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
46 .Brq Cm firewall | altq | one_pass | debug | verbose | dyn_keepalive
49 .Brq Cm firewall | altq | one_pass | debug | verbose | dyn_keepalive
53 .Cm table Ar number Cm add Ar addr Ns Oo / Ns Ar masklen Oc Op Ar value
55 .Cm table Ar number Cm delete Ar addr Ns Op / Ns Ar masklen
65 .Ss DUMMYNET CONFIGURATION (TRAFFIC SHAPER AND PACKET SCHEDULER)
67 .Brq Cm pipe | queue | sched
73 .Brq Cm pipe | queue | sched
74 .Brq Cm delete | list | show
97 utility is the user interface for controlling the
101 traffic shaper/packet scheduler, and the
102 in-kernel NAT services.
104 A firewall configuration, or
108 numbered from 1 to 65535.
109 Packets are passed to the firewall
110 from a number of different places in the protocol stack
111 (depending on the source and destination of the packet,
112 it is possible for the firewall to be
113 invoked multiple times on the same packet).
114 The packet passed to the firewall is compared
115 against each of the rules in the
118 (multiple rules with the same number are permitted, in which case
119 they are processed in order of insertion).
120 When a match is found, the action corresponding to the
121 matching rule is performed.
123 Depending on the action and certain system settings, packets
124 can be reinjected into the firewall at some rule after the
125 matching one for further processing.
127 A ruleset always includes a
129 rule (numbered 65535) which cannot be modified or deleted,
130 and matches all packets.
131 The action associated with the
137 depending on how the kernel is configured.
139 If the ruleset includes one or more rules with the
144 the firewall will have a
146 behaviour, i.e., upon a match it will create
148 i.e. rules that match packets with the same 5-tuple
149 (protocol, source and destination addresses and ports)
150 as the packet which caused their creation.
151 Dynamic rules, which have a limited lifetime, are checked
152 at the first occurrence of a
157 rule, and are typically used to open the firewall on-demand to
158 legitimate traffic only.
160 .Sx STATEFUL FIREWALL
163 Sections below for more information on the stateful behaviour of
166 All rules (including dynamic ones) have a few associated counters:
167 a packet count, a byte count, a log count and a timestamp
168 indicating the time of the last match.
169 Counters can be displayed or reset with
173 Each rule belongs to one of 32 different
177 commands to atomically manipulate sets, such as enable,
178 disable, swap sets, move all rules in a set to another
179 one, delete all rules in a set.
180 These can be useful to
181 install temporary configurations, or to test them.
184 for more information on
187 Rules can be added with the
189 command; deleted individually or in groups with the
191 command, and globally (except those in set 31) with the
193 command; displayed, optionally with the content of the
199 Finally, counters can be reset with the
206 The following general options are available when invoking
208 .Bl -tag -width indent
210 Show counter values when listing rules.
213 command implies this option.
215 Only show the action and the comment, not the body of a rule.
219 When entering or showing rules, print them in compact form,
220 i.e., omitting the "ip from any to any" string
221 when this does not carry any additional information.
223 When listing, show dynamic rules in addition to static ones.
227 is specified, also show expired dynamic rules.
229 Do not ask for confirmation for commands that can cause problems
232 If there is no tty associated with the process, this is implied.
234 When listing a table (see the
236 section below for more information on lookup tables), format values
237 as IP addresses. By default, values are shown as integers.
239 Only check syntax of the command strings, without actually passing
242 Try to resolve addresses and service names in output.
244 Be quiet when executing the
254 This is useful when updating rulesets by executing multiple
258 .Ql sh\ /etc/rc.firewall ) ,
259 or by processing a file with many
261 rules across a remote login session.
262 It also stops a table add or delete
263 from failing if the entry already exists or is not present.
265 The reason why this option may be important is that
266 for some of these actions,
268 may print a message; if the action results in blocking the
269 traffic to the remote client,
270 the remote login session will be closed
271 and the rest of the ruleset will not be processed.
272 Access to the console would then be required to recover.
274 When listing rules, show the
276 each rule belongs to.
277 If this flag is not specified, disabled rules will not be
280 When listing pipes, sort according to one of the four
281 counters (total or current packets or bytes).
283 When listing, show last match timestamp converted with ctime().
285 When listing, show last match timestamp as seconds from the epoch.
286 This form can be more convenient for postprocessing by scripts.
289 .Ss LIST OF RULES AND PREPROCESSING
290 To ease configuration, rules can be put into a file which is
293 as shown in the last synopsis line.
297 The file will be read line by line and applied as arguments to the
301 Optionally, a preprocessor can be specified using
305 is to be piped through.
306 Useful preprocessors include
312 does not start with a slash
314 as its first character, the usual
316 name search is performed.
317 Care should be taken with this in environments where not all
318 file systems are mounted (yet) by the time
320 is being run (e.g.\& when they are mounted over NFS).
323 has been specified, any additional arguments are passed on to the preprocessor
325 This allows for flexible configuration files (like conditionalizing
326 them on the local hostname) and the use of macros to centralize
327 frequently required arguments like IP addresses.
329 .Ss TRAFFIC SHAPER CONFIGURATION
335 commands are used to configure the traffic shaper and packet scheduler.
337 .Sx TRAFFIC SHAPER (DUMMYNET) CONFIGURATION
338 Section below for details.
340 If the world and the kernel get out of sync the
342 ABI may break, preventing you from being able to add any rules.
344 adversely effect the booting process.
349 to temporarily disable the firewall to regain access to the network,
350 allowing you to fix the problem.
352 A packet is checked against the active ruleset in multiple places
353 in the protocol stack, under control of several sysctl variables.
354 These places and variables are shown below, and it is important to
355 have this picture in mind in order to design a correct ruleset.
356 .Bd -literal -offset indent
359 +----------->-----------+
361 [ip(6)_input] [ip(6)_output] net.inet(6).ip(6).fw.enable=1
364 [ether_demux] [ether_output_frame] net.link.ether.ipfw=1
366 +-->--[bdg_forward]-->--+ net.link.bridge.ipfw=1
372 times the same packet goes through the firewall can
373 vary between 0 and 4 depending on packet source and
374 destination, and system configuration.
376 Note that as packets flow through the stack, headers can be
377 stripped or added to it, and so they may or may not be available
379 E.g., incoming packets will include the MAC header when
383 but the same packets will have the MAC header stripped off when
390 Also note that each packet is always checked against the complete ruleset,
391 irrespective of the place where the check occurs, or the source of the packet.
392 If a rule contains some match patterns or actions which are not valid
393 for the place of invocation (e.g.\& trying to match a MAC header within
397 the match pattern will not match, but a
399 operator in front of such patterns
403 match on those packets.
404 It is thus the responsibility of
405 the programmer, if necessary, to write a suitable ruleset to
406 differentiate among the possible places.
408 rules can be useful here, as an example:
409 .Bd -literal -offset indent
410 # packets from ether_demux or bdg_forward
411 ipfw add 10 skipto 1000 all from any to any layer2 in
412 # packets from ip_input
413 ipfw add 10 skipto 2000 all from any to any not layer2 in
414 # packets from ip_output
415 ipfw add 10 skipto 3000 all from any to any not layer2 out
416 # packets from ether_output_frame
417 ipfw add 10 skipto 4000 all from any to any layer2 out
420 (yes, at the moment there is no way to differentiate between
421 ether_demux and bdg_forward).
423 In general, each keyword or argument must be provided as
424 a separate command line argument, with no leading or trailing
426 Keywords are case-sensitive, whereas arguments may
427 or may not be case-sensitive depending on their nature
428 (e.g.\& uid's are, hostnames are not).
430 Some arguments (e.g. port or address lists) are comma-separated
432 In this case, spaces after commas ',' are allowed to make
433 the line more readable.
434 You can also put the entire
435 command (including flags) into a single argument.
436 E.g., the following forms are equivalent:
437 .Bd -literal -offset indent
438 ipfw -q add deny src-ip 10.0.0.0/24,127.0.0.1/8
439 ipfw -q add deny src-ip 10.0.0.0/24, 127.0.0.1/8
440 ipfw "-q add deny src-ip 10.0.0.0/24, 127.0.0.1/8"
443 The format of firewall rules is the following:
444 .Bd -ragged -offset indent
447 .Op Cm set Ar set_number
448 .Op Cm prob Ar match_probability
450 .Op Cm log Op Cm logamount Ar number
460 where the body of the rule specifies which information is used
461 for filtering packets, among the following:
463 .Bl -tag -width "Source and dest. addresses and ports" -offset XXX -compact
464 .It Layer-2 header fields
466 .It IPv4 and IPv6 Protocol
468 .It Source and dest. addresses and ports
472 .It Transmit and receive interface
474 .It Misc. IP header fields
475 Version, type of service, datagram length, identification,
476 fragment flag (non-zero IP offset),
479 .It IPv6 Extension headers
480 Fragmentation, Hop-by-Hop options,
481 Routing Headers, Source routing rthdr0, Mobile IPv6 rthdr2, IPSec options.
483 .It Misc. TCP header fields
484 TCP flags (SYN, FIN, ACK, RST, etc.),
485 sequence number, acknowledgment number,
493 When the packet can be associated with a local socket.
495 Whether a packet came from a divert socket (e.g.,
497 .It Fib annotation state
498 Whether a packet has been tagged for using a specific FIB (routing table)
499 in future forwarding decisions.
502 Note that some of the above information, e.g.\& source MAC or IP addresses and
503 TCP/UDP ports, can be easily spoofed, so filtering on those fields
504 alone might not guarantee the desired results.
505 .Bl -tag -width indent
507 Each rule is associated with a
509 in the range 1..65535, with the latter reserved for the
512 Rules are checked sequentially by rule number.
513 Multiple rules can have the same number, in which case they are
514 checked (and listed) according to the order in which they have
516 If a rule is entered without specifying a number, the kernel will
517 assign one in such a way that the rule becomes the last one
521 Automatic rule numbers are assigned by incrementing the last
522 non-default rule number by the value of the sysctl variable
523 .Ar net.inet.ip.fw.autoinc_step
524 which defaults to 100.
525 If this is not possible (e.g.\& because we would go beyond the
526 maximum allowed rule number), the number of the last
527 non-default value is used instead.
528 .It Cm set Ar set_number
529 Each rule is associated with a
532 Sets can be individually disabled and enabled, so this parameter
533 is of fundamental importance for atomic ruleset manipulation.
534 It can be also used to simplify deletion of groups of rules.
535 If a rule is entered without specifying a set number,
538 Set 31 is special in that it cannot be disabled,
539 and rules in set 31 are not deleted by the
541 command (but you can delete them with the
542 .Nm ipfw delete set 31
544 Set 31 is also used for the
547 .It Cm prob Ar match_probability
548 A match is only declared with the specified probability
549 (floating point number between 0 and 1).
550 This can be useful for a number of applications such as
551 random packet drop or
554 to simulate the effect of multiple paths leading to out-of-order
557 Note: this condition is checked before any other condition, including
558 ones such as keep-state or check-state which might have side effects.
559 .It Cm log Op Cm logamount Ar number
560 Packets matching a rule with the
562 keyword will be made available for logging in two ways:
563 if the sysctl variable
564 .Va net.inet.ip.fw.verbose
565 is set to 0 (default), one can use
569 pseudo interface. There is no overhead if no
571 is attached to the pseudo interface.
574 .Va net.inet.ip.fw.verbose
575 is set to 1, packets will be logged to
579 facility up to a maximum of
584 is specified, the limit is taken from the sysctl variable
585 .Va net.inet.ip.fw.verbose_limit .
586 In both cases, a value of 0 means unlimited logging.
588 Once the limit is reached, logging can be re-enabled by
589 clearing the logging counter or the packet counter for that entry, see the
593 Note: logging is done after all other packet matching conditions
594 have been successfully verified, and before performing the final
595 action (accept, deny, etc.) on the packet.
597 When a packet matches a rule with the
599 keyword, the numeric tag for the given
601 in the range 1..65534 will be attached to the packet.
602 The tag acts as an internal marker (it is not sent out over
603 the wire) that can be used to identify these packets later on.
604 This can be used, for example, to provide trust between interfaces
605 and to start doing policy-based filtering.
606 A packet can have multiple tags at the same time.
607 Tags are "sticky", meaning once a tag is applied to a packet by a
608 matching rule it exists until explicit removal.
609 Tags are kept with the packet everywhere within the kernel, but are
610 lost when packet leaves the kernel, for example, on transmitting
611 packet out to the network or sending packet to a
615 To check for previously applied tags, use the
618 To delete previously applied tag, use the
622 Note: since tags are kept with the packet everywhere in kernelspace,
623 they can be set and unset anywhere in the kernel network subsystem
626 facility), not only by means of the
632 For example, there can be a specialized
634 node doing traffic analyzing and tagging for later inspecting
636 .It Cm untag Ar number
637 When a packet matches a rule with the
639 keyword, the tag with the number
641 is searched among the tags attached to this packet and,
642 if found, removed from it.
643 Other tags bound to packet, if present, are left untouched.
645 When a packet matches a rule with the
647 keyword, the ALTQ identifier for the given
652 Note that this ALTQ tag is only meaningful for packets going "out" of IPFW,
653 and not being rejected or going to divert sockets.
654 Note that if there is insufficient memory at the time the packet is
655 processed, it will not be tagged, so it is wise to make your ALTQ
656 "default" queue policy account for this.
659 rules match a single packet, only the first one adds the ALTQ classification
661 In doing so, traffic may be shaped by using
662 .Cm count Cm altq Ar queue
663 rules for classification early in the ruleset, then later applying
664 the filtering decision.
669 rules may come later and provide the actual filtering decisions in
670 addition to the fallback ALTQ tag.
674 to set up the queues before IPFW will be able to look them up by name,
675 and if the ALTQ disciplines are rearranged, the rules in containing the
676 queue identifiers in the kernel will likely have gone stale and need
678 Stale queue identifiers will probably result in misclassification.
680 All system ALTQ processing can be turned on or off via
685 .Cm disable Ar altq .
687 .Va net.inet.ip.fw.one_pass
688 is irrelevant to ALTQ traffic shaping, as the actual rule action is followed
689 always after adding an ALTQ tag.
692 A rule can be associated with one of the following actions, which
693 will be executed when the packet matches the body of the rule.
694 .Bl -tag -width indent
695 .It Cm allow | accept | pass | permit
696 Allow packets that match rule.
697 The search terminates.
699 Checks the packet against the dynamic ruleset.
700 If a match is found, execute the action associated with
701 the rule which generated this dynamic rule, otherwise
702 move to the next rule.
705 rules do not have a body.
708 rule is found, the dynamic ruleset is checked at the first
714 Update counters for all packets that match rule.
715 The search continues with the next rule.
717 Discard packets that match this rule.
718 The search terminates.
719 .It Cm divert Ar port
720 Divert packets that match this rule to the
724 The search terminates.
725 .It Cm fwd | forward Ar ipaddr | tablearg Ns Op , Ns Ar port
726 Change the next-hop on matching packets to
728 which can be an IP address or a host name.
729 For IPv4, the next hop can also be supplied by the last table
730 looked up for the packet by using the
732 keyword instead of an explicit address.
733 The search terminates if this rule matches.
737 is a local address, then matching packets will be forwarded to
739 (or the port number in the packet if one is not specified in the rule)
740 on the local machine.
744 is not a local address, then the port number
745 (if specified) is ignored, and the packet will be
746 forwarded to the remote address, using the route as found in
747 the local routing table for that IP.
751 rule will not match layer-2 packets (those received
752 on ether_input, ether_output, or bridged).
756 action does not change the contents of the packet at all.
757 In particular, the destination address remains unmodified, so
758 packets forwarded to another system will usually be rejected by that system
759 unless there is a matching rule on that system to capture them.
760 For packets forwarded locally,
761 the local address of the socket will be
762 set to the original destination address of the packet.
765 entry look rather weird but is intended for
766 use with transparent proxy servers.
767 .It Cm nat Ar nat_nr | tablearg
770 (for network address translation, address redirect, etc.):
772 .Sx NETWORK ADDRESS TRANSLATION (NAT)
773 Section for further information.
774 .It Cm pipe Ar pipe_nr
778 (for bandwidth limitation, delay, etc.).
780 .Sx TRAFFIC SHAPER (DUMMYNET) CONFIGURATION
781 Section for further information.
782 The search terminates; however, on exit from the pipe and if
786 .Va net.inet.ip.fw.one_pass
787 is not set, the packet is passed again to the firewall code
788 starting from the next rule.
789 .It Cm queue Ar queue_nr
793 (for bandwidth limitation using WF2Q+).
799 Discard packets that match this rule, and if the
800 packet is a TCP packet, try to send a TCP reset (RST) notice.
801 The search terminates.
803 Discard packets that match this rule, and if the
804 packet is a TCP packet, try to send a TCP reset (RST) notice.
805 The search terminates.
806 .It Cm skipto Ar number | tablearg
807 Skip all subsequent rules numbered less than
809 The search continues with the first rule numbered
812 It is possible to use the
814 keyword with a skipto for a
816 skipto, but care should be used, as no destination caching
817 is possible in this case so the rules are always walked to find it,
820 .It Cm call Ar number | tablearg
821 The current rule number is saved in the internal stack and
822 ruleset processing continues with the first rule numbered
825 If later a rule with the
827 action is encountered, the processing returns to the first rule
830 rule plus one or higher
831 (the same behaviour as with packets returning from
836 This could be used to make somewhat like an assembly language
838 calls to rules with common checks for different interfaces, etc.
840 Rule with any number could be called, not just forward jumps as with
842 So, to prevent endless loops in case of mistakes, both
846 actions don't do any jumps and simply go to the next rule if memory
847 can't be allocated or stack overflowed/undeflowed.
849 Internally stack for rule numbers is implemented using
851 facility and currently has size of 16 entries.
852 As mbuf tags are lost when packet leaves the kernel,
854 should not be used in subroutines to avoid endless loops
855 and other undesired effects.
857 Takes rule number saved to internal stack by the last
859 action and returns ruleset processing to the first rule
860 with number greater than number of corresponding
862 rule. See description of the
864 action for more details.
870 and thus are unconditional, but
872 command-line utility currently requires every action except
875 While it is sometimes useful to return only on some packets,
876 usually you want to print just
879 A workaround for this is to use new syntax and
883 .Bd -literal -offset indent
884 # Add a rule without actual body
885 ipfw add 2999 return via any
887 # List rules without "from any to any" part
891 This cosmetic annoyance may be fixed in future releases.
893 Send a copy of packets matching this rule to the
897 The search continues with the next rule.
898 .It Cm unreach Ar code
899 Discard packets that match this rule, and try to send an ICMP
900 unreachable notice with code
904 is a number from 0 to 255, or one of these aliases:
905 .Cm net , host , protocol , port ,
906 .Cm needfrag , srcfail , net-unknown , host-unknown ,
907 .Cm isolated , net-prohib , host-prohib , tosnet ,
908 .Cm toshost , filter-prohib , host-precedence
910 .Cm precedence-cutoff .
911 The search terminates.
912 .It Cm unreach6 Ar code
913 Discard packets that match this rule, and try to send an ICMPv6
914 unreachable notice with code
918 is a number from 0, 1, 3 or 4, or one of these aliases:
919 .Cm no-route, admin-prohib, address
922 The search terminates.
923 .It Cm netgraph Ar cookie
924 Divert packet into netgraph with given
926 The search terminates.
927 If packet is later returned from netgraph it is either
928 accepted or continues with the next rule, depending on
929 .Va net.inet.ip.fw.one_pass
931 .It Cm ngtee Ar cookie
932 A copy of packet is diverted into netgraph, original
933 packet continues with the next rule.
936 for more information on
941 .It Cm setfib Ar fibnum | tablearg
942 The packet is tagged so as to use the FIB (routing table)
944 in any subsequent forwarding decisions.
945 Initially this is limited to the values 0 through 15, see
947 Processing continues at the next rule.
948 It is possible to use the
950 keyword with a setfib. If tablearg value is not within compiled FIB range packet fib is set to 0.
951 .It Cm setdscp Ar DSCP | number | tablearg
952 Set specified DiffServ codepoint for an IPv4/IPv6 packet.
953 Processing continues at the next rule.
954 Supported values are:
1000 Additionally, DSCP value can be specified by number (0..64).
1001 It is also possible to use the
1003 keyword with setdscp.
1004 If the tablearg value is not within the 0..64 range, lower 6 bits of supplied
1007 Queue and reassemble ip fragments.
1008 If the packet is not fragmented, counters are updated and processing continues with the next rule.
1009 If the packet is the last logical fragment, the packet is reassembled and, if
1010 .Va net.inet.ip.fw.one_pass
1011 is set to 0, processing continues with the next rule, else packet is allowed to pass and search terminates.
1012 If the packet is a fragment in the middle, it is consumed and processing stops immediately.
1014 Fragments handling can be tuned via
1015 .Va net.inet.ip.maxfragpackets
1017 .Va net.inet.ip.maxfragsperpacket
1018 which limit, respectively, the maximum number of processable fragments (default: 800) and
1019 the maximum number of fragments per packet (default: 16).
1021 NOTA BENE: since fragments do not contain port numbers, they should be avoided with the
1024 Alternatively, direction-based (like
1028 ) and source-based (like
1030 ) match patterns can be used to select fragments.
1032 Usually a simple rule like:
1033 .Bd -literal -offset indent
1034 # reassemble incoming fragments
1035 ipfw add reass all from any to any in
1038 is all you need at the beginning of your ruleset.
1041 The body of a rule contains zero or more patterns (such as
1042 specific source and destination addresses or ports,
1043 protocol options, incoming or outgoing interfaces, etc.)
1044 that the packet must match in order to be recognised.
1045 In general, the patterns are connected by (implicit)
1047 operators -- i.e., all must match in order for the
1049 Individual patterns can be prefixed by the
1051 operator to reverse the result of the match, as in
1053 .Dl "ipfw add 100 allow ip from not 1.2.3.4 to any"
1055 Additionally, sets of alternative match patterns
1057 can be constructed by putting the patterns in
1058 lists enclosed between parentheses ( ) or braces { }, and
1061 operator as follows:
1063 .Dl "ipfw add 100 allow ip from { x or not y or z } to any"
1065 Only one level of parentheses is allowed.
1066 Beware that most shells have special meanings for parentheses
1067 or braces, so it is advisable to put a backslash \\ in front of them
1068 to prevent such interpretations.
1070 The body of a rule must in general include a source and destination
1074 can be used in various places to specify that the content of
1075 a required field is irrelevant.
1077 The rule body has the following format:
1078 .Bd -ragged -offset indent
1079 .Op Ar proto Cm from Ar src Cm to Ar dst
1083 The first part (proto from src to dst) is for backward
1084 compatibility with earlier versions of
1088 any match pattern (including MAC headers, IP protocols,
1089 addresses and ports) can be specified in the
1093 Rule fields have the following meaning:
1094 .Bl -tag -width indent
1095 .It Ar proto : protocol | Cm { Ar protocol Cm or ... }
1096 .It Ar protocol : Oo Cm not Oc Ar protocol-name | protocol-number
1097 An IP protocol specified by number or name
1098 (for a complete list see
1099 .Pa /etc/protocols ) ,
1100 or one of the following keywords:
1101 .Bl -tag -width indent
1103 Matches IPv4 packets.
1105 Matches IPv6 packets.
1114 option will be treated as inner protocol.
1122 .Cm { Ar protocol Cm or ... }
1125 is provided for convenience only but its use is deprecated.
1126 .It Ar src No and Ar dst : Bro Cm addr | Cm { Ar addr Cm or ... } Brc Op Oo Cm not Oc Ar ports
1127 An address (or a list, see below)
1128 optionally followed by
1134 with multiple addresses) is provided for convenience only and
1135 its use is discouraged.
1136 .It Ar addr : Oo Cm not Oc Bro
1137 .Cm any | me | me6 |
1138 .Cm table Ns Pq Ar number Ns Op , Ns Ar value
1139 .Ar | addr-list | addr-set
1141 .Bl -tag -width indent
1143 matches any IP address.
1145 matches any IP address configured on an interface in the system.
1147 matches any IPv6 address configured on an interface in the system.
1148 The address list is evaluated at the time the packet is
1150 .It Cm table Ns Pq Ar number Ns Op , Ns Ar value
1151 Matches any IPv4 address for which an entry exists in the lookup table
1153 If an optional 32-bit unsigned
1155 is also specified, an entry will match only if it has this value.
1158 section below for more information on lookup tables.
1160 .It Ar addr-list : ip-addr Ns Op Ns , Ns Ar addr-list
1162 A host or subnet address specified in one of the following ways:
1163 .Bl -tag -width indent
1164 .It Ar numeric-ip | hostname
1165 Matches a single IPv4 address, specified as dotted-quad or a hostname.
1166 Hostnames are resolved at the time the rule is added to the firewall list.
1167 .It Ar addr Ns / Ns Ar masklen
1168 Matches all addresses with base
1170 (specified as an IP address, a network number, or a hostname)
1174 As an example, 1.2.3.4/25 or 1.2.3.0/25 will match
1175 all IP numbers from 1.2.3.0 to 1.2.3.127 .
1176 .It Ar addr Ns : Ns Ar mask
1177 Matches all addresses with base
1179 (specified as an IP address, a network number, or a hostname)
1182 specified as a dotted quad.
1183 As an example, 1.2.3.4:255.0.255.0 or 1.0.3.0:255.0.255.0 will match
1185 This form is advised only for non-contiguous
1187 It is better to resort to the
1188 .Ar addr Ns / Ns Ar masklen
1189 format for contiguous masks, which is more compact and less
1192 .It Ar addr-set : addr Ns Oo Ns / Ns Ar masklen Oc Ns Cm { Ns Ar list Ns Cm }
1193 .It Ar list : Bro Ar num | num-num Brc Ns Op Ns , Ns Ar list
1194 Matches all addresses with base address
1196 (specified as an IP address, a network number, or a hostname)
1197 and whose last byte is in the list between braces { } .
1198 Note that there must be no spaces between braces and
1199 numbers (spaces after commas are allowed).
1200 Elements of the list can be specified as single entries
1204 field is used to limit the size of the set of addresses,
1205 and can have any value between 24 and 32.
1207 it will be assumed as 24.
1209 This format is particularly useful to handle sparse address sets
1210 within a single rule.
1211 Because the matching occurs using a
1212 bitmask, it takes constant time and dramatically reduces
1213 the complexity of rulesets.
1215 As an example, an address specified as 1.2.3.4/24{128,35-55,89}
1216 or 1.2.3.0/24{128,35-55,89}
1217 will match the following IP addresses:
1219 1.2.3.128, 1.2.3.35 to 1.2.3.55, 1.2.3.89 .
1220 .It Ar addr6-list : ip6-addr Ns Op Ns , Ns Ar addr6-list
1222 A host or subnet specified one of the following ways:
1223 .Bl -tag -width indent
1224 .It Ar numeric-ip | hostname
1225 Matches a single IPv6 address as allowed by
1228 Hostnames are resolved at the time the rule is added to the firewall
1230 .It Ar addr Ns / Ns Ar masklen
1231 Matches all IPv6 addresses with base
1233 (specified as allowed by
1241 No support for sets of IPv6 addresses is provided because IPv6 addresses
1242 are typically random past the initial prefix.
1243 .It Ar ports : Bro Ar port | port Ns \&- Ns Ar port Ns Brc Ns Op , Ns Ar ports
1244 For protocols which support port numbers (such as TCP and UDP), optional
1246 may be specified as one or more ports or port ranges, separated
1247 by commas but no spaces, and an optional
1252 notation specifies a range of ports (including boundaries).
1256 may be used instead of numeric port values.
1257 The length of the port list is limited to 30 ports or ranges,
1258 though one can specify larger ranges by using an
1262 section of the rule.
1266 can be used to escape the dash
1268 character in a service name (from a shell, the backslash must be
1269 typed twice to avoid the shell itself interpreting it as an escape
1272 .Dl "ipfw add count tcp from any ftp\e\e-data-ftp to any"
1274 Fragmented packets which have a non-zero offset (i.e., not the first
1275 fragment) will never match a rule which has one or more port
1279 option for details on matching fragmented packets.
1281 .Ss RULE OPTIONS (MATCH PATTERNS)
1282 Additional match patterns can be used within
1284 Zero or more of these so-called
1286 can be present in a rule, optionally prefixed by the
1288 operand, and possibly grouped into
1291 The following match patterns can be used (listed in alphabetical order):
1292 .Bl -tag -width indent
1293 .It Cm // this is a comment.
1294 Inserts the specified text as a comment in the rule.
1295 Everything following // is considered as a comment and stored in the rule.
1296 You can have comment-only rules, which are listed as having a
1298 action followed by the comment.
1303 Matches only packets generated by a divert socket.
1304 .It Cm diverted-loopback
1305 Matches only packets coming from a divert socket back into the IP stack
1307 .It Cm diverted-output
1308 Matches only packets going from a divert socket back outward to the IP
1309 stack output for delivery.
1310 .It Cm dst-ip Ar ip-address
1311 Matches IPv4 packets whose destination IP is one of the address(es)
1312 specified as argument.
1313 .It Bro Cm dst-ip6 | dst-ipv6 Brc Ar ip6-address
1314 Matches IPv6 packets whose destination IP is one of the address(es)
1315 specified as argument.
1316 .It Cm dst-port Ar ports
1317 Matches IP packets whose destination port is one of the port(s)
1318 specified as argument.
1320 Matches TCP packets that have the RST or ACK bits set.
1321 .It Cm ext6hdr Ar header
1322 Matches IPv6 packets containing the extended header given by
1324 Supported headers are:
1330 any type of Routing Header
1332 Source routing Routing Header Type 0
1334 Mobile IPv6 Routing Header Type 2
1338 IPSec authentication headers
1340 and IPsec encapsulated security payload headers
1342 .It Cm fib Ar fibnum
1343 Matches a packet that has been tagged to use
1344 the given FIB (routing table) number.
1345 .It Cm flow-id Ar labels
1346 Matches IPv6 packets containing any of the flow labels given in
1349 is a comma separated list of numeric flow labels.
1351 Matches packets that are fragments and not the first
1352 fragment of an IP datagram.
1353 Note that these packets will not have
1354 the next protocol header (e.g.\& TCP, UDP) so options that look into
1355 these headers cannot match.
1357 Matches all TCP or UDP packets sent by or received for a
1361 may be specified by name or number.
1362 .It Cm jail Ar prisonID
1363 Matches all TCP or UDP packets sent by or received for the
1364 jail whos prison ID is
1366 .It Cm icmptypes Ar types
1367 Matches ICMP packets whose ICMP type is in the list
1369 The list may be specified as any combination of
1370 individual types (numeric) separated by commas.
1371 .Em Ranges are not allowed .
1372 The supported ICMP types are:
1376 destination unreachable
1384 router advertisement
1388 time-to-live exceeded
1400 address mask request
1402 and address mask reply
1404 .It Cm icmp6types Ar types
1405 Matches ICMP6 packets whose ICMP6 type is in the list of
1407 The list may be specified as any combination of
1408 individual types (numeric) separated by commas.
1409 .Em Ranges are not allowed .
1411 Matches incoming or outgoing packets, respectively.
1415 are mutually exclusive (in fact,
1419 .It Cm ipid Ar id-list
1420 Matches IPv4 packets whose
1422 field has value included in
1424 which is either a single value or a list of values or ranges
1425 specified in the same way as
1427 .It Cm iplen Ar len-list
1428 Matches IP packets whose total length, including header and data, is
1431 which is either a single value or a list of values or ranges
1432 specified in the same way as
1434 .It Cm ipoptions Ar spec
1435 Matches packets whose IPv4 header contains the comma separated list of
1436 options specified in
1438 The supported IP options are:
1441 (strict source route),
1443 (loose source route),
1445 (record packet route) and
1448 The absence of a particular option may be denoted
1451 .It Cm ipprecedence Ar precedence
1452 Matches IPv4 packets whose precedence field is equal to
1455 Matches packets that have IPSEC history associated with them
1456 (i.e., the packet comes encapsulated in IPSEC, the kernel
1457 has IPSEC support and IPSEC_FILTERTUNNEL option, and can correctly
1460 Note that specifying
1462 is different from specifying
1464 as the latter will only look at the specific IP protocol field,
1465 irrespective of IPSEC kernel support and the validity of the IPSEC data.
1467 Further note that this flag is silently ignored in kernels without
1469 It does not affect rule processing when given and the
1470 rules are handled as if with no
1473 .It Cm iptos Ar spec
1474 Matches IPv4 packets whose
1476 field contains the comma separated list of
1477 service types specified in
1479 The supported IP types of service are:
1482 .Pq Dv IPTOS_LOWDELAY ,
1484 .Pq Dv IPTOS_THROUGHPUT ,
1486 .Pq Dv IPTOS_RELIABILITY ,
1488 .Pq Dv IPTOS_MINCOST ,
1490 .Pq Dv IPTOS_ECN_CE .
1491 The absence of a particular type may be denoted
1494 .It Cm dscp spec Ns Op , Ns Ar spec
1495 Matches IPv4/IPv6 packets whose
1497 field value is contained in
1500 Multiple values can be specified via
1501 the comma separated list.
1502 Value can be one of keywords used in
1504 action or exact number.
1505 .It Cm ipttl Ar ttl-list
1506 Matches IPv4 packets whose time to live is included in
1508 which is either a single value or a list of values or ranges
1509 specified in the same way as
1511 .It Cm ipversion Ar ver
1512 Matches IP packets whose IP version field is
1515 Upon a match, the firewall will create a dynamic rule, whose
1516 default behaviour is to match bidirectional traffic between
1517 source and destination IP/port using the same protocol.
1518 The rule has a limited lifetime (controlled by a set of
1520 variables), and the lifetime is refreshed every time a matching
1523 Matches only layer2 packets, i.e., those passed to
1525 from ether_demux() and ether_output_frame().
1526 .It Cm limit Bro Cm src-addr | src-port | dst-addr | dst-port Brc Ar N
1527 The firewall will only allow
1529 connections with the same
1530 set of parameters as specified in the rule.
1532 of source and destination addresses and ports can be
1535 only IPv4 flows are supported.
1536 .It Cm lookup Bro Cm dst-ip | dst-port | src-ip | src-port | uid | jail Brc Ar N
1537 Search an entry in lookup table
1539 that matches the field specified as argument.
1540 If not found, the match fails.
1541 Otherwise, the match succeeds and
1543 is set to the value extracted from the table.
1545 This option can be useful to quickly dispatch traffic based on
1546 certain packet fields.
1549 section below for more information on lookup tables.
1550 .It Cm { MAC | mac } Ar dst-mac src-mac
1551 Match packets with a given
1555 addresses, specified as the
1557 keyword (matching any MAC address), or six groups of hex digits
1558 separated by colons,
1559 and optionally followed by a mask indicating the significant bits.
1560 The mask may be specified using either of the following methods:
1561 .Bl -enum -width indent
1565 followed by the number of significant bits.
1566 For example, an address with 33 significant bits could be specified as:
1568 .Dl "MAC 10:20:30:40:50:60/33 any"
1573 followed by a bitmask specified as six groups of hex digits separated
1575 For example, an address in which the last 16 bits are significant could
1578 .Dl "MAC 10:20:30:40:50:60&00:00:00:00:ff:ff any"
1580 Note that the ampersand character has a special meaning in many shells
1581 and should generally be escaped.
1584 Note that the order of MAC addresses (destination first,
1586 the same as on the wire, but the opposite of the one used for
1588 .It Cm mac-type Ar mac-type
1589 Matches packets whose Ethernet Type field
1590 corresponds to one of those specified as argument.
1592 is specified in the same way as
1594 (i.e., one or more comma-separated single values or ranges).
1595 You can use symbolic names for known values such as
1596 .Em vlan , ipv4, ipv6 .
1597 Values can be entered as decimal or hexadecimal (if prefixed by 0x),
1598 and they are always printed as hexadecimal (unless the
1600 option is used, in which case symbolic resolution will be attempted).
1601 .It Cm proto Ar protocol
1602 Matches packets with the corresponding IP protocol.
1603 .It Cm recv | xmit | via Brq Ar ifX | Ar if Ns Cm * | Ar table Ns Pq Ar number Ns Op , Ns Ar value | Ar ipno | Ar any
1604 Matches packets received, transmitted or going through,
1605 respectively, the interface specified by exact name
1606 .Ns No ( Ar ifX Ns No ),
1608 .Ns No ( Ar if Ns Ar * Ns No ),
1609 by IP address, or through some interface.
1613 keyword causes the interface to always be checked.
1620 then only the receive or transmit interface (respectively)
1622 By specifying both, it is possible to match packets based on
1623 both receive and transmit interface, e.g.:
1625 .Dl "ipfw add deny ip from any to any out recv ed0 xmit ed1"
1629 interface can be tested on either incoming or outgoing packets,
1632 interface can only be tested on outgoing packets.
1637 is invalid) whenever
1641 A packet might not have a receive or transmit interface: packets
1642 originating from the local host have no receive interface,
1643 while packets destined for the local host have no transmit
1646 Matches TCP packets that have the SYN bit set but no ACK bit.
1647 This is the short form of
1648 .Dq Li tcpflags\ syn,!ack .
1650 Matches packets that are associated to a local socket and
1651 for which the SO_USER_COOKIE socket option has been set
1652 to a non-zero value. As a side effect, the value of the
1653 option is made available as
1655 value, which in turn can be used as
1660 .It Cm src-ip Ar ip-address
1661 Matches IPv4 packets whose source IP is one of the address(es)
1662 specified as an argument.
1663 .It Cm src-ip6 Ar ip6-address
1664 Matches IPv6 packets whose source IP is one of the address(es)
1665 specified as an argument.
1666 .It Cm src-port Ar ports
1667 Matches IP packets whose source port is one of the port(s)
1668 specified as argument.
1669 .It Cm tagged Ar tag-list
1670 Matches packets whose tags are included in
1672 which is either a single value or a list of values or ranges
1673 specified in the same way as
1675 Tags can be applied to the packet using
1677 rule action parameter (see it's description for details on tags).
1678 .It Cm tcpack Ar ack
1680 Match if the TCP header acknowledgment number field is set to
1682 .It Cm tcpdatalen Ar tcpdatalen-list
1683 Matches TCP packets whose length of TCP data is
1684 .Ar tcpdatalen-list ,
1685 which is either a single value or a list of values or ranges
1686 specified in the same way as
1688 .It Cm tcpflags Ar spec
1690 Match if the TCP header contains the comma separated list of
1693 The supported TCP flags are:
1702 The absence of a particular flag may be denoted
1705 A rule which contains a
1707 specification can never match a fragmented packet which has
1711 option for details on matching fragmented packets.
1712 .It Cm tcpseq Ar seq
1714 Match if the TCP header sequence number field is set to
1716 .It Cm tcpwin Ar tcpwin-list
1717 Matches TCP packets whose header window field is set to
1719 which is either a single value or a list of values or ranges
1720 specified in the same way as
1722 .It Cm tcpoptions Ar spec
1724 Match if the TCP header contains the comma separated list of
1725 options specified in
1727 The supported TCP options are:
1730 (maximum segment size),
1732 (tcp window advertisement),
1736 (rfc1323 timestamp) and
1738 (rfc1644 t/tcp connection count).
1739 The absence of a particular option may be denoted
1743 Match all TCP or UDP packets sent by or received for a
1747 may be matched by name or identification number.
1749 For incoming packets,
1750 a routing table lookup is done on the packet's source address.
1751 If the interface on which the packet entered the system matches the
1752 outgoing interface for the route,
1754 If the interfaces do not match up,
1755 the packet does not match.
1756 All outgoing packets or packets with no incoming interface match.
1758 The name and functionality of the option is intentionally similar to
1759 the Cisco IOS command:
1761 .Dl ip verify unicast reverse-path
1763 This option can be used to make anti-spoofing rules to reject all
1764 packets with source addresses not from this interface.
1768 For incoming packets,
1769 a routing table lookup is done on the packet's source address.
1770 If a route to the source address exists, but not the default route
1771 or a blackhole/reject route, the packet matches.
1772 Otherwise, the packet does not match.
1773 All outgoing packets match.
1775 The name and functionality of the option is intentionally similar to
1776 the Cisco IOS command:
1778 .Dl ip verify unicast source reachable-via any
1780 This option can be used to make anti-spoofing rules to reject all
1781 packets whose source address is unreachable.
1783 For incoming packets, the packet's source address is checked if it
1784 belongs to a directly connected network.
1785 If the network is directly connected, then the interface the packet
1786 came on in is compared to the interface the network is connected to.
1787 When incoming interface and directly connected interface are not the
1788 same, the packet does not match.
1789 Otherwise, the packet does match.
1790 All outgoing packets match.
1792 This option can be used to make anti-spoofing rules to reject all
1793 packets that pretend to be from a directly connected network but do
1794 not come in through that interface.
1795 This option is similar to but more restricted than
1797 because it engages only on packets with source addresses of directly
1798 connected networks instead of all source addresses.
1801 Lookup tables are useful to handle large sparse sets of
1802 addresses or other search keys (e.g. ports, jail IDs, interface names).
1803 In the rest of this section we will use the term ``address''.
1804 There may be up to 4096 different lookup tables, numbered 0 to 4095.
1806 Each entry is represented by an
1807 .Ar addr Ns Op / Ns Ar masklen
1808 and will match all addresses with base
1810 (specified as an IPv4/IPv6 address, a hostname or an unsigned integer)
1816 is not specified, it defaults to 32 for IPv4 and 128 for IPv6.
1817 When looking up an IP address in a table, the most specific
1819 Associated with each entry is a 32-bit unsigned
1821 which can optionally be checked by a rule matching code.
1822 When adding an entry, if
1824 is not specified, it defaults to 0.
1826 An entry can be added to a table
1828 or removed from a table
1830 A table can be examined
1835 Internally, each table is stored in a Radix tree, the same way as
1836 the routing table (see
1839 Lookup tables currently support only ports, jail IDs, IPv4/IPv6 addresses
1840 and interface names. Wildcards is not supported for interface names.
1844 feature provides the ability to use a value, looked up in the table, as
1845 the argument for a rule action, action parameter or rule option.
1846 This can significantly reduce number of rules in some configurations.
1847 If two tables are used in a rule, the result of the second (destination)
1851 argument can be used with the following actions:
1852 .Cm nat, pipe , queue, divert, tee, netgraph, ngtee, fwd, skipto, setfib,
1860 it is possible to supply table entries with values
1861 that are in the form of IP addresses or hostnames.
1864 Section for example usage of tables and the tablearg keyword.
1868 action, the user should be aware that the code will walk the ruleset
1869 up to a rule equal to, or past, the given number, and should therefore try keep the
1870 ruleset compact between the skipto and the target rules.
1872 Each rule belongs to one of 32 different
1875 Set 31 is reserved for the default rule.
1877 By default, rules are put in set 0, unless you use the
1879 attribute when entering a new rule.
1880 Sets can be individually and atomically enabled or disabled,
1881 so this mechanism permits an easy way to store multiple configurations
1882 of the firewall and quickly (and atomically) switch between them.
1883 The command to enable/disable sets is
1884 .Bd -ragged -offset indent
1886 .Cm set Oo Cm disable Ar number ... Oc Op Cm enable Ar number ...
1893 sections can be specified.
1894 Command execution is atomic on all the sets specified in the command.
1895 By default, all sets are enabled.
1897 When you disable a set, its rules behave as if they do not exist
1898 in the firewall configuration, with only one exception:
1899 .Bd -ragged -offset indent
1900 dynamic rules created from a rule before it had been disabled
1901 will still be active until they expire.
1903 dynamic rules you have to explicitly delete the parent rule
1904 which generated them.
1907 The set number of rules can be changed with the command
1908 .Bd -ragged -offset indent
1911 .Brq Cm rule Ar rule-number | old-set
1915 Also, you can atomically swap two rulesets with the command
1916 .Bd -ragged -offset indent
1918 .Cm set swap Ar first-set second-set
1923 Section on some possible uses of sets of rules.
1924 .Sh STATEFUL FIREWALL
1925 Stateful operation is a way for the firewall to dynamically
1926 create rules for specific flows when packets that
1927 match a given pattern are detected.
1928 Support for stateful
1929 operation comes through the
1930 .Cm check-state , keep-state
1936 Dynamic rules are created when a packet matches a
1940 rule, causing the creation of a
1942 rule which will match all and only packets with
1946 .Em src-ip/src-port dst-ip/dst-port
1951 are used here only to denote the initial match addresses, but they
1952 are completely equivalent afterwards).
1953 Dynamic rules will be checked at the first
1954 .Cm check-state, keep-state
1957 occurrence, and the action performed upon a match will be the same
1958 as in the parent rule.
1960 Note that no additional attributes other than protocol and IP addresses
1961 and ports are checked on dynamic rules.
1963 The typical use of dynamic rules is to keep a closed firewall configuration,
1964 but let the first TCP SYN packet from the inside network install a
1965 dynamic rule for the flow so that packets belonging to that session
1966 will be allowed through the firewall:
1968 .Dl "ipfw add check-state"
1969 .Dl "ipfw add allow tcp from my-subnet to any setup keep-state"
1970 .Dl "ipfw add deny tcp from any to any"
1972 A similar approach can be used for UDP, where an UDP packet coming
1973 from the inside will install a dynamic rule to let the response through
1976 .Dl "ipfw add check-state"
1977 .Dl "ipfw add allow udp from my-subnet to any keep-state"
1978 .Dl "ipfw add deny udp from any to any"
1980 Dynamic rules expire after some time, which depends on the status
1981 of the flow and the setting of some
1985 .Sx SYSCTL VARIABLES
1987 For TCP sessions, dynamic rules can be instructed to periodically
1988 send keepalive packets to refresh the state of the rule when it is
1993 for more examples on how to use dynamic rules.
1994 .Sh TRAFFIC SHAPER (DUMMYNET) CONFIGURATION
1996 is also the user interface for the
1998 traffic shaper, packet scheduler and network emulator, a subsystem that
1999 can artificially queue, delay or drop packets
2000 emulating the behaviour of certain network links
2001 or queueing systems.
2004 operates by first using the firewall to select packets
2005 using any match pattern that can be used in
2008 Matching packets are then passed to either of two
2009 different objects, which implement the traffic regulation:
2010 .Bl -hang -offset XXXX
2016 with given bandwidth and propagation delay,
2017 driven by a FIFO scheduler and a single queue with programmable
2018 queue size and packet loss rate.
2019 Packets are appended to the queue as they come out from
2021 and then transferred in FIFO order to the link at the desired rate.
2025 is an abstraction used to implement packet scheduling
2026 using one of several packet scheduling algorithms.
2029 are first grouped into flows according to a mask on the 5-tuple.
2030 Flows are then passed to the scheduler associated to the
2032 and each flow uses scheduling parameters (weight and others)
2033 as configured in the
2036 A scheduler in turn is connected to an emulated link,
2037 and arbitrates the link's bandwidth among backlogged flows according to
2038 weights and to the features of the scheduling algorithm in use.
2043 can be used to set hard limits to the bandwidth that a flow can use, whereas
2045 can be used to determine how different flows share the available bandwidth.
2047 A graphical representation of the binding of queues,
2048 flows, schedulers and links is below.
2049 .Bd -literal -offset indent
2050 (flow_mask|sched_mask) sched_mask
2051 +---------+ weight Wx +-------------+
2052 | |->-[flow]-->--| |-+
2053 -->--| QUEUE x | ... | | |
2054 | |->-[flow]-->--| SCHEDuler N | |
2056 ... | +--[LINK N]-->--
2057 +---------+ weight Wy | | +--[LINK N]-->--
2058 | |->-[flow]-->--| | |
2059 -->--| QUEUE y | ... | | |
2060 | |->-[flow]-->--| | |
2061 +---------+ +-------------+ |
2064 It is important to understand the role of the SCHED_MASK
2065 and FLOW_MASK, which are configured through the commands
2066 .Dl "ipfw sched N config mask SCHED_MASK ..."
2068 .Dl "ipfw queue X config mask FLOW_MASK ..." .
2070 The SCHED_MASK is used to assign flows to one or more
2071 scheduler instances, one for each
2072 value of the packet's 5-tuple after applying SCHED_MASK.
2073 As an example, using ``src-ip 0xffffff00'' creates one instance
2074 for each /24 destination subnet.
2076 The FLOW_MASK, together with the SCHED_MASK, is used to split
2077 packets into flows. As an example, using
2078 ``src-ip 0x000000ff''
2079 together with the previous SCHED_MASK makes a flow for
2080 each individual source address. In turn, flows for each /24
2081 subnet will be sent to the same scheduler instance.
2083 The above diagram holds even for the
2085 case, with the only restriction that a
2087 only supports a SCHED_MASK, and forces the use of a FIFO
2088 scheduler (these are for backward compatibility reasons;
2089 in fact, internally, a
2091 pipe is implemented exactly as above).
2093 There are two modes of
2101 mode tries to emulate a real link: the
2103 scheduler ensures that the packet will not leave the pipe faster than it
2104 would on the real link with a given bandwidth.
2107 mode allows certain packets to bypass the
2109 scheduler (if packet flow does not exceed pipe's bandwidth).
2110 This is the reason why the
2112 mode requires less CPU cycles per packet (on average) and packet latency
2113 can be significantly lower in comparison to a real link with the same
2119 mode can be enabled by setting the
2120 .Va net.inet.ip.dummynet.io_fast
2122 variable to a non-zero value.
2124 .Ss PIPE, QUEUE AND SCHEDULER CONFIGURATION
2130 configuration commands are the following:
2131 .Bd -ragged -offset indent
2132 .Cm pipe Ar number Cm config Ar pipe-configuration
2134 .Cm queue Ar number Cm config Ar queue-configuration
2136 .Cm sched Ar number Cm config Ar sched-configuration
2139 The following parameters can be configured for a pipe:
2141 .Bl -tag -width indent -compact
2142 .It Cm bw Ar bandwidth | device
2143 Bandwidth, measured in
2146 .Brq Cm bit/s | Byte/s .
2149 A value of 0 (default) means unlimited bandwidth.
2150 The unit must immediately follow the number, as in
2152 .Dl "ipfw pipe 1 config bw 300Kbit/s"
2154 If a device name is specified instead of a numeric value, as in
2156 .Dl "ipfw pipe 1 config bw tun0"
2158 then the transmit clock is supplied by the specified device.
2159 At the moment only the
2161 device supports this
2162 functionality, for use in conjunction with
2165 .It Cm delay Ar ms-delay
2166 Propagation delay, measured in milliseconds.
2167 The value is rounded to the next multiple of the clock tick
2168 (typically 10ms, but it is a good practice to run kernels
2170 .Dq "options HZ=1000"
2172 the granularity to 1ms or less).
2173 The default value is 0, meaning no delay.
2175 .It Cm burst Ar size
2176 If the data to be sent exceeds the pipe's bandwidth limit
2177 (and the pipe was previously idle), up to
2179 bytes of data are allowed to bypass the
2181 scheduler, and will be sent as fast as the physical link allows.
2182 Any additional data will be transmitted at the rate specified
2186 The burst size depends on how long the pipe has been idle;
2187 the effective burst size is calculated as follows:
2194 .It Cm profile Ar filename
2195 A file specifying the additional overhead incurred in the transmission
2196 of a packet on the link.
2198 Some link types introduce extra delays in the transmission
2199 of a packet, e.g. because of MAC level framing, contention on
2200 the use of the channel, MAC level retransmissions and so on.
2201 From our point of view, the channel is effectively unavailable
2202 for this extra time, which is constant or variable depending
2203 on the link type. Additionally, packets may be dropped after this
2204 time (e.g. on a wireless link after too many retransmissions).
2205 We can model the additional delay with an empirical curve
2206 that represents its distribution.
2207 .Bd -literal -offset indent
2208 cumulative probability
2218 +-------*------------------->
2221 The empirical curve may have both vertical and horizontal lines.
2222 Vertical lines represent constant delay for a range of
2224 Horizontal lines correspond to a discontinuity in the delay
2225 distribution: the pipe will use the largest delay for a
2228 The file format is the following, with whitespace acting as
2229 a separator and '#' indicating the beginning a comment:
2230 .Bl -tag -width indent
2231 .It Cm name Ar identifier
2232 optional name (listed by "ipfw pipe show")
2233 to identify the delay distribution;
2235 the bandwidth used for the pipe.
2236 If not specified here, it must be present
2237 explicitly as a configuration parameter for the pipe;
2238 .It Cm loss-level Ar L
2239 the probability above which packets are lost.
2240 (0.0 <= L <= 1.0, default 1.0 i.e. no loss);
2242 the number of samples used in the internal
2243 representation of the curve (2..1024; default 100);
2244 .It Cm "delay prob" | "prob delay"
2245 One of these two lines is mandatory and defines
2246 the format of the following lines with data points.
2248 2 or more lines representing points in the curve,
2249 with either delay or probability first, according
2250 to the chosen format.
2251 The unit for delay is milliseconds.
2252 Data points do not need to be sorted.
2253 Also, the number of actual lines can be different
2254 from the value of the "samples" parameter:
2256 utility will sort and interpolate
2257 the curve as needed.
2260 Example of a profile file:
2261 .Bd -literal -offset indent
2266 0 200 # minimum overhead is 200ms
2272 #configuration file end
2276 The following parameters can be configured for a queue:
2278 .Bl -tag -width indent -compact
2279 .It Cm pipe Ar pipe_nr
2280 Connects a queue to the specified pipe.
2281 Multiple queues (with the same or different weights) can be connected to
2282 the same pipe, which specifies the aggregate rate for the set of queues.
2284 .It Cm weight Ar weight
2285 Specifies the weight to be used for flows matching this queue.
2286 The weight must be in the range 1..100, and defaults to 1.
2289 The following case-insensitive parameters can be configured for a
2292 .Bl -tag -width indent -compact
2293 .It Cm type Ar {fifo | wf2q+ | rr | qfq}
2294 specifies the scheduling algorithm to use.
2295 .Bl -tag -width indent -compact
2297 is just a FIFO scheduler (which means that all packets
2298 are stored in the same queue as they arrive to the scheduler).
2299 FIFO has O(1) per-packet time complexity, with very low
2300 constants (estimate 60-80ns on a 2GHz desktop machine)
2301 but gives no service guarantees.
2303 implements the WF2Q+ algorithm, which is a Weighted Fair Queueing
2304 algorithm which permits flows to share bandwidth according to
2305 their weights. Note that weights are not priorities; even a flow
2306 with a minuscule weight will never starve.
2307 WF2Q+ has O(log N) per-packet processing cost, where N is the number
2308 of flows, and is the default algorithm used by previous versions
2311 implements the Deficit Round Robin algorithm, which has O(1) processing
2312 costs (roughly, 100-150ns per packet)
2313 and permits bandwidth allocation according to weights, but
2314 with poor service guarantees.
2316 implements the QFQ algorithm, which is a very fast variant of
2317 WF2Q+, with similar service guarantees and O(1) processing
2318 costs (roughly, 200-250ns per packet).
2322 In addition to the type, all parameters allowed for a pipe can also
2323 be specified for a scheduler.
2325 Finally, the following parameters can be configured for both
2328 .Bl -tag -width XXXX -compact
2329 .It Cm buckets Ar hash-table-size
2330 Specifies the size of the hash table used for storing the
2332 Default value is 64 controlled by the
2335 .Va net.inet.ip.dummynet.hash_size ,
2336 allowed range is 16 to 65536.
2338 .It Cm mask Ar mask-specifier
2339 Packets sent to a given pipe or queue by an
2341 rule can be further classified into multiple flows, each of which is then
2345 A flow identifier is constructed by masking the IP addresses,
2346 ports and protocol types as specified with the
2348 options in the configuration of the pipe or queue.
2349 For each different flow identifier, a new pipe or queue is created
2350 with the same parameters as the original object, and matching packets
2355 are used, each flow will get the same bandwidth as defined by the pipe,
2358 are used, each flow will share the parent's pipe bandwidth evenly
2359 with other flows generated by the same queue (note that other queues
2360 with different weights might be connected to the same pipe).
2362 Available mask specifiers are a combination of one or more of the following:
2364 .Cm dst-ip Ar mask ,
2365 .Cm dst-ip6 Ar mask ,
2366 .Cm src-ip Ar mask ,
2367 .Cm src-ip6 Ar mask ,
2368 .Cm dst-port Ar mask ,
2369 .Cm src-port Ar mask ,
2370 .Cm flow-id Ar mask ,
2375 where the latter means all bits in all fields are significant.
2378 When a packet is dropped by a
2380 queue or pipe, the error
2381 is normally reported to the caller routine in the kernel, in the
2382 same way as it happens when a device queue fills up.
2384 option reports the packet as successfully delivered, which can be
2385 needed for some experimental setups where you want to simulate
2386 loss or congestion at a remote router.
2388 .It Cm plr Ar packet-loss-rate
2391 .Ar packet-loss-rate
2392 is a floating-point number between 0 and 1, with 0 meaning no
2393 loss, 1 meaning 100% loss.
2394 The loss rate is internally represented on 31 bits.
2396 .It Cm queue Brq Ar slots | size Ns Cm Kbytes
2401 Default value is 50 slots, which
2402 is the typical queue size for Ethernet devices.
2403 Note that for slow speed links you should keep the queue
2404 size short or your traffic might be affected by a significant
2406 E.g., 50 max-sized ethernet packets (1500 bytes) mean 600Kbit
2407 or 20s of queue on a 30Kbit/s pipe.
2408 Even worse effects can result if you get packets from an
2409 interface with a much larger MTU, e.g.\& the loopback interface
2410 with its 16KB packets.
2414 .Em net.inet.ip.dummynet.pipe_byte_limit
2416 .Em net.inet.ip.dummynet.pipe_slot_limit
2417 control the maximum lengths that can be specified.
2419 .It Cm red | gred Ar w_q Ns / Ns Ar min_th Ns / Ns Ar max_th Ns / Ns Ar max_p
2420 Make use of the RED (Random Early Detection) queue management algorithm.
2425 point numbers between 0 and 1 (0 not included), while
2429 are integer numbers specifying thresholds for queue management
2430 (thresholds are computed in bytes if the queue has been defined
2431 in bytes, in slots otherwise).
2434 also supports the gentle RED variant (gred).
2437 variables can be used to control the RED behaviour:
2438 .Bl -tag -width indent
2439 .It Va net.inet.ip.dummynet.red_lookup_depth
2440 specifies the accuracy in computing the average queue
2441 when the link is idle (defaults to 256, must be greater than zero)
2442 .It Va net.inet.ip.dummynet.red_avg_pkt_size
2443 specifies the expected average packet size (defaults to 512, must be
2445 .It Va net.inet.ip.dummynet.red_max_pkt_size
2446 specifies the expected maximum packet size, only used when queue
2447 thresholds are in bytes (defaults to 1500, must be greater than zero).
2451 When used with IPv6 data,
2453 currently has several limitations.
2454 Information necessary to route link-local packets to an
2455 interface is not available after processing by
2457 so those packets are dropped in the output path.
2458 Care should be taken to ensure that link-local packets are not passed to
2461 Here are some important points to consider when designing your
2465 Remember that you filter both packets going
2469 Most connections need packets going in both directions.
2471 Remember to test very carefully.
2472 It is a good idea to be near the console when doing this.
2473 If you cannot be near the console,
2474 use an auto-recovery script such as the one in
2475 .Pa /usr/share/examples/ipfw/change_rules.sh .
2477 Do not forget the loopback interface.
2482 There are circumstances where fragmented datagrams are unconditionally
2484 TCP packets are dropped if they do not contain at least 20 bytes of
2485 TCP header, UDP packets are dropped if they do not contain a full 8
2486 byte UDP header, and ICMP packets are dropped if they do not contain
2487 4 bytes of ICMP header, enough to specify the ICMP type, code, and
2489 These packets are simply logged as
2491 since there may not be enough good data in the packet to produce a
2492 meaningful log entry.
2494 Another type of packet is unconditionally dropped, a TCP packet with a
2495 fragment offset of one.
2496 This is a valid packet, but it only has one use, to try
2497 to circumvent firewalls.
2498 When logging is enabled, these packets are
2499 reported as being dropped by rule -1.
2501 If you are logged in over a network, loading the
2505 is probably not as straightforward as you would think.
2506 The following command line is recommended:
2507 .Bd -literal -offset indent
2509 ipfw add 32000 allow ip from any to any
2512 Along the same lines, doing an
2513 .Bd -literal -offset indent
2517 in similar surroundings is also a bad idea.
2521 filter list may not be modified if the system security level
2522 is set to 3 or higher
2525 for information on system security levels).
2527 .Sh PACKET DIVERSION
2530 socket bound to the specified port will receive all packets
2531 diverted to that port.
2532 If no socket is bound to the destination port, or if the divert module is
2533 not loaded, or if the kernel was not compiled with divert socket support,
2534 the packets are dropped.
2535 .Sh NETWORK ADDRESS TRANSLATION (NAT)
2538 support in-kernel NAT using the kernel version of
2541 The nat configuration command is the following:
2542 .Bd -ragged -offset indent
2547 .Ar nat-configuration
2551 The following parameters can be configured:
2552 .Bl -tag -width indent
2553 .It Cm ip Ar ip_address
2554 Define an ip address to use for aliasing.
2556 Use ip address of NIC for aliasing, dynamically changing
2557 it if NIC's ip address changes.
2559 Enable logging on this nat instance.
2561 Deny any incoming connection from outside world.
2563 Try to leave the alias port numbers unchanged from
2564 the actual local port numbers.
2566 Traffic on the local network not originating from an
2567 unregistered address spaces will be ignored.
2569 Reset table of the packet aliasing engine on address change.
2571 Reverse the way libalias handles aliasing.
2573 Obey transparent proxy rules only, packet aliasing is not performed.
2575 Skip instance in case of global state lookup (see below).
2578 Some specials value can be supplied instead of
2580 .Bl -tag -width indent
2582 Looks up translation state in all configured nat instances.
2583 If an entry is found, packet is aliased according to that entry.
2584 If no entry was found in any of the instances, packet is passed unchanged,
2585 and no new entry will be created.
2587 .Sx MULTIPLE INSTANCES
2590 for more information.
2592 Uses argument supplied in lookup table. See
2594 section below for more information on lookup tables.
2597 To let the packet continue after being (de)aliased, set the sysctl variable
2598 .Va net.inet.ip.fw.one_pass
2600 For more information about aliasing modes, refer to
2604 for some examples about nat usage.
2605 .Ss REDIRECT AND LSNAT SUPPORT IN IPFW
2606 Redirect and LSNAT support follow closely the syntax used in
2610 for some examples on how to do redirect and lsnat.
2611 .Ss SCTP NAT SUPPORT
2612 SCTP nat can be configured in a similar manner to TCP through the
2615 The main difference is that
2617 does not do port translation.
2618 Since the local and global side ports will be the same,
2619 there is no need to specify both.
2620 Ports are redirected as follows:
2621 .Bd -ragged -offset indent
2627 .Cm redirect_port sctp
2628 .Ar ip_address [,addr_list] {[port | port-port] [,ports]}
2634 configuration can be done in real-time through the
2637 All may be changed dynamically, though the hash_table size will only
2642 .Sx SYSCTL VARIABLES
2645 Tunables can be set in
2651 before ipfw module gets loaded.
2652 .Bl -tag -width indent
2653 .It Va net.inet.ip.fw.default_to_accept: No 0
2654 Defines ipfw last rule behavior. This value overrides
2655 .Cd "options IPFW_DEFAULT_TO_(ACCEPT|DENY)"
2656 from kernel configuration file.
2657 .It Va net.inet.ip.fw.tables_max: No 128
2658 Defines number of tables available in ipfw. Number cannot exceed 65534.
2660 .Sh SYSCTL VARIABLES
2663 variables controls the behaviour of the firewall and
2665 .Pq Nm dummynet , bridge , sctp nat .
2666 These are shown below together with their default value
2667 (but always check with the
2669 command what value is actually in use) and meaning:
2670 .Bl -tag -width indent
2671 .It Va net.inet.ip.alias.sctp.accept_global_ootb_addip: No 0
2674 responds to receipt of global OOTB ASCONF-AddIP:
2675 .Bl -tag -width indent
2677 No response (unless a partially matching association exists -
2678 ports and vtags match but global address does not)
2681 will accept and process all OOTB global AddIP messages.
2684 Option 1 should never be selected as this forms a security risk.
2686 establish multiple fake associations by sending AddIP messages.
2687 .It Va net.inet.ip.alias.sctp.chunk_proc_limit: No 5
2688 Defines the maximum number of chunks in an SCTP packet that will be parsed for a
2689 packet that matches an existing association.
2690 This value is enforced to be greater or equal than
2691 .Cm net.inet.ip.alias.sctp.initialising_chunk_proc_limit .
2693 a DoS risk yet setting too low a value may result in important control chunks in
2694 the packet not being located and parsed.
2695 .It Va net.inet.ip.alias.sctp.error_on_ootb: No 1
2698 responds to any Out-of-the-Blue (OOTB) packets with ErrorM packets.
2699 An OOTB packet is a packet that arrives with no existing association
2702 and is not an INIT or ASCONF-AddIP packet:
2703 .Bl -tag -width indent
2705 ErrorM is never sent in response to OOTB packets.
2707 ErrorM is only sent to OOTB packets received on the local side.
2709 ErrorM is sent to the local side and on the global side ONLY if there is a
2710 partial match (ports and vtags match but the source global IP does not).
2711 This value is only useful if the
2713 is tracking global IP addresses.
2715 ErrorM is sent in response to all OOTB packets on both the local and global side
2719 At the moment the default is 0, since the ErrorM packet is not yet
2720 supported by most SCTP stacks.
2721 When it is supported, and if not tracking
2722 global addresses, we recommend setting this value to 1 to allow
2723 multi-homed local hosts to function with the
2725 To track global addresses, we recommend setting this value to 2 to
2726 allow global hosts to be informed when they need to (re)send an
2728 Value 3 should never be chosen (except for debugging) as the
2730 will respond to all OOTB global packets (a DoS risk).
2731 .It Va net.inet.ip.alias.sctp.hashtable_size: No 2003
2732 Size of hash tables used for
2734 lookups (100 < prime_number > 1000001).
2737 size for any future created
2739 instance and therefore must be set prior to creating a
2742 The table sizes may be changed to suit specific needs.
2743 If there will be few
2744 concurrent associations, and memory is scarce, you may make these smaller.
2745 If there will be many thousands (or millions) of concurrent associations, you
2746 should make these larger.
2747 A prime number is best for the table size.
2749 update function will adjust your input value to the next highest prime number.
2750 .It Va net.inet.ip.alias.sctp.holddown_time: No 0
2751 Hold association in table for this many seconds after receiving a
2753 This allows endpoints to correct shutdown gracefully if a
2754 shutdown_complete is lost and retransmissions are required.
2755 .It Va net.inet.ip.alias.sctp.init_timer: No 15
2756 Timeout value while waiting for (INIT-ACK|AddIP-ACK).
2757 This value cannot be 0.
2758 .It Va net.inet.ip.alias.sctp.initialising_chunk_proc_limit: No 2
2759 Defines the maximum number of chunks in an SCTP packet that will be parsed when
2760 no existing association exists that matches that packet.
2762 will only be an INIT or ASCONF-AddIP packet.
2763 A higher value may become a DoS
2764 risk as malformed packets can consume processing resources.
2765 .It Va net.inet.ip.alias.sctp.param_proc_limit: No 25
2766 Defines the maximum number of parameters within a chunk that will be parsed in a
2768 As for other similar sysctl variables, larger values pose a DoS risk.
2769 .It Va net.inet.ip.alias.sctp.log_level: No 0
2770 Level of detail in the system log messages (0 \- minimal, 1 \- event,
2771 2 \- info, 3 \- detail, 4 \- debug, 5 \- max debug). May be a good
2772 option in high loss environments.
2773 .It Va net.inet.ip.alias.sctp.shutdown_time: No 15
2774 Timeout value while waiting for SHUTDOWN-COMPLETE.
2775 This value cannot be 0.
2776 .It Va net.inet.ip.alias.sctp.track_global_addresses: No 0
2777 Enables/disables global IP address tracking within the
2780 upper limit on the number of addresses tracked for each association:
2781 .Bl -tag -width indent
2783 Global tracking is disabled
2785 Enables tracking, the maximum number of addresses tracked for each
2786 association is limited to this value
2789 This variable is fully dynamic, the new value will be adopted for all newly
2790 arriving associations, existing associations are treated as they were previously.
2791 Global tracking will decrease the number of collisions within the
2794 of increased processing load, memory usage, complexity, and possible
2797 problems in complex networks with multiple
2799 We recommend not tracking
2800 global IP addresses, this will still result in a fully functional
2802 .It Va net.inet.ip.alias.sctp.up_timer: No 300
2803 Timeout value to keep an association up with no traffic.
2804 This value cannot be 0.
2805 .It Va net.inet.ip.dummynet.expire : No 1
2806 Lazily delete dynamic pipes/queue once they have no pending traffic.
2807 You can disable this by setting the variable to 0, in which case
2808 the pipes/queues will only be deleted when the threshold is reached.
2809 .It Va net.inet.ip.dummynet.hash_size : No 64
2810 Default size of the hash table used for dynamic pipes/queues.
2811 This value is used when no
2813 option is specified when configuring a pipe/queue.
2814 .It Va net.inet.ip.dummynet.io_fast : No 0
2815 If set to a non-zero value,
2820 operation (see above) is enabled.
2821 .It Va net.inet.ip.dummynet.io_pkt
2822 Number of packets passed to
2824 .It Va net.inet.ip.dummynet.io_pkt_drop
2825 Number of packets dropped by
2827 .It Va net.inet.ip.dummynet.io_pkt_fast
2828 Number of packets bypassed by the
2831 .It Va net.inet.ip.dummynet.max_chain_len : No 16
2832 Target value for the maximum number of pipes/queues in a hash bucket.
2834 .Cm max_chain_len*hash_size
2835 is used to determine the threshold over which empty pipes/queues
2836 will be expired even when
2837 .Cm net.inet.ip.dummynet.expire=0 .
2838 .It Va net.inet.ip.dummynet.red_lookup_depth : No 256
2839 .It Va net.inet.ip.dummynet.red_avg_pkt_size : No 512
2840 .It Va net.inet.ip.dummynet.red_max_pkt_size : No 1500
2841 Parameters used in the computations of the drop probability
2842 for the RED algorithm.
2843 .It Va net.inet.ip.dummynet.pipe_byte_limit : No 1048576
2844 .It Va net.inet.ip.dummynet.pipe_slot_limit : No 100
2845 The maximum queue size that can be specified in bytes or packets.
2846 These limits prevent accidental exhaustion of resources such as mbufs.
2847 If you raise these limits,
2848 you should make sure the system is configured so that sufficient resources
2850 .It Va net.inet.ip.fw.autoinc_step : No 100
2851 Delta between rule numbers when auto-generating them.
2852 The value must be in the range 1..1000.
2853 .It Va net.inet.ip.fw.curr_dyn_buckets : Va net.inet.ip.fw.dyn_buckets
2854 The current number of buckets in the hash table for dynamic rules
2856 .It Va net.inet.ip.fw.debug : No 1
2857 Controls debugging messages produced by
2859 .It Va net.inet.ip.fw.default_rule : No 65535
2860 The default rule number (read-only).
2862 .Nm , the default rule is the last one, so its number
2863 can also serve as the highest number allowed for a rule.
2864 .It Va net.inet.ip.fw.dyn_buckets : No 256
2865 The number of buckets in the hash table for dynamic rules.
2866 Must be a power of 2, up to 65536.
2867 It only takes effect when all dynamic rules have expired, so you
2868 are advised to use a
2870 command to make sure that the hash table is resized.
2871 .It Va net.inet.ip.fw.dyn_count : No 3
2872 Current number of dynamic rules
2874 .It Va net.inet.ip.fw.dyn_keepalive : No 1
2875 Enables generation of keepalive packets for
2877 rules on TCP sessions.
2878 A keepalive is generated to both
2879 sides of the connection every 5 seconds for the last 20
2880 seconds of the lifetime of the rule.
2881 .It Va net.inet.ip.fw.dyn_max : No 8192
2882 Maximum number of dynamic rules.
2883 When you hit this limit, no more dynamic rules can be
2884 installed until old ones expire.
2885 .It Va net.inet.ip.fw.dyn_ack_lifetime : No 300
2886 .It Va net.inet.ip.fw.dyn_syn_lifetime : No 20
2887 .It Va net.inet.ip.fw.dyn_fin_lifetime : No 1
2888 .It Va net.inet.ip.fw.dyn_rst_lifetime : No 1
2889 .It Va net.inet.ip.fw.dyn_udp_lifetime : No 5
2890 .It Va net.inet.ip.fw.dyn_short_lifetime : No 30
2891 These variables control the lifetime, in seconds, of dynamic
2893 Upon the initial SYN exchange the lifetime is kept short,
2894 then increased after both SYN have been seen, then decreased
2895 again during the final FIN exchange or when a RST is received.
2897 .Em dyn_fin_lifetime
2899 .Em dyn_rst_lifetime
2900 must be strictly lower than 5 seconds, the period of
2901 repetition of keepalives.
2902 The firewall enforces that.
2903 .It Va net.inet.ip.fw.enable : No 1
2904 Enables the firewall.
2905 Setting this variable to 0 lets you run your machine without
2906 firewall even if compiled in.
2907 .It Va net.inet6.ip6.fw.enable : No 1
2908 provides the same functionality as above for the IPv6 case.
2909 .It Va net.inet.ip.fw.one_pass : No 1
2910 When set, the packet exiting from the
2914 node is not passed though the firewall again.
2915 Otherwise, after an action, the packet is
2916 reinjected into the firewall at the next rule.
2917 .It Va net.inet.ip.fw.tables_max : No 128
2918 Maximum number of tables.
2919 .It Va net.inet.ip.fw.verbose : No 1
2920 Enables verbose messages.
2921 .It Va net.inet.ip.fw.verbose_limit : No 0
2922 Limits the number of messages produced by a verbose firewall.
2923 .It Va net.inet6.ip6.fw.deny_unknown_exthdrs : No 1
2924 If enabled packets with unknown IPv6 Extension Headers will be denied.
2925 .It Va net.link.ether.ipfw : No 0
2926 Controls whether layer-2 packets are passed to
2929 .It Va net.link.bridge.ipfw : No 0
2930 Controls whether bridged packets are passed to
2936 There are far too many possible uses of
2938 so this Section will only give a small set of examples.
2940 .Ss BASIC PACKET FILTERING
2941 This command adds an entry which denies all tcp packets from
2942 .Em cracker.evil.org
2943 to the telnet port of
2945 from being forwarded by the host:
2947 .Dl "ipfw add deny tcp from cracker.evil.org to wolf.tambov.su telnet"
2949 This one disallows any connection from the entire cracker's
2952 .Dl "ipfw add deny ip from 123.45.67.0/24 to my.host.org"
2954 A first and efficient way to limit access (not using dynamic rules)
2955 is the use of the following rules:
2957 .Dl "ipfw add allow tcp from any to any established"
2958 .Dl "ipfw add allow tcp from net1 portlist1 to net2 portlist2 setup"
2959 .Dl "ipfw add allow tcp from net3 portlist3 to net3 portlist3 setup"
2961 .Dl "ipfw add deny tcp from any to any"
2963 The first rule will be a quick match for normal TCP packets,
2964 but it will not match the initial SYN packet, which will be
2967 rules only for selected source/destination pairs.
2968 All other SYN packets will be rejected by the final
2972 If you administer one or more subnets, you can take advantage
2973 of the address sets and or-blocks and write extremely
2974 compact rulesets which selectively enable services to blocks
2975 of clients, as below:
2977 .Dl "goodguys=\*q{ 10.1.2.0/24{20,35,66,18} or 10.2.3.0/28{6,3,11} }\*q"
2978 .Dl "badguys=\*q10.1.2.0/24{8,38,60}\*q"
2980 .Dl "ipfw add allow ip from ${goodguys} to any"
2981 .Dl "ipfw add deny ip from ${badguys} to any"
2982 .Dl "... normal policies ..."
2986 option could be used to do automated anti-spoofing by adding the
2987 following to the top of a ruleset:
2989 .Dl "ipfw add deny ip from any to any not verrevpath in"
2991 This rule drops all incoming packets that appear to be coming to the
2992 system on the wrong interface.
2993 For example, a packet with a source
2994 address belonging to a host on a protected internal network would be
2995 dropped if it tried to enter the system from an external interface.
2999 option could be used to do similar but more restricted anti-spoofing
3000 by adding the following to the top of a ruleset:
3002 .Dl "ipfw add deny ip from any to any not antispoof in"
3004 This rule drops all incoming packets that appear to be coming from another
3005 directly connected system but on the wrong interface.
3006 For example, a packet with a source address of
3007 .Li 192.168.0.0/24 ,
3016 option could be used to (re)mark user traffic,
3017 by adding the following to the appropriate place in ruleset:
3019 .Dl "ipfw add setdscp be ip from any to any dscp af11,af21"
3021 This rule drops all incoming packets that appear to be coming from another
3022 directly connected system but on the wrong interface.
3023 For example, a packet with a source address of
3024 .Li 192.168.0.0/24 ,
3031 In order to protect a site from flood attacks involving fake
3032 TCP packets, it is safer to use dynamic rules:
3034 .Dl "ipfw add check-state"
3035 .Dl "ipfw add deny tcp from any to any established"
3036 .Dl "ipfw add allow tcp from my-net to any setup keep-state"
3038 This will let the firewall install dynamic rules only for
3039 those connection which start with a regular SYN packet coming
3040 from the inside of our network.
3041 Dynamic rules are checked when encountering the first
3048 rule should usually be placed near the beginning of the
3049 ruleset to minimize the amount of work scanning the ruleset.
3050 Your mileage may vary.
3052 To limit the number of connections a user can open
3053 you can use the following type of rules:
3055 .Dl "ipfw add allow tcp from my-net/24 to any setup limit src-addr 10"
3056 .Dl "ipfw add allow tcp from any to me setup limit src-addr 4"
3058 The former (assuming it runs on a gateway) will allow each host
3059 on a /24 network to open at most 10 TCP connections.
3060 The latter can be placed on a server to make sure that a single
3061 client does not use more than 4 simultaneous connections.
3064 stateful rules can be subject to denial-of-service attacks
3065 by a SYN-flood which opens a huge number of dynamic rules.
3066 The effects of such attacks can be partially limited by
3069 variables which control the operation of the firewall.
3071 Here is a good usage of the
3073 command to see accounting records and timestamp information:
3077 or in short form without timestamps:
3081 which is equivalent to:
3085 Next rule diverts all incoming packets from 192.168.2.0/24
3086 to divert port 5000:
3088 .Dl ipfw divert 5000 ip from 192.168.2.0/24 to any in
3091 The following rules show some of the applications of
3095 for simulations and the like.
3097 This rule drops random incoming packets with a probability
3100 .Dl "ipfw add prob 0.05 deny ip from any to any in"
3102 A similar effect can be achieved making use of
3106 .Dl "ipfw add pipe 10 ip from any to any"
3107 .Dl "ipfw pipe 10 config plr 0.05"
3109 We can use pipes to artificially limit bandwidth, e.g.\& on a
3110 machine acting as a router, if we want to limit traffic from
3111 local clients on 192.168.2.0/24 we do:
3113 .Dl "ipfw add pipe 1 ip from 192.168.2.0/24 to any out"
3114 .Dl "ipfw pipe 1 config bw 300Kbit/s queue 50KBytes"
3116 note that we use the
3118 modifier so that the rule is not used twice.
3119 Remember in fact that
3121 rules are checked both on incoming and outgoing packets.
3123 Should we want to simulate a bidirectional link with bandwidth
3124 limitations, the correct way is the following:
3126 .Dl "ipfw add pipe 1 ip from any to any out"
3127 .Dl "ipfw add pipe 2 ip from any to any in"
3128 .Dl "ipfw pipe 1 config bw 64Kbit/s queue 10Kbytes"
3129 .Dl "ipfw pipe 2 config bw 64Kbit/s queue 10Kbytes"
3131 The above can be very useful, e.g.\& if you want to see how
3132 your fancy Web page will look for a residential user who
3133 is connected only through a slow link.
3134 You should not use only one pipe for both directions, unless
3135 you want to simulate a half-duplex medium (e.g.\& AppleTalk,
3137 It is not necessary that both pipes have the same configuration,
3138 so we can also simulate asymmetric links.
3140 Should we want to verify network performance with the RED queue
3141 management algorithm:
3143 .Dl "ipfw add pipe 1 ip from any to any"
3144 .Dl "ipfw pipe 1 config bw 500Kbit/s queue 100 red 0.002/30/80/0.1"
3146 Another typical application of the traffic shaper is to
3147 introduce some delay in the communication.
3148 This can significantly affect applications which do a lot of Remote
3149 Procedure Calls, and where the round-trip-time of the
3150 connection often becomes a limiting factor much more than
3153 .Dl "ipfw add pipe 1 ip from any to any out"
3154 .Dl "ipfw add pipe 2 ip from any to any in"
3155 .Dl "ipfw pipe 1 config delay 250ms bw 1Mbit/s"
3156 .Dl "ipfw pipe 2 config delay 250ms bw 1Mbit/s"
3158 Per-flow queueing can be useful for a variety of purposes.
3159 A very simple one is counting traffic:
3161 .Dl "ipfw add pipe 1 tcp from any to any"
3162 .Dl "ipfw add pipe 1 udp from any to any"
3163 .Dl "ipfw add pipe 1 ip from any to any"
3164 .Dl "ipfw pipe 1 config mask all"
3166 The above set of rules will create queues (and collect
3167 statistics) for all traffic.
3168 Because the pipes have no limitations, the only effect is
3169 collecting statistics.
3170 Note that we need 3 rules, not just the last one, because
3173 tries to match IP packets it will not consider ports, so we
3174 would not see connections on separate ports as different
3177 A more sophisticated example is limiting the outbound traffic
3178 on a net with per-host limits, rather than per-network limits:
3180 .Dl "ipfw add pipe 1 ip from 192.168.2.0/24 to any out"
3181 .Dl "ipfw add pipe 2 ip from any to 192.168.2.0/24 in"
3182 .Dl "ipfw pipe 1 config mask src-ip 0x000000ff bw 200Kbit/s queue 20Kbytes"
3183 .Dl "ipfw pipe 2 config mask dst-ip 0x000000ff bw 200Kbit/s queue 20Kbytes"
3185 In the following example, we need to create several traffic bandwidth
3186 classes and we need different hosts/networks to fall into different classes.
3187 We create one pipe for each class and configure them accordingly.
3188 Then we create a single table and fill it with IP subnets and addresses.
3189 For each subnet/host we set the argument equal to the number of the pipe
3191 Then we classify traffic using a single rule:
3193 .Dl "ipfw pipe 1 config bw 1000Kbyte/s"
3194 .Dl "ipfw pipe 4 config bw 4000Kbyte/s"
3196 .Dl "ipfw table 1 add 192.168.2.0/24 1"
3197 .Dl "ipfw table 1 add 192.168.0.0/27 4"
3198 .Dl "ipfw table 1 add 192.168.0.2 1"
3200 .Dl "ipfw add pipe tablearg ip from table(1) to any"
3204 action, the table entries may include hostnames and IP addresses.
3206 .Dl "ipfw table 1 add 192.168.2.0/24 10.23.2.1"
3207 .Dl "ipfw table 1 add 192.168.0.0/27 router1.dmz"
3209 .Dl "ipfw add 100 fwd tablearg ip from any to table(1)"
3211 In the following example per-interface firewall is created:
3213 .Dl "ipfw table 10 add vlan20 12000"
3214 .Dl "ipfw table 10 add vlan30 13000"
3215 .Dl "ipfw table 20 add vlan20 22000"
3216 .Dl "ipfw table 20 add vlan30 23000"
3218 .Dl "ipfw add 100 ipfw skipto tablearg ip from any to any recv 'table(10)' in"
3219 .Dl "ipfw add 200 ipfw skipto tablearg ip from any to any xmit 'table(10)' out"
3221 To add a set of rules atomically, e.g.\& set 18:
3223 .Dl "ipfw set disable 18"
3224 .Dl "ipfw add NN set 18 ... # repeat as needed"
3225 .Dl "ipfw set enable 18"
3227 To delete a set of rules atomically the command is simply:
3229 .Dl "ipfw delete set 18"
3231 To test a ruleset and disable it and regain control if something goes wrong:
3233 .Dl "ipfw set disable 18"
3234 .Dl "ipfw add NN set 18 ... # repeat as needed"
3235 .Dl "ipfw set enable 18; echo done; sleep 30 && ipfw set disable 18"
3237 Here if everything goes well, you press control-C before the "sleep"
3238 terminates, and your ruleset will be left active.
3239 Otherwise, e.g.\& if
3240 you cannot access your box, the ruleset will be disabled after
3241 the sleep terminates thus restoring the previous situation.
3243 To show rules of the specific set:
3245 .Dl "ipfw set 18 show"
3247 To show rules of the disabled set:
3249 .Dl "ipfw -S set 18 show"
3251 To clear a specific rule counters of the specific set:
3253 .Dl "ipfw set 18 zero NN"
3255 To delete a specific rule of the specific set:
3257 .Dl "ipfw set 18 delete NN"
3258 .Ss NAT, REDIRECT AND LSNAT
3259 First redirect all the traffic to nat instance 123:
3261 .Dl "ipfw add nat 123 all from any to any"
3263 Then to configure nat instance 123 to alias all the outgoing traffic with ip
3264 192.168.0.123, blocking all incoming connections, trying to keep
3265 same ports on both sides, clearing aliasing table on address change
3266 and keeping a log of traffic/link statistics:
3268 .Dl "ipfw nat 123 config ip 192.168.0.123 log deny_in reset same_ports"
3270 Or to change address of instance 123, aliasing table will be cleared (see
3273 .Dl "ipfw nat 123 config ip 10.0.0.1"
3275 To see configuration of nat instance 123:
3277 .Dl "ipfw nat 123 show config"
3279 To show logs of all the instances in range 111-999:
3281 .Dl "ipfw nat 111-999 show"
3283 To see configurations of all instances:
3285 .Dl "ipfw nat show config"
3287 Or a redirect rule with mixed modes could looks like:
3289 .Dl "ipfw nat 123 config redirect_addr 10.0.0.1 10.0.0.66"
3290 .Dl " redirect_port tcp 192.168.0.1:80 500"
3291 .Dl " redirect_proto udp 192.168.1.43 192.168.1.1"
3292 .Dl " redirect_addr 192.168.0.10,192.168.0.11"
3293 .Dl " 10.0.0.100 # LSNAT"
3294 .Dl " redirect_port tcp 192.168.0.1:80,192.168.0.10:22"
3297 or it could be split in:
3299 .Dl "ipfw nat 1 config redirect_addr 10.0.0.1 10.0.0.66"
3300 .Dl "ipfw nat 2 config redirect_port tcp 192.168.0.1:80 500"
3301 .Dl "ipfw nat 3 config redirect_proto udp 192.168.1.43 192.168.1.1"
3302 .Dl "ipfw nat 4 config redirect_addr 192.168.0.10,192.168.0.11,192.168.0.12"
3304 .Dl "ipfw nat 5 config redirect_port tcp"
3305 .Dl " 192.168.0.1:80,192.168.0.10:22,192.168.0.20:25 500"
3327 utility first appeared in
3332 Stateful extensions were introduced in
3335 was introduced in Summer 2002.
3337 .An Ugen J. S. Antsilevich ,
3338 .An Poul-Henning Kamp ,
3344 API based upon code written by
3348 Dummynet has been introduced by Luigi Rizzo in 1997-1998.
3350 Some early work (1999-2000) on the
3352 traffic shaper supported by Akamba Corp.
3354 The ipfw core (ipfw2) has been completely redesigned and
3355 reimplemented by Luigi Rizzo in summer 2002. Further
3357 options have been added by various developer over the years.
3360 In-kernel NAT support written by
3361 .An Paolo Pisati Aq piso@FreeBSD.org
3362 as part of a Summer of Code 2005 project.
3366 support has been developed by
3367 .An The Centre for Advanced Internet Architectures (CAIA) Aq http://www.caia.swin.edu.au .
3368 The primary developers and maintainers are David Hayes and Jason But.
3369 For further information visit:
3370 .Aq http://www.caia.swin.edu.au/urp/SONATA
3372 Delay profiles have been developed by Alessandro Cerri and
3373 Luigi Rizzo, supported by the
3374 European Commission within Projects Onelab and Onelab2.
3376 The syntax has grown over the years and sometimes it might be confusing.
3377 Unfortunately, backward compatibility prevents cleaning up mistakes
3378 made in the definition of the syntax.
3382 Misconfiguring the firewall can put your computer in an unusable state,
3383 possibly shutting down network services and requiring console access to
3384 regain control of it.
3386 Incoming packet fragments diverted by
3388 are reassembled before delivery to the socket.
3389 The action used on those packet is the one from the
3390 rule which matches the first fragment of the packet.
3392 Packets diverted to userland, and then reinserted by a userland process
3393 may lose various packet attributes.
3394 The packet source interface name
3395 will be preserved if it is shorter than 8 bytes and the userland process
3396 saves and reuses the sockaddr_in
3399 otherwise, it may be lost.
3400 If a packet is reinserted in this manner, later rules may be incorrectly
3401 applied, making the order of
3403 rules in the rule sequence very important.
3405 Dummynet drops all packets with IPv6 link-local addresses.
3411 may not behave as expected.
3412 In particular, incoming SYN packets may
3413 have no uid or gid associated with them since they do not yet belong
3414 to a TCP connection, and the uid/gid associated with a packet may not
3415 be as expected if the associated process calls
3417 or similar system calls.
3419 Rule syntax is subject to the command line environment and some patterns
3420 may need to be escaped with the backslash character
3421 or quoted appropriately.
3423 Due to the architecture of
3425 ipfw nat is not compatible with the TCP segmentation offloading (TSO).
3426 Thus, to reliably nat your network traffic, please disable TSO
3430 ICMP error messages are not implicitly matched by dynamic rules
3431 for the respective conversations.
3432 To avoid failures of network error detection and path MTU discovery,
3433 ICMP error messages may need to be allowed explicitly through static
3440 actions may lead to confusing behaviour if ruleset has mistakes,
3441 and/or interaction with other subsystems (netgraph, dummynet, etc.) is used.
3442 One possible case for this is packet leaving
3444 in subroutine on the input pass, while later on output encountering unpaired
3447 As the call stack is kept intact after input pass, packet will suddenly
3448 return to the rule number used on input pass, not on output one.
3449 Order of processing should be checked carefully to avoid such mistakes.