costs (roughly, 100-150ns per packet)
2756 and permits bandwidth allocation according to weights, but
2757 with poor service guarantees.
2759 implements the QFQ algorithm, which is a very fast variant of
2760 WF2Q+, with similar service guarantees and O(1) processing
2761 costs (roughly, 200-250ns per packet).
2763 implements the FQ-CoDel (FlowQueue-CoDel) scheduler/AQM algorithm, which
2764 uses a modified Deficit Round Robin scheduler to manage two lists of sub-queues
2765 (old sub-queues and new sub-queues) for providing brief periods of priority to
2766 lightweight or short burst flows.
2767 By default, the total number of sub-queues is 1024.
2768 FQ-CoDel's internal, dynamically
2769 created sub-queues are controlled by separate instances of CoDel AQM.
2771 implements the FQ-PIE (FlowQueue-PIE) scheduler/AQM algorithm, which similar to
2773 but uses per sub-queue PIE AQM instance to control the queue delay.
2777 inherits AQM parameters and options from
2781 inherits AQM parameters and options from
2784 Additionally, both of
2788 have shared scheduler parameters which are:
2789 .Bl -tag -width indent
2792 specifies the quantum (credit) of the scheduler.
2794 is the number of bytes a queue can serve before being moved to the tail
2796 The default is 1514 bytes, and the maximum accepable value
2800 specifies the hard size limit (in unit of packets) of all queues managed by an
2801 instance of the scheduler.
2802 The default value of
2804 is 10240 packets, and the maximum accepable value is 20480 packets.
2807 specifies the total number of flow queues (sub-queues) that fq_*
2808 creates and manages.
2809 By default, 1024 sub-queues are created when an instance
2810 of the fq_{codel/pie} scheduler is created.
2811 The maximum accepable value is
2815 Note that any token after
2819 is considered a parameter for fq_{codel/pie}.
2820 So, ensure all scheduler
2821 configuration options not related to fq_{codel/pie} are written before
2826 In addition to the type, all parameters allowed for a pipe can also
2827 be specified for a scheduler.
2829 Finally, the following parameters can be configured for both
2832 .Bl -tag -width XXXX -compact
2833 .It Cm buckets Ar hash-table-size
2834 Specifies the size of the hash table used for storing the
2836 Default value is 64 controlled by the
2839 .Va net.inet.ip.dummynet.hash_size ,
2840 allowed range is 16 to 65536.
2842 .It Cm mask Ar mask-specifier
2843 Packets sent to a given pipe or queue by an
2845 rule can be further classified into multiple flows, each of which is then
2849 A flow identifier is constructed by masking the IP addresses,
2850 ports and protocol types as specified with the
2852 options in the configuration of the pipe or queue.
2853 For each different flow identifier, a new pipe or queue is created
2854 with the same parameters as the original object, and matching packets
2859 are used, each flow will get the same bandwidth as defined by the pipe,
2862 are used, each flow will share the parent's pipe bandwidth evenly
2863 with other flows generated by the same queue (note that other queues
2864 with different weights might be connected to the same pipe).
2866 Available mask specifiers are a combination of one or more of the following:
2868 .Cm dst-ip Ar mask ,
2869 .Cm dst-ip6 Ar mask ,
2870 .Cm src-ip Ar mask ,
2871 .Cm src-ip6 Ar mask ,
2872 .Cm dst-port Ar mask ,
2873 .Cm src-port Ar mask ,
2874 .Cm flow-id Ar mask ,
2879 where the latter means all bits in all fields are significant.
2882 When a packet is dropped by a
2884 queue or pipe, the error
2885 is normally reported to the caller routine in the kernel, in the
2886 same way as it happens when a device queue fills up.
2888 option reports the packet as successfully delivered, which can be
2889 needed for some experimental setups where you want to simulate
2890 loss or congestion at a remote router.
2892 .It Cm plr Ar packet-loss-rate
2895 .Ar packet-loss-rate
2896 is a floating-point number between 0 and 1, with 0 meaning no
2897 loss, 1 meaning 100% loss.
2898 The loss rate is internally represented on 31 bits.
2900 .It Cm queue Brq Ar slots | size Ns Cm Kbytes
2905 Default value is 50 slots, which
2906 is the typical queue size for Ethernet devices.
2907 Note that for slow speed links you should keep the queue
2908 size short or your traffic might be affected by a significant
2910 E.g., 50 max-sized ethernet packets (1500 bytes) mean 600Kbit
2911 or 20s of queue on a 30Kbit/s pipe.
2912 Even worse effects can result if you get packets from an
2913 interface with a much larger MTU, e.g.\& the loopback interface
2914 with its 16KB packets.
2918 .Em net.inet.ip.dummynet.pipe_byte_limit
2920 .Em net.inet.ip.dummynet.pipe_slot_limit
2921 control the maximum lengths that can be specified.
2923 .It Cm red | gred Ar w_q Ns / Ns Ar min_th Ns / Ns Ar max_th Ns / Ns Ar max_p
2925 Make use of the RED (Random Early Detection) queue management algorithm.
2930 point numbers between 0 and 1 (inclusive), while
2934 are integer numbers specifying thresholds for queue management
2935 (thresholds are computed in bytes if the queue has been defined
2936 in bytes, in slots otherwise).
2937 The two parameters can also be of the same value if needed. The
2939 also supports the gentle RED variant (gred) and ECN (Explicit Congestion
2940 Notification) as optional. Three
2942 variables can be used to control the RED behaviour:
2943 .Bl -tag -width indent
2944 .It Va net.inet.ip.dummynet.red_lookup_depth
2945 specifies the accuracy in computing the average queue
2946 when the link is idle (defaults to 256, must be greater than zero)
2947 .It Va net.inet.ip.dummynet.red_avg_pkt_size
2948 specifies the expected average packet size (defaults to 512, must be
2950 .It Va net.inet.ip.dummynet.red_max_pkt_size
2951 specifies the expected maximum packet size, only used when queue
2952 thresholds are in bytes (defaults to 1500, must be greater than zero).
2955 .It Cm codel Oo Cm target Ar time Oc Oo Cm interval Ar time Oc Oo Cm ecn |
2957 Make use of the CoDel (Controlled-Delay) queue management algorithm.
2959 is interpreted as milliseconds by default but seconds (s), milliseconds (ms) or
2960 microseconds (us) can be specified instead.
2961 CoDel drops or marks (ECN) packets
2962 depending on packet sojourn time in the queue.
2965 (5ms by default) is the minimum acceptable persistent queue delay that CoDel
2967 CoDel does not drop packets directly after packets sojourn time becomes
2974 (100ms default) before dropping.
2977 should be set to maximum RTT for all expected connections.
2979 enables (disabled by default) packet marking (instead of dropping) for
2980 ECN-enabled TCP flows when queue delay becomes high.
2982 Note that any token after
2984 is considered a parameter for CoDel.
2985 So, ensure all pipe/queue
2986 configuration options are written before
2993 .Va net.inet.ip.dummynet.codel.target
2995 .Va net.inet.ip.dummynet.codel.interval
2996 can be used to set CoDel default parameters.
2998 .It Cm pie Oo Cm target Ar time Oc Oo Cm tupdate Ar time Oc Oo
2999 .Cm alpha Ar n Oc Oo Cm beta Ar n Oc Oo Cm max_burst Ar time Oc Oo
3000 .Cm max_ecnth Ar n Oc Oo Cm ecn | Cm noecn Oc Oo Cm capdrop |
3001 .Cm nocapdrop Oc Oo Cm drand | Cm nodrand Oc Oo Cm onoff
3002 .Oc Oo Cm dre | Cm ts Oc
3003 Make use of the PIE (Proportional Integral controller Enhanced) queue management
3005 PIE drops or marks packets depending on a calculated drop probability during
3006 en-queue process, with the aim of achieving high throughput while keeping queue
3008 At regular time intervals of
3011 (15ms by default) a background process (re)calculates the probability based on queue delay
3015 (15ms by default) and queue delay trends.
3016 PIE approximates current queue
3017 delay by using a departure rate estimation method, or (optionally) by using a
3018 packet timestamp method similar to CoDel.
3020 is interpreted as milliseconds by default but seconds (s), milliseconds (ms) or
3021 microseconds (us) can be specified instead.
3022 The other PIE parameters and options are as follows:
3023 .Bl -tag -width indent
3026 is a floating point number between 0 and 7 which specifies the weight of queue
3027 delay deviations that is used in drop probability calculation.
3028 0.125 is the default.
3031 is a floating point number between 0 and 7 which specifies is the weight of queue
3032 delay trend that is used in drop probability calculation.
3033 1.25 is the default.
3034 .It Cm max_burst Ar time
3035 The maximum period of time that PIE does not drop/mark packets.
3037 default and 10s is the maximum value.
3038 .It Cm max_ecnth Ar n
3039 Even when ECN is enabled, PIE drops packets instead of marking them when drop
3040 probability becomes higher than ECN probability threshold
3042 , the default is 0.1 (i.e 10%) and 1 is the maximum value.
3044 enable or disable ECN marking for ECN-enabled TCP flows.
3045 Disabled by default.
3046 .It Cm capdrop | nocapdrop
3047 enable or disable cap drop adjustment.
3048 Cap drop adjustment is enabled by default.
3049 .It Cm drand | nodrand
3050 enable or disable drop probability de-randomisation.
3051 De-randomisation eliminates
3052 the problem of dropping packets too close or too far.
3053 De-randomisation is enabled by default.
3055 enable turning PIE on and off depending on queue load.
3056 If this option is enabled,
3057 PIE turnes on when over 1/3 of queue becomes full.
3058 This option is disabled by
3061 Calculate queue delay using departure rate estimation
3069 Note that any token after
3071 is considered a parameter for PIE.
3072 So ensure all pipe/queue
3073 the configuration options are written before
3077 variables can be used to control the
3081 .Sx SYSCTL VARIABLES
3082 section for more details.
3085 When used with IPv6 data,
3087 currently has several limitations.
3088 Information necessary to route link-local packets to an
3089 interface is not available after processing by
3091 so those packets are dropped in the output path.
3092 Care should be taken to ensure that link-local packets are not passed to
3095 Here are some important points to consider when designing your
3099 Remember that you filter both packets going
3103 Most connections need packets going in both directions.
3105 Remember to test very carefully.
3106 It is a good idea to be near the console when doing this.
3107 If you cannot be near the console,
3108 use an auto-recovery script such as the one in
3109 .Pa /usr/share/examples/ipfw/change_rules.sh .
3111 Do not forget the loopback interface.
3116 There are circumstances where fragmented datagrams are unconditionally
3118 TCP packets are dropped if they do not contain at least 20 bytes of
3119 TCP header, UDP packets are dropped if they do not contain a full 8
3120 byte UDP header, and ICMP packets are dropped if they do not contain
3121 4 bytes of ICMP header, enough to specify the ICMP type, code, and
3123 These packets are simply logged as
3125 since there may not be enough good data in the packet to produce a
3126 meaningful log entry.
3128 Another type of packet is unconditionally dropped, a TCP packet with a
3129 fragment offset of one.
3130 This is a valid packet, but it only has one use, to try
3131 to circumvent firewalls.
3132 When logging is enabled, these packets are
3133 reported as being dropped by rule -1.
3135 If you are logged in over a network, loading the
3139 is probably not as straightforward as you would think.
3140 The following command line is recommended:
3141 .Bd -literal -offset indent
3143 ipfw add 32000 allow ip from any to any
3146 Along the same lines, doing an
3147 .Bd -literal -offset indent
3151 in similar surroundings is also a bad idea.
3155 filter list may not be modified if the system security level
3156 is set to 3 or higher
3159 for information on system security levels).
3161 .Sh PACKET DIVERSION
3164 socket bound to the specified port will receive all packets
3165 diverted to that port.
3166 If no socket is bound to the destination port, or if the divert module is
3167 not loaded, or if the kernel was not compiled with divert socket support,
3168 the packets are dropped.
3169 .Sh NETWORK ADDRESS TRANSLATION (NAT)
3171 support in-kernel NAT using the kernel version of
3174 The nat configuration command is the following:
3175 .Bd -ragged -offset indent
3180 .Ar nat-configuration
3184 The following parameters can be configured:
3185 .Bl -tag -width indent
3186 .It Cm ip Ar ip_address
3187 Define an ip address to use for aliasing.
3189 Use ip address of NIC for aliasing, dynamically changing
3190 it if NIC's ip address changes.
3192 Enable logging on this nat instance.
3194 Deny any incoming connection from outside world.
3196 Try to leave the alias port numbers unchanged from
3197 the actual local port numbers.
3199 Traffic on the local network not originating from an
3200 unregistered address spaces will be ignored.
3202 Reset table of the packet aliasing engine on address change.
3204 Reverse the way libalias handles aliasing.
3206 Obey transparent proxy rules only, packet aliasing is not performed.
3208 Skip instance in case of global state lookup (see below).
3211 Some specials value can be supplied instead of
3213 .Bl -tag -width indent
3215 Looks up translation state in all configured nat instances.
3216 If an entry is found, packet is aliased according to that entry.
3217 If no entry was found in any of the instances, packet is passed unchanged,
3218 and no new entry will be created.
3220 .Sx MULTIPLE INSTANCES
3223 for more information.
3225 Uses argument supplied in lookup table.
3228 section below for more information on lookup tables.
3231 To let the packet continue after being (de)aliased, set the sysctl variable
3232 .Va net.inet.ip.fw.one_pass
3234 For more information about aliasing modes, refer to
3238 for some examples about nat usage.
3239 .Ss REDIRECT AND LSNAT SUPPORT IN IPFW
3240 Redirect and LSNAT support follow closely the syntax used in
3244 for some examples on how to do redirect and lsnat.
3245 .Ss SCTP NAT SUPPORT
3246 SCTP nat can be configured in a similar manner to TCP through the
3249 The main difference is that
3251 does not do port translation.
3252 Since the local and global side ports will be the same,
3253 there is no need to specify both.
3254 Ports are redirected as follows:
3255 .Bd -ragged -offset indent
3261 .Cm redirect_port sctp
3262 .Ar ip_address [,addr_list] {[port | port-port] [,ports]}
3268 configuration can be done in real-time through the
3271 All may be changed dynamically, though the hash_table size will only
3276 .Sx SYSCTL VARIABLES
3278 .Sh IPv6/IPv4 NETWORK ADDRESS AND PROTOCOL TRANSLATION
3280 supports in-kernel IPv6/IPv4 network address and protocol translation.
3281 Stateful NAT64 translation allows IPv6-only clients to contact IPv4 servers
3282 using unicast TCP, UDP or ICMP protocols.
3283 One or more IPv4 addresses assigned to a stateful NAT64 translator are shared
3284 among serveral IPv6-only clients.
3285 When stateful NAT64 is used in conjunction with DNS64, no changes are usually
3286 required in the IPv6 client or the IPv4 server.
3289 should be loaded or kernel should have
3290 .Cm options IPFIREWALL_NAT64
3291 to be able use stateful NAT64 translator.
3293 Stateful NAT64 uses a bunch of memory for several types of objects.
3294 When IPv6 client initiates connection, NAT64 translator creates a host entry
3295 in the states table.
3296 Each host entry has a number of ports group entries allocated on demand.
3297 Ports group entries contains connection state entries.
3298 There are several options to control limits and lifetime for these objects.
3300 NAT64 translator follows RFC7915 when does ICMPv6/ICMP translation,
3301 unsupported message types will be silently dropped.
3302 IPv6 needs several ICMPv6 message types to be explicitly allowed for correct
3304 Make sure that ND6 neighbor solicitation (ICMPv6 type 135) and neighbor
3305 advertisement (ICMPv6 type 136) messages will not be handled by translation
3308 After translation NAT64 translator sends packets through corresponding netisr
3310 Thus translator host should be configured as IPv4 and IPv6 router.
3312 The stateful NAT64 configuration command is the following:
3313 .Bd -ragged -offset indent
3322 The following parameters can be configured:
3323 .Bl -tag -width indent
3324 .It Cm prefix4 Ar ipv4_prefix/plen
3325 The IPv4 prefix with mask defines the pool of IPv4 addresses used as
3326 source address after translation.
3327 Stateful NAT64 module translates IPv6 source address of client to one
3328 IPv4 address from this pool.
3329 Note that incoming IPv4 packets that don't have corresponding state entry
3330 in the states table will be dropped by translator.
3331 Make sure that translation rules handle packets, destined to configured prefix.
3332 .It Cm prefix6 Ar ipv6_prefix/length
3333 The IPv6 prefix defines IPv4-embedded IPv6 addresses used by translator
3334 to represent IPv4 addresses. This IPv6 prefix should be configured in DNS64.
3335 The translator implementation follows RFC6052, that restricts the length of
3336 prefixes to one of following: 32, 40, 48, 56, 64, or 96.
3337 The Well-Known IPv6 Prefix 64:ff9b:: must be 96 bits long.
3338 .It Cm max_ports Ar number
3339 Maximum number of ports reserved for upper level protocols to one IPv6 client.
3340 All reserved ports are divided into chunks between supported protocols.
3341 The number of connections from one IPv6 client is limited by this option.
3342 Note that closed TCP connections still remain in the list of connections until
3344 interval will not expire.
3347 .It Cm host_del_age Ar seconds
3348 The number of seconds until the host entry for a IPv6 client will be deleted
3349 and all its resources will be released due to inactivity.
3352 .It Cm pg_del_age Ar seconds
3353 The number of seconds until a ports group with unused state entries will
3357 .It Cm tcp_syn_age Ar seconds
3358 The number of seconds while a state entry for TCP connection with only SYN
3360 If TCP connection establishing will not be finished,
3361 state entry will be deleted.
3364 .It Cm tcp_est_age Ar seconds
3365 The number of seconds while a state entry for established TCP connection
3369 .It Cm tcp_close_age Ar seconds
3370 The number of seconds while a state entry for closed TCP connection
3372 Keeping state entries for closed connections is needed, because IPv4 servers
3373 typically keep closed connections in a TIME_WAIT state for a several minutes.
3374 Since translator's IPv4 addresses are shared among all IPv6 clients,
3375 new connections from the same addresses and ports may be rejected by server,
3376 because these connections are still in a TIME_WAIT state.
3377 Keeping them in translator's state table protects from such rejects.
3380 .It Cm udp_age Ar seconds
3381 The number of seconds while translator keeps state entry in a waiting for
3382 reply to the sent UDP datagram.
3385 .It Cm icmp_age Ar seconds
3386 The number of seconds while translator keeps state entry in a waiting for
3387 reply to the sent ICMP message.
3391 Turn on logging of all handled packets via BPF through
3395 is a pseudo interface and can be created after a boot manually with
3398 Note that it has different purpose than
3401 Translators sends to BPF an additional information with each packet.
3404 you are able to see each handled packet before and after translation.
3406 Turn off logging of all handled packets via BPF.
3409 To inspect a states table of stateful NAT64 the following command can be used:
3410 .Bd -ragged -offset indent
3419 Stateless NAT64 translator doesn't use a states table for translation
3420 and converts IPv4 addresses to IPv6 and vice versa solely based on the
3421 mappings taken from configured lookup tables.
3422 Since a states table doesn't used by stateless translator,
3423 it can be configured to pass IPv4 clients to IPv6-only servers.
3425 The stateless NAT64 configuration command is the following:
3426 .Bd -ragged -offset indent
3435 The following parameters can be configured:
3436 .Bl -tag -width indent
3437 .It Cm prefix6 Ar ipv6_prefix/length
3438 The IPv6 prefix defines IPv4-embedded IPv6 addresses used by translator
3439 to represent IPv4 addresses. This IPv6 prefix should be configured in DNS64.
3440 .It Cm table4 Ar table46
3443 contains mapping how IPv4 addresses should be translated to IPv6 addresses.
3444 .It Cm table6 Ar table64
3447 contains mapping how IPv6 addresses should be translated to IPv4 addresses.
3449 Turn on logging of all handled packets via BPF through
3453 Turn off logging of all handled packets via BPF.
3456 Note that the behavior of stateless translator with respect to not matched
3457 packets differs from stateful translator.
3458 If corresponding addresses was not found in the lookup tables, the packet
3459 will not be dropped and the search continues.
3460 .Sh IPv6-to-IPv6 NETWORK PREFIX TRANSLATION (NPTv6)
3462 supports in-kernel IPv6-to-IPv6 network prefix translation as described
3466 should be loaded or kernel should has
3467 .Cm options IPFIREWALL_NPTV6
3468 to be able use NPTv6 translator.
3470 The NPTv6 configuration command is the following:
3471 .Bd -ragged -offset indent
3480 The following parameters can be configured:
3481 .Bl -tag -width indent
3482 .It Cm int_prefix Ar ipv6_prefix
3483 IPv6 prefix used in internal network.
3484 NPTv6 module translates source address when it matches this prefix.
3485 .It Cm ext_prefix Ar ipv6_prefix
3486 IPv6 prefix used in external network.
3487 NPTv6 module translates destination address when it matches this prefix.
3488 .It Cm prefixlen Ar length
3489 The length of specified IPv6 prefixes. It must be in range from 8 to 64.
3492 Note that the prefix translation rules are silently ignored when IPv6 packet
3493 forwarding is disabled.
3494 To enable the packet forwarding, set the sysctl variable
3495 .Va net.inet6.ip6.forwarding
3498 To let the packet continue after being translated, set the sysctl variable
3499 .Va net.inet.ip.fw.one_pass
3502 Tunables can be set in
3508 before ipfw module gets loaded.
3509 .Bl -tag -width indent
3510 .It Va net.inet.ip.fw.default_to_accept: No 0
3511 Defines ipfw last rule behavior.
3512 This value overrides
3513 .Cd "options IPFW_DEFAULT_TO_(ACCEPT|DENY)"
3514 from kernel configuration file.
3515 .It Va net.inet.ip.fw.tables_max: No 128
3516 Defines number of tables available in ipfw.
3517 Number cannot exceed 65534.
3519 .Sh SYSCTL VARIABLES
3522 variables controls the behaviour of the firewall and
3524 .Pq Nm dummynet , bridge , sctp nat .
3525 These are shown below together with their default value
3526 (but always check with the
3528 command what value is actually in use) and meaning:
3529 .Bl -tag -width indent
3530 .It Va net.inet.ip.alias.sctp.accept_global_ootb_addip: No 0
3533 responds to receipt of global OOTB ASCONF-AddIP:
3534 .Bl -tag -width indent
3536 No response (unless a partially matching association exists -
3537 ports and vtags match but global address does not)
3540 will accept and process all OOTB global AddIP messages.
3543 Option 1 should never be selected as this forms a security risk.
3545 establish multiple fake associations by sending AddIP messages.
3546 .It Va net.inet.ip.alias.sctp.chunk_proc_limit: No 5
3547 Defines the maximum number of chunks in an SCTP packet that will be
3549 packet that matches an existing association.
3550 This value is enforced to be greater or equal than
3551 .Cm net.inet.ip.alias.sctp.initialising_chunk_proc_limit .
3553 a DoS risk yet setting too low a value may result in
3554 important control chunks in
3555 the packet not being located and parsed.
3556 .It Va net.inet.ip.alias.sctp.error_on_ootb: No 1
3559 responds to any Out-of-the-Blue (OOTB) packets with ErrorM packets.
3560 An OOTB packet is a packet that arrives with no existing association
3563 and is not an INIT or ASCONF-AddIP packet:
3564 .Bl -tag -width indent
3566 ErrorM is never sent in response to OOTB packets.
3568 ErrorM is only sent to OOTB packets received on the local side.
3570 ErrorM is sent to the local side and on the global side ONLY if there is a
3571 partial match (ports and vtags match but the source global IP does not).
3572 This value is only useful if the
3574 is tracking global IP addresses.
3576 ErrorM is sent in response to all OOTB packets on both
3577 the local and global side
3581 At the moment the default is 0, since the ErrorM packet is not yet
3582 supported by most SCTP stacks.
3583 When it is supported, and if not tracking
3584 global addresses, we recommend setting this value to 1 to allow
3585 multi-homed local hosts to function with the
3587 To track global addresses, we recommend setting this value to 2 to
3588 allow global hosts to be informed when they need to (re)send an
3590 Value 3 should never be chosen (except for debugging) as the
3592 will respond to all OOTB global packets (a DoS risk).
3593 .It Va net.inet.ip.alias.sctp.hashtable_size: No 2003
3594 Size of hash tables used for
3596 lookups (100 < prime_number > 1000001).
3599 size for any future created
3601 instance and therefore must be set prior to creating a
3604 The table sizes may be changed to suit specific needs.
3605 If there will be few
3606 concurrent associations, and memory is scarce, you may make these smaller.
3607 If there will be many thousands (or millions) of concurrent associations, you
3608 should make these larger.
3609 A prime number is best for the table size.
3611 update function will adjust your input value to the next highest prime number.
3612 .It Va net.inet.ip.alias.sctp.holddown_time: No 0
3613 Hold association in table for this many seconds after receiving a
3615 This allows endpoints to correct shutdown gracefully if a
3616 shutdown_complete is lost and retransmissions are required.
3617 .It Va net.inet.ip.alias.sctp.init_timer: No 15
3618 Timeout value while waiting for (INIT-ACK|AddIP-ACK).
3619 This value cannot be 0.
3620 .It Va net.inet.ip.alias.sctp.initialising_chunk_proc_limit: No 2
3621 Defines the maximum number of chunks in an SCTP packet that will be parsed when
3622 no existing association exists that matches that packet.
3624 will only be an INIT or ASCONF-AddIP packet.
3625 A higher value may become a DoS
3626 risk as malformed packets can consume processing resources.
3627 .It Va net.inet.ip.alias.sctp.param_proc_limit: No 25
3628 Defines the maximum number of parameters within a chunk that will be
3631 As for other similar sysctl variables, larger values pose a DoS risk.
3632 .It Va net.inet.ip.alias.sctp.log_level: No 0
3633 Level of detail in the system log messages (0 \- minimal, 1 \- event,
3634 2 \- info, 3 \- detail, 4 \- debug, 5 \- max debug).
3636 option in high loss environments.
3637 .It Va net.inet.ip.alias.sctp.shutdown_time: No 15
3638 Timeout value while waiting for SHUTDOWN-COMPLETE.
3639 This value cannot be 0.
3640 .It Va net.inet.ip.alias.sctp.track_global_addresses: No 0
3641 Enables/disables global IP address tracking within the
3644 upper limit on the number of addresses tracked for each association:
3645 .Bl -tag -width indent
3647 Global tracking is disabled
3649 Enables tracking, the maximum number of addresses tracked for each
3650 association is limited to this value
3653 This variable is fully dynamic, the new value will be adopted for all newly
3654 arriving associations, existing associations are treated
3655 as they were previously.
3656 Global tracking will decrease the number of collisions within the
3659 of increased processing load, memory usage, complexity, and possible
3662 problems in complex networks with multiple
3664 We recommend not tracking
3665 global IP addresses, this will still result in a fully functional
3667 .It Va net.inet.ip.alias.sctp.up_timer: No 300
3668 Timeout value to keep an association up with no traffic.
3669 This value cannot be 0.
3670 .It Va net.inet.ip.dummynet.codel.interval : No 100000
3673 AQM interval in microseconds.
3674 The value must be in the range 1..5000000.
3675 .It Va net.inet.ip.dummynet.codel.target : No 5000
3678 AQM target delay time in microseconds (the minimum acceptable persistent queue
3680 The value must be in the range 1..5000000.
3681 .It Va net.inet.ip.dummynet.expire : No 1
3682 Lazily delete dynamic pipes/queue once they have no pending traffic.
3683 You can disable this by setting the variable to 0, in which case
3684 the pipes/queues will only be deleted when the threshold is reached.
3685 .It Va net.inet.ip.dummynet.fqcodel.flows : No 1024
3686 Defines the default total number of flow queues (sub-queues) that
3688 creates and manages.
3689 The value must be in the range 1..65536.
3690 .It Va net.inet.ip.dummynet.fqcodel.interval : No 100000
3693 scheduler/AQM interval in microseconds.
3694 The value must be in the range 1..5000000.
3695 .It Va net.inet.ip.dummynet.fqcodel.limit : No 10240
3696 The default hard size limit (in unit of packet) of all queues managed by an
3700 The value must be in the range 1..20480.
3701 .It Va net.inet.ip.dummynet.fqcodel.quantum : No 1514
3702 The default quantum (credit) of the
3705 The value must be in the range 1..9000.
3706 .It Va net.inet.ip.dummynet.fqcodel.target : No 5000
3709 scheduler/AQM target delay time in microseconds (the minimum acceptable
3710 persistent queue delay).
3711 The value must be in the range 1..5000000.
3712 .It Va net.inet.ip.dummynet.fqpie.alpha : No 125
3715 parameter (scaled by 1000) for
3718 The value must be in the range 1..7000.
3719 .It Va net.inet.ip.dummynet.fqpie.beta : No 1250
3722 parameter (scaled by 1000) for
3725 The value must be in the range 1..7000.
3726 .It Va net.inet.ip.dummynet.fqpie.flows : No 1024
3727 Defines the default total number of flow queues (sub-queues) that
3729 creates and manages.
3730 The value must be in the range 1..65536.
3731 .It Va net.inet.ip.dummynet.fqpie.limit : No 10240
3732 The default hard size limit (in unit of packet) of all queues managed by an
3736 The value must be in the range 1..20480.
3737 .It Va net.inet.ip.dummynet.fqpie.max_burst : No 150000
3738 The default maximum period of microseconds that
3740 scheduler/AQM does not drop/mark packets.
3741 The value must be in the range 1..10000000.
3742 .It Va net.inet.ip.dummynet.fqpie.max_ecnth : No 99
3743 The default maximum ECN probability threshold (scaled by 1000) for
3746 The value must be in the range 1..7000.
3747 .It Va net.inet.ip.dummynet.fqpie.quantum : No 1514
3748 The default quantum (credit) of the
3751 The value must be in the range 1..9000.
3752 .It Va net.inet.ip.dummynet.fqpie.target : No 15000
3757 in unit of microsecond.
3758 The value must be in the range 1..5000000.
3759 .It Va net.inet.ip.dummynet.fqpie.tupdate : No 15000
3764 in unit of microsecond.
3765 The value must be in the range 1..5000000.
3766 .It Va net.inet.ip.dummynet.hash_size : No 64
3767 Default size of the hash table used for dynamic pipes/queues.
3768 This value is used when no
3770 option is specified when configuring a pipe/queue.
3771 .It Va net.inet.ip.dummynet.io_fast : No 0
3772 If set to a non-zero value,
3777 operation (see above) is enabled.
3778 .It Va net.inet.ip.dummynet.io_pkt
3779 Number of packets passed to
3781 .It Va net.inet.ip.dummynet.io_pkt_drop
3782 Number of packets dropped by
3784 .It Va net.inet.ip.dummynet.io_pkt_fast
3785 Number of packets bypassed by the
3788 .It Va net.inet.ip.dummynet.max_chain_len : No 16
3789 Target value for the maximum number of pipes/queues in a hash bucket.
3791 .Cm max_chain_len*hash_size
3792 is used to determine the threshold over which empty pipes/queues
3793 will be expired even when
3794 .Cm net.inet.ip.dummynet.expire=0 .
3795 .It Va net.inet.ip.dummynet.red_lookup_depth : No 256
3796 .It Va net.inet.ip.dummynet.red_avg_pkt_size : No 512
3797 .It Va net.inet.ip.dummynet.red_max_pkt_size : No 1500
3798 Parameters used in the computations of the drop probability
3799 for the RED algorithm.
3800 .It Va net.inet.ip.dummynet.pie.alpha : No 125
3803 parameter (scaled by 1000) for
3806 The value must be in the range 1..7000.
3807 .It Va net.inet.ip.dummynet.pie.beta : No 1250
3810 parameter (scaled by 1000) for
3813 The value must be in the range 1..7000.
3814 .It Va net.inet.ip.dummynet.pie.max_burst : No 150000
3815 The default maximum period of microseconds that
3817 AQM does not drop/mark packets.
3818 The value must be in the range 1..10000000.
3819 .It Va net.inet.ip.dummynet.pie.max_ecnth : No 99
3820 The default maximum ECN probability threshold (scaled by 1000) for
3823 The value must be in the range 1..7000.
3824 .It Va net.inet.ip.dummynet.pie.target : No 15000
3829 AQM in unit of microsecond.
3830 The value must be in the range 1..5000000.
3831 .It Va net.inet.ip.dummynet.pie.tupdate : No 15000
3836 AQM in unit of microsecond.
3837 The value must be in the range 1..5000000.
3838 .It Va net.inet.ip.dummynet.pipe_byte_limit : No 1048576
3839 .It Va net.inet.ip.dummynet.pipe_slot_limit : No 100
3840 The maximum queue size that can be specified in bytes or packets.
3841 These limits prevent accidental exhaustion of resources such as mbufs.
3842 If you raise these limits,
3843 you should make sure the system is configured so that sufficient resources
3845 .It Va net.inet.ip.fw.autoinc_step : No 100
3846 Delta between rule numbers when auto-generating them.
3847 The value must be in the range 1..1000.
3848 .It Va net.inet.ip.fw.curr_dyn_buckets : Va net.inet.ip.fw.dyn_buckets
3849 The current number of buckets in the hash table for dynamic rules
3851 .It Va net.inet.ip.fw.debug : No 1
3852 Controls debugging messages produced by
3854 .It Va net.inet.ip.fw.default_rule : No 65535
3855 The default rule number (read-only).
3857 .Nm , the default rule is the last one, so its number
3858 can also serve as the highest number allowed for a rule.
3859 .It Va net.inet.ip.fw.dyn_buckets : No 256
3860 The number of buckets in the hash table for dynamic rules.
3861 Must be a power of 2, up to 65536.
3862 It only takes effect when all dynamic rules have expired, so you
3863 are advised to use a
3865 command to make sure that the hash table is resized.
3866 .It Va net.inet.ip.fw.dyn_count : No 3
3867 Current number of dynamic rules
3869 .It Va net.inet.ip.fw.dyn_keepalive : No 1
3870 Enables generation of keepalive packets for
3872 rules on TCP sessions.
3873 A keepalive is generated to both
3874 sides of the connection every 5 seconds for the last 20
3875 seconds of the lifetime of the rule.
3876 .It Va net.inet.ip.fw.dyn_max : No 8192
3877 Maximum number of dynamic rules.
3878 When you hit this limit, no more dynamic rules can be
3879 installed until old ones expire.
3880 .It Va net.inet.ip.fw.dyn_ack_lifetime : No 300
3881 .It Va net.inet.ip.fw.dyn_syn_lifetime : No 20
3882 .It Va net.inet.ip.fw.dyn_fin_lifetime : No 1
3883 .It Va net.inet.ip.fw.dyn_rst_lifetime : No 1
3884 .It Va net.inet.ip.fw.dyn_udp_lifetime : No 5
3885 .It Va net.inet.ip.fw.dyn_short_lifetime : No 30
3886 These variables control the lifetime, in seconds, of dynamic
3888 Upon the initial SYN exchange the lifetime is kept short,
3889 then increased after both SYN have been seen, then decreased
3890 again during the final FIN exchange or when a RST is received.
3892 .Em dyn_fin_lifetime
3894 .Em dyn_rst_lifetime
3895 must be strictly lower than 5 seconds, the period of
3896 repetition of keepalives.
3897 The firewall enforces that.
3898 .It Va net.inet.ip.fw.dyn_keep_states: No 0
3899 Keep dynamic states on rule/set deletion.
3900 States are relinked to default rule (65535).
3901 This can be handly for ruleset reload.
3902 Turned off by default.
3903 .It Va net.inet.ip.fw.enable : No 1
3904 Enables the firewall.
3905 Setting this variable to 0 lets you run your machine without
3906 firewall even if compiled in.
3907 .It Va net.inet6.ip6.fw.enable : No 1
3908 provides the same functionality as above for the IPv6 case.
3909 .It Va net.inet.ip.fw.one_pass : No 1
3910 When set, the packet exiting from the
3914 node is not passed though the firewall again.
3915 Otherwise, after an action, the packet is
3916 reinjected into the firewall at the next rule.
3917 .It Va net.inet.ip.fw.tables_max : No 128
3918 Maximum number of tables.
3919 .It Va net.inet.ip.fw.verbose : No 1
3920 Enables verbose messages.
3921 .It Va net.inet.ip.fw.verbose_limit : No 0
3922 Limits the number of messages produced by a verbose firewall.
3923 .It Va net.inet6.ip6.fw.deny_unknown_exthdrs : No 1
3924 If enabled packets with unknown IPv6 Extension Headers will be denied.
3925 .It Va net.link.ether.ipfw : No 0
3926 Controls whether layer-2 packets are passed to
3929 .It Va net.link.bridge.ipfw : No 0
3930 Controls whether bridged packets are passed to
3934 .Sh INTERNAL DIAGNOSTICS
3935 There are some commands that may be useful to understand current state
3936 of certain subsystems inside kernel module.
3937 These commands provide debugging output which may change without notice.
3939 Currently the following commands are available as
3942 .Bl -tag -width indent
3944 Lists all interface which are currently tracked by
3946 with their in-kernel status.
3948 List all table lookup algorithms currently available.
3951 There are far too many possible uses of
3953 so this Section will only give a small set of examples.
3955 .Ss BASIC PACKET FILTERING
3956 This command adds an entry which denies all tcp packets from
3957 .Em cracker.evil.org
3958 to the telnet port of
3960 from being forwarded by the host:
3962 .Dl "ipfw add deny tcp from cracker.evil.org to wolf.tambov.su telnet"
3964 This one disallows any connection from the entire cracker's
3967 .Dl "ipfw add deny ip from 123.45.67.0/24 to my.host.org"
3969 A first and efficient way to limit access (not using dynamic rules)
3970 is the use of the following rules:
3972 .Dl "ipfw add allow tcp from any to any established"
3973 .Dl "ipfw add allow tcp from net1 portlist1 to net2 portlist2 setup"
3974 .Dl "ipfw add allow tcp from net3 portlist3 to net3 portlist3 setup"
3976 .Dl "ipfw add deny tcp from any to any"
3978 The first rule will be a quick match for normal TCP packets,
3979 but it will not match the initial SYN packet, which will be
3982 rules only for selected source/destination pairs.
3983 All other SYN packets will be rejected by the final
3987 If you administer one or more subnets, you can take advantage
3988 of the address sets and or-blocks and write extremely
3989 compact rulesets which selectively enable services to blocks
3990 of clients, as below:
3992 .Dl "goodguys=\*q{ 10.1.2.0/24{20,35,66,18} or 10.2.3.0/28{6,3,11} }\*q"
3993 .Dl "badguys=\*q10.1.2.0/24{8,38,60}\*q"
3995 .Dl "ipfw add allow ip from ${goodguys} to any"
3996 .Dl "ipfw add deny ip from ${badguys} to any"
3997 .Dl "... normal policies ..."
4001 option could be used to do automated anti-spoofing by adding the
4002 following to the top of a ruleset:
4004 .Dl "ipfw add deny ip from any to any not verrevpath in"
4006 This rule drops all incoming packets that appear to be coming to the
4007 system on the wrong interface.
4008 For example, a packet with a source
4009 address belonging to a host on a protected internal network would be
4010 dropped if it tried to enter the system from an external interface.
4014 option could be used to do similar but more restricted anti-spoofing
4015 by adding the following to the top of a ruleset:
4017 .Dl "ipfw add deny ip from any to any not antispoof in"
4019 This rule drops all incoming packets that appear to be coming from another
4020 directly connected system but on the wrong interface.
4021 For example, a packet with a source address of
4022 .Li 192.168.0.0/24 ,
4031 option could be used to (re)mark user traffic,
4032 by adding the following to the appropriate place in ruleset:
4034 .Dl "ipfw add setdscp be ip from any to any dscp af11,af21"
4035 .Ss SELECTIVE MIRRORING
4036 If your network has network traffic analyzer
4037 connected to your host directly via dedicated interface
4038 or remotely via RSPAN vlan, you can selectively mirror
4039 some ethernet layer2 frames to the analyzer.
4041 First, make sure your firewall is already configured and runs.
4042 Then, enable layer2 processing if not already enabled:
4044 .Dl "sysctl net.link.ether.ipfw=1"
4046 Next, load needed additional kernel modules:
4048 .Dl "kldload ng_ether ng_ipfw"
4050 Optionally, make system load these modules automatically
4053 .Dl sysrc kld_list+="ng_ether ng_ipfw"
4057 kernel module to transmit mirrored copies of layer2 frames
4058 out via vlan900 interface:
4060 .Dl "ngctl connect ipfw: vlan900: 1 lower"
4062 Think of "1" here as of "mirroring instance index" and vlan900 is its
4064 You can have arbitrary number of instances.
4069 At last, actually start mirroring of selected frames using "instance 1".
4070 For frames incoming from em0 interface:
4072 .Dl "ipfw add ngtee 1 ip from any to 192.168.0.1 layer2 in recv em0"
4074 For frames outgoing to em0 interface:
4076 .Dl "ipfw add ngtee 1 ip from any to 192.168.0.1 layer2 out xmit em0"
4078 For both incoming and outgoing frames while flowing through em0:
4080 .Dl "ipfw add ngtee 1 ip from any to 192.168.0.1 layer2 via em0"
4082 Make sure you do not perform mirroring for already duplicated frames
4083 or kernel may hang as there is no safety net.
4085 In order to protect a site from flood attacks involving fake
4086 TCP packets, it is safer to use dynamic rules:
4088 .Dl "ipfw add check-state"
4089 .Dl "ipfw add deny tcp from any to any established"
4090 .Dl "ipfw add allow tcp from my-net to any setup keep-state"
4092 This will let the firewall install dynamic rules only for
4093 those connection which start with a regular SYN packet coming
4094 from the inside of our network.
4095 Dynamic rules are checked when encountering the first
4104 rule should usually be placed near the beginning of the
4105 ruleset to minimize the amount of work scanning the ruleset.
4106 Your mileage may vary.
4108 For more complex scenarios with dynamic rules
4112 can be used to precisely control creation and checking of dynamic rules.
4113 Example of usage of these options are provided in
4114 .Sx NETWORK ADDRESS TRANSLATION (NAT)
4117 To limit the number of connections a user can open
4118 you can use the following type of rules:
4120 .Dl "ipfw add allow tcp from my-net/24 to any setup limit src-addr 10"
4121 .Dl "ipfw add allow tcp from any to me setup limit src-addr 4"
4123 The former (assuming it runs on a gateway) will allow each host
4124 on a /24 network to open at most 10 TCP connections.
4125 The latter can be placed on a server to make sure that a single
4126 client does not use more than 4 simultaneous connections.
4129 stateful rules can be subject to denial-of-service attacks
4130 by a SYN-flood which opens a huge number of dynamic rules.
4131 The effects of such attacks can be partially limited by
4134 variables which control the operation of the firewall.
4136 Here is a good usage of the
4138 command to see accounting records and timestamp information:
4142 or in short form without timestamps:
4146 which is equivalent to:
4150 Next rule diverts all incoming packets from 192.168.2.0/24
4151 to divert port 5000:
4153 .Dl ipfw divert 5000 ip from 192.168.2.0/24 to any in
4155 The following rules show some of the applications of
4159 for simulations and the like.
4161 This rule drops random incoming packets with a probability
4164 .Dl "ipfw add prob 0.05 deny ip from any to any in"
4166 A similar effect can be achieved making use of
4170 .Dl "ipfw add pipe 10 ip from any to any"
4171 .Dl "ipfw pipe 10 config plr 0.05"
4173 We can use pipes to artificially limit bandwidth, e.g.\& on a
4174 machine acting as a router, if we want to limit traffic from
4175 local clients on 192.168.2.0/24 we do:
4177 .Dl "ipfw add pipe 1 ip from 192.168.2.0/24 to any out"
4178 .Dl "ipfw pipe 1 config bw 300Kbit/s queue 50KBytes"
4180 note that we use the
4182 modifier so that the rule is not used twice.
4183 Remember in fact that
4185 rules are checked both on incoming and outgoing packets.
4187 Should we want to simulate a bidirectional link with bandwidth
4188 limitations, the correct way is the following:
4190 .Dl "ipfw add pipe 1 ip from any to any out"
4191 .Dl "ipfw add pipe 2 ip from any to any in"
4192 .Dl "ipfw pipe 1 config bw 64Kbit/s queue 10Kbytes"
4193 .Dl "ipfw pipe 2 config bw 64Kbit/s queue 10Kbytes"
4195 The above can be very useful, e.g.\& if you want to see how
4196 your fancy Web page will look for a residential user who
4197 is connected only through a slow link.
4198 You should not use only one pipe for both directions, unless
4199 you want to simulate a half-duplex medium (e.g.\& AppleTalk,
4201 It is not necessary that both pipes have the same configuration,
4202 so we can also simulate asymmetric links.
4204 Should we want to verify network performance with the RED queue
4205 management algorithm:
4207 .Dl "ipfw add pipe 1 ip from any to any"
4208 .Dl "ipfw pipe 1 config bw 500Kbit/s queue 100 red 0.002/30/80/0.1"
4210 Another typical application of the traffic shaper is to
4211 introduce some delay in the communication.
4212 This can significantly affect applications which do a lot of Remote
4213 Procedure Calls, and where the round-trip-time of the
4214 connection often becomes a limiting factor much more than
4217 .Dl "ipfw add pipe 1 ip from any to any out"
4218 .Dl "ipfw add pipe 2 ip from any to any in"
4219 .Dl "ipfw pipe 1 config delay 250ms bw 1Mbit/s"
4220 .Dl "ipfw pipe 2 config delay 250ms bw 1Mbit/s"
4222 Per-flow queueing can be useful for a variety of purposes.
4223 A very simple one is counting traffic:
4225 .Dl "ipfw add pipe 1 tcp from any to any"
4226 .Dl "ipfw add pipe 1 udp from any to any"
4227 .Dl "ipfw add pipe 1 ip from any to any"
4228 .Dl "ipfw pipe 1 config mask all"
4230 The above set of rules will create queues (and collect
4231 statistics) for all traffic.
4232 Because the pipes have no limitations, the only effect is
4233 collecting statistics.
4234 Note that we need 3 rules, not just the last one, because
4237 tries to match IP packets it will not consider ports, so we
4238 would not see connections on separate ports as different
4241 A more sophisticated example is limiting the outbound traffic
4242 on a net with per-host limits, rather than per-network limits:
4244 .Dl "ipfw add pipe 1 ip from 192.168.2.0/24 to any out"
4245 .Dl "ipfw add pipe 2 ip from any to 192.168.2.0/24 in"
4246 .Dl "ipfw pipe 1 config mask src-ip 0x000000ff bw 200Kbit/s queue 20Kbytes"
4247 .Dl "ipfw pipe 2 config mask dst-ip 0x000000ff bw 200Kbit/s queue 20Kbytes"
4249 In the following example, we need to create several traffic bandwidth
4250 classes and we need different hosts/networks to fall into different classes.
4251 We create one pipe for each class and configure them accordingly.
4252 Then we create a single table and fill it with IP subnets and addresses.
4253 For each subnet/host we set the argument equal to the number of the pipe
4255 Then we classify traffic using a single rule:
4257 .Dl "ipfw pipe 1 config bw 1000Kbyte/s"
4258 .Dl "ipfw pipe 4 config bw 4000Kbyte/s"
4260 .Dl "ipfw table T1 create type addr"
4261 .Dl "ipfw table T1 add 192.168.2.0/24 1"
4262 .Dl "ipfw table T1 add 192.168.0.0/27 4"
4263 .Dl "ipfw table T1 add 192.168.0.2 1"
4265 .Dl "ipfw add pipe tablearg ip from 'table(T1)' to any"
4269 action, the table entries may include hostnames and IP addresses.
4271 .Dl "ipfw table T2 create type addr ftype ip"
4272 .Dl "ipfw table T2 add 192.168.2.0/24 10.23.2.1"
4273 .Dl "ipfw table T21 add 192.168.0.0/27 router1.dmz"
4275 .Dl "ipfw add 100 fwd tablearg ip from any to table(1)"
4277 In the following example per-interface firewall is created:
4279 .Dl "ipfw table IN create type iface valtype skipto,fib"
4280 .Dl "ipfw table IN add vlan20 12000,12"
4281 .Dl "ipfw table IN add vlan30 13000,13"
4282 .Dl "ipfw table OUT create type iface valtype skipto"
4283 .Dl "ipfw table OUT add vlan20 22000"
4284 .Dl "ipfw table OUT add vlan30 23000"
4286 .Dl "ipfw add 100 setfib tablearg ip from any to any recv 'table(IN)' in"
4287 .Dl "ipfw add 200 skipto tablearg ip from any to any recv 'table(IN)' in"
4288 .Dl "ipfw add 300 skipto tablearg ip from any to any xmit 'table(OUT)' out"
4290 The following example illustrate usage of flow tables:
4292 .Dl "ipfw table fl create type flow:src-ip,proto,dst-ip,dst-port"
4293 .Dl "ipfw table fl add 2a02:6b8:77::88,tcp,2a02:6b8:77::99,80 11"
4294 .Dl "ipfw table fl add 10.0.0.1,udp,10.0.0.2,53 12"
4296 .Dl "ipfw add 100 allow ip from any to any flow 'table(fl,11)' recv ix0"
4298 To add a set of rules atomically, e.g.\& set 18:
4300 .Dl "ipfw set disable 18"
4301 .Dl "ipfw add NN set 18 ... # repeat as needed"
4302 .Dl "ipfw set enable 18"
4304 To delete a set of rules atomically the command is simply:
4306 .Dl "ipfw delete set 18"
4308 To test a ruleset and disable it and regain control if something goes wrong:
4310 .Dl "ipfw set disable 18"
4311 .Dl "ipfw add NN set 18 ... # repeat as needed"
4312 .Dl "ipfw set enable 18; echo done; sleep 30 && ipfw set disable 18"
4314 Here if everything goes well, you press control-C before the "sleep"
4315 terminates, and your ruleset will be left active.
4316 Otherwise, e.g.\& if
4317 you cannot access your box, the ruleset will be disabled after
4318 the sleep terminates thus restoring the previous situation.
4320 To show rules of the specific set:
4322 .Dl "ipfw set 18 show"
4324 To show rules of the disabled set:
4326 .Dl "ipfw -S set 18 show"
4328 To clear a specific rule counters of the specific set:
4330 .Dl "ipfw set 18 zero NN"
4332 To delete a specific rule of the specific set:
4334 .Dl "ipfw set 18 delete NN"
4335 .Ss NAT, REDIRECT AND LSNAT
4336 First redirect all the traffic to nat instance 123:
4338 .Dl "ipfw add nat 123 all from any to any"
4340 Then to configure nat instance 123 to alias all the outgoing traffic with ip
4341 192.168.0.123, blocking all incoming connections, trying to keep
4342 same ports on both sides, clearing aliasing table on address change
4343 and keeping a log of traffic/link statistics:
4345 .Dl "ipfw nat 123 config ip 192.168.0.123 log deny_in reset same_ports"
4347 Or to change address of instance 123, aliasing table will be cleared (see
4350 .Dl "ipfw nat 123 config ip 10.0.0.1"
4352 To see configuration of nat instance 123:
4354 .Dl "ipfw nat 123 show config"
4356 To show logs of all the instances in range 111-999:
4358 .Dl "ipfw nat 111-999 show"
4360 To see configurations of all instances:
4362 .Dl "ipfw nat show config"
4364 Or a redirect rule with mixed modes could looks like:
4366 .Dl "ipfw nat 123 config redirect_addr 10.0.0.1 10.0.0.66"
4367 .Dl " redirect_port tcp 192.168.0.1:80 500"
4368 .Dl " redirect_proto udp 192.168.1.43 192.168.1.1"
4369 .Dl " redirect_addr 192.168.0.10,192.168.0.11"
4370 .Dl " 10.0.0.100 # LSNAT"
4371 .Dl " redirect_port tcp 192.168.0.1:80,192.168.0.10:22"
4374 or it could be split in:
4376 .Dl "ipfw nat 1 config redirect_addr 10.0.0.1 10.0.0.66"
4377 .Dl "ipfw nat 2 config redirect_port tcp 192.168.0.1:80 500"
4378 .Dl "ipfw nat 3 config redirect_proto udp 192.168.1.43 192.168.1.1"
4379 .Dl "ipfw nat 4 config redirect_addr 192.168.0.10,192.168.0.11,192.168.0.12"
4381 .Dl "ipfw nat 5 config redirect_port tcp"
4382 .Dl " 192.168.0.1:80,192.168.0.10:22,192.168.0.20:25 500"
4384 Sometimes you may want to mix NAT and dynamic rules. It could be achived with
4388 options. Problem is, you need to create dynamic rule before NAT and check it
4389 after NAT actions (or vice versa) to have consistent addresses and ports.
4392 option will trigger activation of existing dynamic state, and action of such
4393 rule will be performed as soon as rule is matched. In case of NAT and
4395 rule packet need to be passed to NAT, not allowed as soon is possible.
4397 There is example of set of rules to achive this. Bear in mind that this
4398 is exmaple only and it is not very usefult by itself.
4400 On way out, after all checks place this rules:
4402 .Dl "ipfw add allow record-state skip-action"
4403 .Dl "ipfw add nat 1"
4405 And on way in there should be something like this:
4407 .Dl "ipfw add nat 1"
4408 .Dl "ipfw add check-state"
4410 Please note, that first rule on way out doesn't allow packet and doesn't
4411 execute existing dynamic rules. All it does, create new dynamic rule with
4413 action, if it is not created yet. Later, this dynamic rule is used on way
4417 .Ss CONFIGURING CODEL, PIE, FQ-CODEL and FQ-PIE AQM
4421 AQM can be configured for
4431 AQM using default configuration for traffic from 192.168.0.0/24 and 1Mbits/s
4434 .Dl "ipfw pipe 1 config bw 1mbits/s codel"
4435 .Dl "ipfw add 100 pipe 1 ip from 192.168.0.0/24 to any"
4441 AQM using different configurations parameters for traffic from
4442 192.168.0.0/24 and 1Mbits/s rate limit, we do:
4444 .Dl "ipfw pipe 1 config bw 1mbits/s"
4445 .Dl "ipfw queue 1 config pipe 1 codel target 8ms interval 160ms ecn"
4446 .Dl "ipfw add 100 queue 1 ip from 192.168.0.0/24 to any"
4452 AQM using default configuration for traffic from 192.168.0.0/24 and 1Mbits/s
4455 .Dl "ipfw pipe 1 config bw 1mbits/s pie"
4456 .Dl "ipfw add 100 pipe 1 ip from 192.168.0.0/24 to any"
4462 AQM using different configuration parameters for traffic from
4463 192.168.0.0/24 and 1Mbits/s rate limit, we do:
4465 .Dl "ipfw pipe 1 config bw 1mbits/s"
4466 .Dl "ipfw queue 1 config pipe 1 pie target 20ms tupdate 30ms ecn"
4467 .Dl "ipfw add 100 queue 1 ip from 192.168.0.0/24 to any"
4472 AQM can be configured for
4478 scheduler using different configurations parameters for traffic from
4479 192.168.0.0/24 and 1Mbits/s rate limit, we do:
4481 .Dl "ipfw pipe 1 config bw 1mbits/s"
4482 .Dl "ipfw sched 1 config pipe 1 type fq_codel"
4483 .Dl "ipfw queue 1 config sched 1"
4484 .Dl "ipfw add 100 queue 1 ip from 192.168.0.0/24 to any"
4488 default configuration for a
4490 such as disable ECN and change the
4494 .Dl "ipfw sched 1 config pipe 1 type fq_codel target 10ms noecn"
4500 scheduler using different configurations parameters for traffic from
4501 192.168.0.0/24 and 1Mbits/s rate limit, we do:
4503 .Dl "ipfw pipe 1 config bw 1mbits/s"
4504 .Dl "ipfw sched 1 config pipe 1 type fq_pie"
4505 .Dl "ipfw queue 1 config sched 1"
4506 .Dl "ipfw add 100 queue 1 ip from 192.168.0.0/24 to any"
4508 The configurations of
4511 can be changed in a similar way as for
4535 utility first appeared in
4540 Stateful extensions were introduced in
4543 was introduced in Summer 2002.
4545 .An Ugen J. S. Antsilevich ,
4546 .An Poul-Henning Kamp ,
4550 .An Rasool Al-Saadi .
4553 API based upon code written by
4557 Dummynet has been introduced by Luigi Rizzo in 1997-1998.
4559 Some early work (1999-2000) on the
4561 traffic shaper supported by Akamba Corp.
4563 The ipfw core (ipfw2) has been completely redesigned and
4564 reimplemented by Luigi Rizzo in summer 2002.
4567 options have been added by various developer over the years.
4570 In-kernel NAT support written by
4571 .An Paolo Pisati Aq Mt piso@FreeBSD.org
4572 as part of a Summer of Code 2005 project.
4576 support has been developed by
4577 .An The Centre for Advanced Internet Architectures (CAIA) Aq http://www.caia.swin.edu.au .
4578 The primary developers and maintainers are David Hayes and Jason But.
4579 For further information visit:
4580 .Aq http://www.caia.swin.edu.au/urp/SONATA
4582 Delay profiles have been developed by Alessandro Cerri and
4583 Luigi Rizzo, supported by the
4584 European Commission within Projects Onelab and Onelab2.
4586 CoDel, PIE, FQ-CoDel and FQ-PIE AQM for Dummynet have been implemented by
4587 .An The Centre for Advanced Internet Architectures (CAIA)
4588 in 2016, supported by The Comcast Innovation Fund.
4589 The primary developer is
4592 The syntax has grown over the years and sometimes it might be confusing.
4593 Unfortunately, backward compatibility prevents cleaning up mistakes
4594 made in the definition of the syntax.
4598 Misconfiguring the firewall can put your computer in an unusable state,
4599 possibly shutting down network services and requiring console access to
4600 regain control of it.
4602 Incoming packet fragments diverted by
4604 are reassembled before delivery to the socket.
4605 The action used on those packet is the one from the
4606 rule which matches the first fragment of the packet.
4608 Packets diverted to userland, and then reinserted by a userland process
4609 may lose various packet attributes.
4610 The packet source interface name
4611 will be preserved if it is shorter than 8 bytes and the userland process
4612 saves and reuses the sockaddr_in
4615 otherwise, it may be lost.
4616 If a packet is reinserted in this manner, later rules may be incorrectly
4617 applied, making the order of
4619 rules in the rule sequence very important.
4621 Dummynet drops all packets with IPv6 link-local addresses.
4627 may not behave as expected.
4628 In particular, incoming SYN packets may
4629 have no uid or gid associated with them since they do not yet belong
4630 to a TCP connection, and the uid/gid associated with a packet may not
4631 be as expected if the associated process calls
4633 or similar system calls.
4635 Rule syntax is subject to the command line environment and some patterns
4636 may need to be escaped with the backslash character
4637 or quoted appropriately.
4639 Due to the architecture of
4641 ipfw nat is not compatible with the TCP segmentation offloading (TSO).
4642 Thus, to reliably nat your network traffic, please disable TSO
4646 ICMP error messages are not implicitly matched by dynamic rules
4647 for the respective conversations.
4648 To avoid failures of network error detection and path MTU discovery,
4649 ICMP error messages may need to be allowed explicitly through static
4656 actions may lead to confusing behaviour if ruleset has mistakes,
4657 and/or interaction with other subsystems (netgraph, dummynet, etc.) is used.
4658 One possible case for this is packet leaving
4660 in subroutine on the input pass, while later on output encountering unpaired
4663 As the call stack is kept intact after input pass, packet will suddenly
4664 return to the rule number used on input pass, not on output one.
4665 Order of processing should be checked carefully to avoid such mistakes.