xref: /freebsd/sbin/ipfw/ipfw.8 (revision fee1489eb14e022d52bea01ff4399bc6d7db3015)
1.\"
2.\" $FreeBSD$
3.\"
4.Dd July 9, 2018
5.Dt IPFW 8
6.Os
7.Sh NAME
8.Nm ipfw
9.Nd User interface for firewall, traffic shaper, packet scheduler,
10in-kernel NAT.
11.Sh SYNOPSIS
12.Ss FIREWALL CONFIGURATION
13.Nm
14.Op Fl cq
15.Cm add
16.Ar rule
17.Nm
18.Op Fl acdefnNStT
19.Op Cm set Ar N
20.Brq Cm list | show
21.Op Ar rule | first-last ...
22.Nm
23.Op Fl f | q
24.Op Cm set Ar N
25.Cm flush
26.Nm
27.Op Fl q
28.Op Cm set Ar N
29.Brq Cm delete | zero | resetlog
30.Op Ar number ...
31.Pp
32.Nm
33.Cm set Oo Cm disable Ar number ... Oc Op Cm enable Ar number ...
34.Nm
35.Cm set move
36.Op Cm rule
37.Ar number Cm to Ar number
38.Nm
39.Cm set swap Ar number number
40.Nm
41.Cm set show
42.Ss SYSCTL SHORTCUTS
43.Nm
44.Cm enable
45.Brq Cm firewall | altq | one_pass | debug | verbose | dyn_keepalive
46.Nm
47.Cm disable
48.Brq Cm firewall | altq | one_pass | debug | verbose | dyn_keepalive
49.Ss LOOKUP TABLES
50.Nm
51.Oo Cm set Ar N Oc Cm table Ar name Cm create Ar create-options
52.Nm
53.Oo Cm set Ar N Oc Cm table
54.Brq Ar name | all
55.Cm destroy
56.Nm
57.Oo Cm set Ar N Oc Cm table Ar name Cm modify Ar modify-options
58.Nm
59.Oo Cm set Ar N Oc Cm table Ar name Cm swap Ar name
60.Nm
61.Oo Cm set Ar N Oc Cm table Ar name Cm add Ar table-key Op Ar value
62.Nm
63.Oo Cm set Ar N Oc Cm table Ar name Cm add Op Ar table-key Ar value ...
64.Nm
65.Oo Cm set Ar N Oc Cm table Ar name Cm atomic add Op Ar table-key Ar value ...
66.Nm
67.Oo Cm set Ar N Oc Cm table Ar name Cm delete Op Ar table-key ...
68.Nm
69.Oo Cm set Ar N Oc Cm table Ar name Cm lookup Ar addr
70.Nm
71.Oo Cm set Ar N Oc Cm table Ar name Cm lock
72.Nm
73.Oo Cm set Ar N Oc Cm table Ar name Cm unlock
74.Nm
75.Oo Cm set Ar N Oc Cm table
76.Brq Ar name | all
77.Cm list
78.Nm
79.Oo Cm set Ar N Oc Cm table
80.Brq Ar name | all
81.Cm info
82.Nm
83.Oo Cm set Ar N Oc Cm table
84.Brq Ar name | all
85.Cm detail
86.Nm
87.Oo Cm set Ar N Oc Cm table
88.Brq Ar name | all
89.Cm flush
90.Ss DUMMYNET CONFIGURATION (TRAFFIC SHAPER AND PACKET SCHEDULER)
91.Nm
92.Brq Cm pipe | queue | sched
93.Ar number
94.Cm config
95.Ar config-options
96.Nm
97.Op Fl s Op Ar field
98.Brq Cm pipe | queue | sched
99.Brq Cm delete | list | show
100.Op Ar number ...
101.Ss IN-KERNEL NAT
102.Nm
103.Op Fl q
104.Cm nat
105.Ar number
106.Cm config
107.Ar config-options
108.Pp
109.Nm
110.Op Fl cfnNqS
111.Oo
112.Fl p Ar preproc
113.Oo
114.Ar preproc-flags
115.Oc
116.Oc
117.Ar pathname
118.Ss STATEFUL IPv6/IPv4 NETWORK ADDRESS AND PROTOCOL TRANSLATION
119.Nm
120.Oo Cm set Ar N Oc Cm nat64lsn Ar name Cm create Ar create-options
121.Nm
122.Oo Cm set Ar N Oc Cm nat64lsn Ar name Cm config Ar config-options
123.Nm
124.Oo Cm set Ar N Oc Cm nat64lsn
125.Brq Ar name | all
126.Brq Cm list | show
127.Op Cm states
128.Nm
129.Oo Cm set Ar N Oc Cm nat64lsn
130.Brq Ar name | all
131.Cm destroy
132.Nm
133.Oo Cm set Ar N Oc Cm nat64lsn Ar name Cm stats Op Cm reset
134.Ss STATELESS IPv6/IPv4 NETWORK ADDRESS AND PROTOCOL TRANSLATION
135.Nm
136.Oo Cm set Ar N Oc Cm nat64stl Ar name Cm create Ar create-options
137.Nm
138.Oo Cm set Ar N Oc Cm nat64stl Ar name Cm config Ar config-options
139.Nm
140.Oo Cm set Ar N Oc Cm nat64stl
141.Brq Ar name | all
142.Brq Cm list | show
143.Nm
144.Oo Cm set Ar N Oc Cm nat64stl
145.Brq Ar name | all
146.Cm destroy
147.Nm
148.Oo Cm set Ar N Oc Cm nat64stl Ar name Cm stats Op Cm reset
149.Ss IPv6-to-IPv6 NETWORK PREFIX TRANSLATION
150.Nm
151.Oo Cm set Ar N Oc Cm nptv6 Ar name Cm create Ar create-options
152.Nm
153.Oo Cm set Ar N Oc Cm nptv6
154.Brq Ar name | all
155.Brq Cm list | show
156.Nm
157.Oo Cm set Ar N Oc Cm nptv6
158.Brq Ar name | all
159.Cm destroy
160.Nm
161.Oo Cm set Ar N Oc Cm nptv6 Ar name Cm stats Op Cm reset
162.Ss INTERNAL DIAGNOSTICS
163.Nm
164.Cm internal iflist
165.Nm
166.Cm internal talist
167.Nm
168.Cm internal vlist
169.Sh DESCRIPTION
170The
171.Nm
172utility is the user interface for controlling the
173.Xr ipfw 4
174firewall, the
175.Xr dummynet 4
176traffic shaper/packet scheduler, and the
177in-kernel NAT services.
178.Pp
179A firewall configuration, or
180.Em ruleset ,
181is made of a list of
182.Em rules
183numbered from 1 to 65535.
184Packets are passed to the firewall
185from a number of different places in the protocol stack
186(depending on the source and destination of the packet,
187it is possible for the firewall to be
188invoked multiple times on the same packet).
189The packet passed to the firewall is compared
190against each of the rules in the
191.Em ruleset ,
192in rule-number order
193(multiple rules with the same number are permitted, in which case
194they are processed in order of insertion).
195When a match is found, the action corresponding to the
196matching rule is performed.
197.Pp
198Depending on the action and certain system settings, packets
199can be reinjected into the firewall at some rule after the
200matching one for further processing.
201.Pp
202A ruleset always includes a
203.Em default
204rule (numbered 65535) which cannot be modified or deleted,
205and matches all packets.
206The action associated with the
207.Em default
208rule can be either
209.Cm deny
210or
211.Cm allow
212depending on how the kernel is configured.
213.Pp
214If the ruleset includes one or more rules with the
215.Cm keep-state ,
216.Cm record-state ,
217.Cm limit
218or
219.Cm set-limit
220option,
221the firewall will have a
222.Em stateful
223behaviour, i.e., upon a match it will create
224.Em dynamic rules ,
225i.e., rules that match packets with the same 5-tuple
226(protocol, source and destination addresses and ports)
227as the packet which caused their creation.
228Dynamic rules, which have a limited lifetime, are checked
229at the first occurrence of a
230.Cm check-state ,
231.Cm keep-state
232or
233.Cm limit
234rule, and are typically used to open the firewall on-demand to
235legitimate traffic only.
236Please, note, that
237.Cm keep-state
238amd
239.Cm limit
240imply implicit
241.Cm check-state
242for all packets (not only these matched by the rule) but
243.Cm record-state
244and
245.Cm set-limit
246have no implicit
247.Cm check-state .
248See the
249.Sx STATEFUL FIREWALL
250and
251.Sx EXAMPLES
252Sections below for more information on the stateful behaviour of
253.Nm .
254.Pp
255All rules (including dynamic ones) have a few associated counters:
256a packet count, a byte count, a log count and a timestamp
257indicating the time of the last match.
258Counters can be displayed or reset with
259.Nm
260commands.
261.Pp
262Each rule belongs to one of 32 different
263.Em sets
264, and there are
265.Nm
266commands to atomically manipulate sets, such as enable,
267disable, swap sets, move all rules in a set to another
268one, delete all rules in a set.
269These can be useful to
270install temporary configurations, or to test them.
271See Section
272.Sx SETS OF RULES
273for more information on
274.Em sets .
275.Pp
276Rules can be added with the
277.Cm add
278command; deleted individually or in groups with the
279.Cm delete
280command, and globally (except those in set 31) with the
281.Cm flush
282command; displayed, optionally with the content of the
283counters, using the
284.Cm show
285and
286.Cm list
287commands.
288Finally, counters can be reset with the
289.Cm zero
290and
291.Cm resetlog
292commands.
293.Pp
294.Ss COMMAND OPTIONS
295The following general options are available when invoking
296.Nm :
297.Bl -tag -width indent
298.It Fl a
299Show counter values when listing rules.
300The
301.Cm show
302command implies this option.
303.It Fl b
304Only show the action and the comment, not the body of a rule.
305Implies
306.Fl c .
307.It Fl c
308When entering or showing rules, print them in compact form,
309i.e., omitting the "ip from any to any" string
310when this does not carry any additional information.
311.It Fl d
312When listing, show dynamic rules in addition to static ones.
313.It Fl e
314When listing and
315.Fl d
316is specified, also show expired dynamic rules.
317.It Fl f
318Do not ask for confirmation for commands that can cause problems
319if misused, i.e.,
320.Cm flush .
321If there is no tty associated with the process, this is implied.
322.It Fl i
323When listing a table (see the
324.Sx LOOKUP TABLES
325section below for more information on lookup tables), format values
326as IP addresses.
327By default, values are shown as integers.
328.It Fl n
329Only check syntax of the command strings, without actually passing
330them to the kernel.
331.It Fl N
332Try to resolve addresses and service names in output.
333.It Fl q
334Be quiet when executing the
335.Cm add ,
336.Cm nat ,
337.Cm zero ,
338.Cm resetlog
339or
340.Cm flush
341commands;
342(implies
343.Fl f ) .
344This is useful when updating rulesets by executing multiple
345.Nm
346commands in a script
347(e.g.,
348.Ql sh\ /etc/rc.firewall ) ,
349or by processing a file with many
350.Nm
351rules across a remote login session.
352It also stops a table add or delete
353from failing if the entry already exists or is not present.
354.Pp
355The reason why this option may be important is that
356for some of these actions,
357.Nm
358may print a message; if the action results in blocking the
359traffic to the remote client,
360the remote login session will be closed
361and the rest of the ruleset will not be processed.
362Access to the console would then be required to recover.
363.It Fl S
364When listing rules, show the
365.Em set
366each rule belongs to.
367If this flag is not specified, disabled rules will not be
368listed.
369.It Fl s Op Ar field
370When listing pipes, sort according to one of the four
371counters (total or current packets or bytes).
372.It Fl t
373When listing, show last match timestamp converted with ctime().
374.It Fl T
375When listing, show last match timestamp as seconds from the epoch.
376This form can be more convenient for postprocessing by scripts.
377.El
378.Ss LIST OF RULES AND PREPROCESSING
379To ease configuration, rules can be put into a file which is
380processed using
381.Nm
382as shown in the last synopsis line.
383An absolute
384.Ar pathname
385must be used.
386The file will be read line by line and applied as arguments to the
387.Nm
388utility.
389.Pp
390Optionally, a preprocessor can be specified using
391.Fl p Ar preproc
392where
393.Ar pathname
394is to be piped through.
395Useful preprocessors include
396.Xr cpp 1
397and
398.Xr m4 1 .
399If
400.Ar preproc
401does not start with a slash
402.Pq Ql /
403as its first character, the usual
404.Ev PATH
405name search is performed.
406Care should be taken with this in environments where not all
407file systems are mounted (yet) by the time
408.Nm
409is being run (e.g.\& when they are mounted over NFS).
410Once
411.Fl p
412has been specified, any additional arguments are passed on to the preprocessor
413for interpretation.
414This allows for flexible configuration files (like conditionalizing
415them on the local hostname) and the use of macros to centralize
416frequently required arguments like IP addresses.
417.Ss TRAFFIC SHAPER CONFIGURATION
418The
419.Nm
420.Cm pipe , queue
421and
422.Cm sched
423commands are used to configure the traffic shaper and packet scheduler.
424See the
425.Sx TRAFFIC SHAPER (DUMMYNET) CONFIGURATION
426Section below for details.
427.Pp
428If the world and the kernel get out of sync the
429.Nm
430ABI may break, preventing you from being able to add any rules.
431This can adversely affect the booting process.
432You can use
433.Nm
434.Cm disable
435.Cm firewall
436to temporarily disable the firewall to regain access to the network,
437allowing you to fix the problem.
438.Sh PACKET FLOW
439A packet is checked against the active ruleset in multiple places
440in the protocol stack, under control of several sysctl variables.
441These places and variables are shown below, and it is important to
442have this picture in mind in order to design a correct ruleset.
443.Bd -literal -offset indent
444       ^    to upper layers    V
445       |                       |
446       +----------->-----------+
447       ^                       V
448 [ip(6)_input]           [ip(6)_output]     net.inet(6).ip(6).fw.enable=1
449       |                       |
450       ^                       V
451 [ether_demux]        [ether_output_frame]  net.link.ether.ipfw=1
452       |                       |
453       +-->--[bdg_forward]-->--+            net.link.bridge.ipfw=1
454       ^                       V
455       |      to devices       |
456.Ed
457.Pp
458The number of
459times the same packet goes through the firewall can
460vary between 0 and 4 depending on packet source and
461destination, and system configuration.
462.Pp
463Note that as packets flow through the stack, headers can be
464stripped or added to it, and so they may or may not be available
465for inspection.
466E.g., incoming packets will include the MAC header when
467.Nm
468is invoked from
469.Cm ether_demux() ,
470but the same packets will have the MAC header stripped off when
471.Nm
472is invoked from
473.Cm ip_input()
474or
475.Cm ip6_input() .
476.Pp
477Also note that each packet is always checked against the complete ruleset,
478irrespective of the place where the check occurs, or the source of the packet.
479If a rule contains some match patterns or actions which are not valid
480for the place of invocation (e.g.\& trying to match a MAC header within
481.Cm ip_input
482or
483.Cm ip6_input ),
484the match pattern will not match, but a
485.Cm not
486operator in front of such patterns
487.Em will
488cause the pattern to
489.Em always
490match on those packets.
491It is thus the responsibility of
492the programmer, if necessary, to write a suitable ruleset to
493differentiate among the possible places.
494.Cm skipto
495rules can be useful here, as an example:
496.Bd -literal -offset indent
497# packets from ether_demux or bdg_forward
498ipfw add 10 skipto 1000 all from any to any layer2 in
499# packets from ip_input
500ipfw add 10 skipto 2000 all from any to any not layer2 in
501# packets from ip_output
502ipfw add 10 skipto 3000 all from any to any not layer2 out
503# packets from ether_output_frame
504ipfw add 10 skipto 4000 all from any to any layer2 out
505.Ed
506.Pp
507(yes, at the moment there is no way to differentiate between
508ether_demux and bdg_forward).
509.Sh SYNTAX
510In general, each keyword or argument must be provided as
511a separate command line argument, with no leading or trailing
512spaces.
513Keywords are case-sensitive, whereas arguments may
514or may not be case-sensitive depending on their nature
515(e.g.\& uid's are, hostnames are not).
516.Pp
517Some arguments (e.g., port or address lists) are comma-separated
518lists of values.
519In this case, spaces after commas ',' are allowed to make
520the line more readable.
521You can also put the entire
522command (including flags) into a single argument.
523E.g., the following forms are equivalent:
524.Bd -literal -offset indent
525ipfw -q add deny src-ip 10.0.0.0/24,127.0.0.1/8
526ipfw -q add deny src-ip 10.0.0.0/24, 127.0.0.1/8
527ipfw "-q add deny src-ip 10.0.0.0/24, 127.0.0.1/8"
528.Ed
529.Sh RULE FORMAT
530The format of firewall rules is the following:
531.Bd -ragged -offset indent
532.Bk -words
533.Op Ar rule_number
534.Op Cm set Ar set_number
535.Op Cm prob Ar match_probability
536.Ar action
537.Op Cm log Op Cm logamount Ar number
538.Op Cm altq Ar queue
539.Oo
540.Bro Cm tag | untag
541.Brc Ar number
542.Oc
543.Ar body
544.Ek
545.Ed
546.Pp
547where the body of the rule specifies which information is used
548for filtering packets, among the following:
549.Pp
550.Bl -tag -width "Source and dest. addresses and ports" -offset XXX -compact
551.It Layer-2 header fields
552When available
553.It IPv4 and IPv6 Protocol
554SCTP, TCP, UDP, ICMP, etc.
555.It Source and dest. addresses and ports
556.It Direction
557See Section
558.Sx PACKET FLOW
559.It Transmit and receive interface
560By name or address
561.It Misc. IP header fields
562Version, type of service, datagram length, identification,
563fragment flag (non-zero IP offset),
564Time To Live
565.It IP options
566.It IPv6 Extension headers
567Fragmentation, Hop-by-Hop options,
568Routing Headers, Source routing rthdr0, Mobile IPv6 rthdr2, IPSec options.
569.It IPv6 Flow-ID
570.It Misc. TCP header fields
571TCP flags (SYN, FIN, ACK, RST, etc.),
572sequence number, acknowledgment number,
573window
574.It TCP options
575.It ICMP types
576for ICMP packets
577.It ICMP6 types
578for ICMP6 packets
579.It User/group ID
580When the packet can be associated with a local socket.
581.It Divert status
582Whether a packet came from a divert socket (e.g.,
583.Xr natd 8 ) .
584.It Fib annotation state
585Whether a packet has been tagged for using a specific FIB (routing table)
586in future forwarding decisions.
587.El
588.Pp
589Note that some of the above information, e.g.\& source MAC or IP addresses and
590TCP/UDP ports, can be easily spoofed, so filtering on those fields
591alone might not guarantee the desired results.
592.Bl -tag -width indent
593.It Ar rule_number
594Each rule is associated with a
595.Ar rule_number
596in the range 1..65535, with the latter reserved for the
597.Em default
598rule.
599Rules are checked sequentially by rule number.
600Multiple rules can have the same number, in which case they are
601checked (and listed) according to the order in which they have
602been added.
603If a rule is entered without specifying a number, the kernel will
604assign one in such a way that the rule becomes the last one
605before the
606.Em default
607rule.
608Automatic rule numbers are assigned by incrementing the last
609non-default rule number by the value of the sysctl variable
610.Ar net.inet.ip.fw.autoinc_step
611which defaults to 100.
612If this is not possible (e.g.\& because we would go beyond the
613maximum allowed rule number), the number of the last
614non-default value is used instead.
615.It Cm set Ar set_number
616Each rule is associated with a
617.Ar set_number
618in the range 0..31.
619Sets can be individually disabled and enabled, so this parameter
620is of fundamental importance for atomic ruleset manipulation.
621It can be also used to simplify deletion of groups of rules.
622If a rule is entered without specifying a set number,
623set 0 will be used.
624.br
625Set 31 is special in that it cannot be disabled,
626and rules in set 31 are not deleted by the
627.Nm ipfw flush
628command (but you can delete them with the
629.Nm ipfw delete set 31
630command).
631Set 31 is also used for the
632.Em default
633rule.
634.It Cm prob Ar match_probability
635A match is only declared with the specified probability
636(floating point number between 0 and 1).
637This can be useful for a number of applications such as
638random packet drop or
639(in conjunction with
640.Nm dummynet )
641to simulate the effect of multiple paths leading to out-of-order
642packet delivery.
643.Pp
644Note: this condition is checked before any other condition, including
645ones such as
646.Cm keep-state
647or
648.Cm check-state
649which might have
650side effects.
651.It Cm log Op Cm logamount Ar number
652Packets matching a rule with the
653.Cm log
654keyword will be made available for logging in two ways:
655if the sysctl variable
656.Va net.inet.ip.fw.verbose
657is set to 0 (default), one can use
658.Xr bpf 4
659attached to the
660.Li ipfw0
661pseudo interface.
662This pseudo interface can be created after a boot
663manually by using the following command:
664.Bd -literal -offset indent
665# ifconfig ipfw0 create
666.Ed
667.Pp
668Or, automatically at boot time by adding the following
669line to the
670.Xr rc.conf 5
671file:
672.Bd -literal -offset indent
673firewall_logif="YES"
674.Ed
675.Pp
676There is no overhead if no
677.Xr bpf 4
678is attached to the pseudo interface.
679.Pp
680If
681.Va net.inet.ip.fw.verbose
682is set to 1, packets will be logged to
683.Xr syslogd 8
684with a
685.Dv LOG_SECURITY
686facility up to a maximum of
687.Cm logamount
688packets.
689If no
690.Cm logamount
691is specified, the limit is taken from the sysctl variable
692.Va net.inet.ip.fw.verbose_limit .
693In both cases, a value of 0 means unlimited logging.
694.Pp
695Once the limit is reached, logging can be re-enabled by
696clearing the logging counter or the packet counter for that entry, see the
697.Cm resetlog
698command.
699.Pp
700Note: logging is done after all other packet matching conditions
701have been successfully verified, and before performing the final
702action (accept, deny, etc.) on the packet.
703.It Cm tag Ar number
704When a packet matches a rule with the
705.Cm tag
706keyword, the numeric tag for the given
707.Ar number
708in the range 1..65534 will be attached to the packet.
709The tag acts as an internal marker (it is not sent out over
710the wire) that can be used to identify these packets later on.
711This can be used, for example, to provide trust between interfaces
712and to start doing policy-based filtering.
713A packet can have multiple tags at the same time.
714Tags are "sticky", meaning once a tag is applied to a packet by a
715matching rule it exists until explicit removal.
716Tags are kept with the packet everywhere within the kernel, but are
717lost when packet leaves the kernel, for example, on transmitting
718packet out to the network or sending packet to a
719.Xr divert 4
720socket.
721.Pp
722To check for previously applied tags, use the
723.Cm tagged
724rule option.
725To delete previously applied tag, use the
726.Cm untag
727keyword.
728.Pp
729Note: since tags are kept with the packet everywhere in kernelspace,
730they can be set and unset anywhere in the kernel network subsystem
731(using the
732.Xr mbuf_tags 9
733facility), not only by means of the
734.Xr ipfw 4
735.Cm tag
736and
737.Cm untag
738keywords.
739For example, there can be a specialized
740.Xr netgraph 4
741node doing traffic analyzing and tagging for later inspecting
742in firewall.
743.It Cm untag Ar number
744When a packet matches a rule with the
745.Cm untag
746keyword, the tag with the number
747.Ar number
748is searched among the tags attached to this packet and,
749if found, removed from it.
750Other tags bound to packet, if present, are left untouched.
751.It Cm altq Ar queue
752When a packet matches a rule with the
753.Cm altq
754keyword, the ALTQ identifier for the given
755.Ar queue
756(see
757.Xr altq 4 )
758will be attached.
759Note that this ALTQ tag is only meaningful for packets going "out" of IPFW,
760and not being rejected or going to divert sockets.
761Note that if there is insufficient memory at the time the packet is
762processed, it will not be tagged, so it is wise to make your ALTQ
763"default" queue policy account for this.
764If multiple
765.Cm altq
766rules match a single packet, only the first one adds the ALTQ classification
767tag.
768In doing so, traffic may be shaped by using
769.Cm count Cm altq Ar queue
770rules for classification early in the ruleset, then later applying
771the filtering decision.
772For example,
773.Cm check-state
774and
775.Cm keep-state
776rules may come later and provide the actual filtering decisions in
777addition to the fallback ALTQ tag.
778.Pp
779You must run
780.Xr pfctl 8
781to set up the queues before IPFW will be able to look them up by name,
782and if the ALTQ disciplines are rearranged, the rules in containing the
783queue identifiers in the kernel will likely have gone stale and need
784to be reloaded.
785Stale queue identifiers will probably result in misclassification.
786.Pp
787All system ALTQ processing can be turned on or off via
788.Nm
789.Cm enable Ar altq
790and
791.Nm
792.Cm disable Ar altq .
793The usage of
794.Va net.inet.ip.fw.one_pass
795is irrelevant to ALTQ traffic shaping, as the actual rule action is followed
796always after adding an ALTQ tag.
797.El
798.Ss RULE ACTIONS
799A rule can be associated with one of the following actions, which
800will be executed when the packet matches the body of the rule.
801.Bl -tag -width indent
802.It Cm allow | accept | pass | permit
803Allow packets that match rule.
804The search terminates.
805.It Cm check-state Op Ar :flowname | Cm :any
806Checks the packet against the dynamic ruleset.
807If a match is found, execute the action associated with
808the rule which generated this dynamic rule, otherwise
809move to the next rule.
810.br
811.Cm Check-state
812rules do not have a body.
813If no
814.Cm check-state
815rule is found, the dynamic ruleset is checked at the first
816.Cm keep-state
817or
818.Cm limit
819rule.
820The
821.Ar :flowname
822is symbolic name assigned to dynamic rule by
823.Cm keep-state
824opcode.
825The special flowname
826.Cm :any
827can be used to ignore states flowname when matching.
828The
829.Cm :default
830keyword is special name used for compatibility with old rulesets.
831.It Cm count
832Update counters for all packets that match rule.
833The search continues with the next rule.
834.It Cm deny | drop
835Discard packets that match this rule.
836The search terminates.
837.It Cm divert Ar port
838Divert packets that match this rule to the
839.Xr divert 4
840socket bound to port
841.Ar port .
842The search terminates.
843.It Cm fwd | forward Ar ipaddr | tablearg Ns Op , Ns Ar port
844Change the next-hop on matching packets to
845.Ar ipaddr ,
846which can be an IP address or a host name.
847The next hop can also be supplied by the last table
848looked up for the packet by using the
849.Cm tablearg
850keyword instead of an explicit address.
851The search terminates if this rule matches.
852.Pp
853If
854.Ar ipaddr
855is a local address, then matching packets will be forwarded to
856.Ar port
857(or the port number in the packet if one is not specified in the rule)
858on the local machine.
859.br
860If
861.Ar ipaddr
862is not a local address, then the port number
863(if specified) is ignored, and the packet will be
864forwarded to the remote address, using the route as found in
865the local routing table for that IP.
866.br
867A
868.Ar fwd
869rule will not match layer-2 packets (those received
870on ether_input, ether_output, or bridged).
871.br
872The
873.Cm fwd
874action does not change the contents of the packet at all.
875In particular, the destination address remains unmodified, so
876packets forwarded to another system will usually be rejected by that system
877unless there is a matching rule on that system to capture them.
878For packets forwarded locally,
879the local address of the socket will be
880set to the original destination address of the packet.
881This makes the
882.Xr netstat 1
883entry look rather weird but is intended for
884use with transparent proxy servers.
885.It Cm nat Ar nat_nr | tablearg
886Pass packet to a
887nat instance
888(for network address translation, address redirect, etc.):
889see the
890.Sx NETWORK ADDRESS TRANSLATION (NAT)
891Section for further information.
892.It Cm nat64lsn Ar name
893Pass packet to a stateful NAT64 instance (for IPv6/IPv4 network address and
894protocol translation): see the
895.Sx IPv6/IPv4 NETWORK ADDRESS AND PROTOCOL TRANSLATION
896Section for further information.
897.It Cm nat64stl Ar name
898Pass packet to a stateless NAT64 instance (for IPv6/IPv4 network address and
899protocol translation): see the
900.Sx IPv6/IPv4 NETWORK ADDRESS AND PROTOCOL TRANSLATION
901Section for further information.
902.It Cm nptv6 Ar name
903Pass packet to a NPTv6 instance (for IPv6-to-IPv6 network prefix translation):
904see the
905.Sx IPv6-to-IPv6 NETWORK PREFIX TRANSLATION (NPTv6)
906Section for further information.
907.It Cm pipe Ar pipe_nr
908Pass packet to a
909.Nm dummynet
910.Dq pipe
911(for bandwidth limitation, delay, etc.).
912See the
913.Sx TRAFFIC SHAPER (DUMMYNET) CONFIGURATION
914Section for further information.
915The search terminates; however, on exit from the pipe and if
916the
917.Xr sysctl 8
918variable
919.Va net.inet.ip.fw.one_pass
920is not set, the packet is passed again to the firewall code
921starting from the next rule.
922.It Cm queue Ar queue_nr
923Pass packet to a
924.Nm dummynet
925.Dq queue
926(for bandwidth limitation using WF2Q+).
927.It Cm reject
928(Deprecated).
929Synonym for
930.Cm unreach host .
931.It Cm reset
932Discard packets that match this rule, and if the
933packet is a TCP packet, try to send a TCP reset (RST) notice.
934The search terminates.
935.It Cm reset6
936Discard packets that match this rule, and if the
937packet is a TCP packet, try to send a TCP reset (RST) notice.
938The search terminates.
939.It Cm skipto Ar number | tablearg
940Skip all subsequent rules numbered less than
941.Ar number .
942The search continues with the first rule numbered
943.Ar number
944or higher.
945It is possible to use the
946.Cm tablearg
947keyword with a skipto for a
948.Em computed
949skipto. Skipto may work either in O(log(N)) or in O(1) depending
950on amount of memory and/or sysctl variables.
951See the
952.Sx SYSCTL VARIABLES
953section for more details.
954.It Cm call Ar number | tablearg
955The current rule number is saved in the internal stack and
956ruleset processing continues with the first rule numbered
957.Ar number
958or higher.
959If later a rule with the
960.Cm return
961action is encountered, the processing returns to the first rule
962with number of this
963.Cm call
964rule plus one or higher
965(the same behaviour as with packets returning from
966.Xr divert 4
967socket after a
968.Cm divert
969action).
970This could be used to make somewhat like an assembly language
971.Dq subroutine
972calls to rules with common checks for different interfaces, etc.
973.Pp
974Rule with any number could be called, not just forward jumps as with
975.Cm skipto .
976So, to prevent endless loops in case of mistakes, both
977.Cm call
978and
979.Cm return
980actions don't do any jumps and simply go to the next rule if memory
981cannot be allocated or stack overflowed/underflowed.
982.Pp
983Internally stack for rule numbers is implemented using
984.Xr mbuf_tags 9
985facility and currently has size of 16 entries.
986As mbuf tags are lost when packet leaves the kernel,
987.Cm divert
988should not be used in subroutines to avoid endless loops
989and other undesired effects.
990.It Cm return
991Takes rule number saved to internal stack by the last
992.Cm call
993action and returns ruleset processing to the first rule
994with number greater than number of corresponding
995.Cm call
996rule.
997See description of the
998.Cm call
999action for more details.
1000.Pp
1001Note that
1002.Cm return
1003rules usually end a
1004.Dq subroutine
1005and thus are unconditional, but
1006.Nm
1007command-line utility currently requires every action except
1008.Cm check-state
1009to have body.
1010While it is sometimes useful to return only on some packets,
1011usually you want to print just
1012.Dq return
1013for readability.
1014A workaround for this is to use new syntax and
1015.Fl c
1016switch:
1017.Bd -literal -offset indent
1018# Add a rule without actual body
1019ipfw add 2999 return via any
1020
1021# List rules without "from any to any" part
1022ipfw -c list
1023.Ed
1024.Pp
1025This cosmetic annoyance may be fixed in future releases.
1026.It Cm tee Ar port
1027Send a copy of packets matching this rule to the
1028.Xr divert 4
1029socket bound to port
1030.Ar port .
1031The search continues with the next rule.
1032.It Cm unreach Ar code
1033Discard packets that match this rule, and try to send an ICMP
1034unreachable notice with code
1035.Ar code ,
1036where
1037.Ar code
1038is a number from 0 to 255, or one of these aliases:
1039.Cm net , host , protocol , port ,
1040.Cm needfrag , srcfail , net-unknown , host-unknown ,
1041.Cm isolated , net-prohib , host-prohib , tosnet ,
1042.Cm toshost , filter-prohib , host-precedence
1043or
1044.Cm precedence-cutoff .
1045The search terminates.
1046.It Cm unreach6 Ar code
1047Discard packets that match this rule, and try to send an ICMPv6
1048unreachable notice with code
1049.Ar code ,
1050where
1051.Ar code
1052is a number from 0, 1, 3 or 4, or one of these aliases:
1053.Cm no-route, admin-prohib, address
1054or
1055.Cm port .
1056The search terminates.
1057.It Cm netgraph Ar cookie
1058Divert packet into netgraph with given
1059.Ar cookie .
1060The search terminates.
1061If packet is later returned from netgraph it is either
1062accepted or continues with the next rule, depending on
1063.Va net.inet.ip.fw.one_pass
1064sysctl variable.
1065.It Cm ngtee Ar cookie
1066A copy of packet is diverted into netgraph, original
1067packet continues with the next rule.
1068See
1069.Xr ng_ipfw 4
1070for more information on
1071.Cm netgraph
1072and
1073.Cm ngtee
1074actions.
1075.It Cm setfib Ar fibnum | tablearg
1076The packet is tagged so as to use the FIB (routing table)
1077.Ar fibnum
1078in any subsequent forwarding decisions.
1079In the current implementation, this is limited to the values 0 through 15, see
1080.Xr setfib 2 .
1081Processing continues at the next rule.
1082It is possible to use the
1083.Cm tablearg
1084keyword with setfib.
1085If the tablearg value is not within the compiled range of fibs,
1086the packet's fib is set to 0.
1087.It Cm setdscp Ar DSCP | number | tablearg
1088Set specified DiffServ codepoint for an IPv4/IPv6 packet.
1089Processing continues at the next rule.
1090Supported values are:
1091.Pp
1092.Cm cs0
1093.Pq Dv 000000 ,
1094.Cm cs1
1095.Pq Dv 001000 ,
1096.Cm cs2
1097.Pq Dv 010000 ,
1098.Cm cs3
1099.Pq Dv 011000 ,
1100.Cm cs4
1101.Pq Dv 100000 ,
1102.Cm cs5
1103.Pq Dv 101000 ,
1104.Cm cs6
1105.Pq Dv 110000 ,
1106.Cm cs7
1107.Pq Dv 111000 ,
1108.Cm af11
1109.Pq Dv 001010 ,
1110.Cm af12
1111.Pq Dv 001100 ,
1112.Cm af13
1113.Pq Dv 001110 ,
1114.Cm af21
1115.Pq Dv 010010 ,
1116.Cm af22
1117.Pq Dv 010100 ,
1118.Cm af23
1119.Pq Dv 010110 ,
1120.Cm af31
1121.Pq Dv 011010 ,
1122.Cm af32
1123.Pq Dv 011100 ,
1124.Cm af33
1125.Pq Dv 011110 ,
1126.Cm af41
1127.Pq Dv 100010 ,
1128.Cm af42
1129.Pq Dv 100100 ,
1130.Cm af43
1131.Pq Dv 100110 ,
1132.Cm ef
1133.Pq Dv 101110 ,
1134.Cm be
1135.Pq Dv 000000 .
1136Additionally, DSCP value can be specified by number (0..64).
1137It is also possible to use the
1138.Cm tablearg
1139keyword with setdscp.
1140If the tablearg value is not within the 0..64 range, lower 6 bits of supplied
1141value are used.
1142.It Cm tcp-setmss Ar mss
1143Set the Maximum Segment Size (MSS) in the TCP segment to value
1144.Ar mss .
1145The kernel module
1146.Cm ipfw_pmod
1147should be loaded or kernel should have
1148.Cm options IPFIREWALL_PMOD
1149to be able use this action.
1150This command does not change a packet if original MSS value is lower than
1151specified value.
1152Both TCP over IPv4 and over IPv6 are supported.
1153Regardless of matched a packet or not by the
1154.Cm tcp-setmss
1155rule, the search continues with the next rule.
1156.It Cm reass
1157Queue and reassemble IPv4 fragments.
1158If the packet is not fragmented, counters are updated and
1159processing continues with the next rule.
1160If the packet is the last logical fragment, the packet is reassembled and, if
1161.Va net.inet.ip.fw.one_pass
1162is set to 0, processing continues with the next rule.
1163Otherwise, the packet is allowed to pass and the search terminates.
1164If the packet is a fragment in the middle of a logical group of fragments,
1165it is consumed and
1166processing stops immediately.
1167.Pp
1168Fragment handling can be tuned via
1169.Va net.inet.ip.maxfragpackets
1170and
1171.Va net.inet.ip.maxfragsperpacket
1172which limit, respectively, the maximum number of processable
1173fragments (default: 800) and
1174the maximum number of fragments per packet (default: 16).
1175.Pp
1176NOTA BENE: since fragments do not contain port numbers,
1177they should be avoided with the
1178.Nm reass
1179rule.
1180Alternatively, direction-based (like
1181.Nm in
1182/
1183.Nm out
1184) and source-based (like
1185.Nm via
1186) match patterns can be used to select fragments.
1187.Pp
1188Usually a simple rule like:
1189.Bd -literal -offset indent
1190# reassemble incoming fragments
1191ipfw add reass all from any to any in
1192.Ed
1193.Pp
1194is all you need at the beginning of your ruleset.
1195.It Cm abort
1196Discard packets that match this rule, and if the packet is an SCTP packet,
1197try to send an SCTP packet containing an ABORT chunk.
1198The search terminates.
1199.It Cm abort6
1200Discard packets that match this rule, and if the packet is an SCTP packet,
1201try to send an SCTP packet containing an ABORT chunk.
1202The search terminates.
1203.El
1204.Ss RULE BODY
1205The body of a rule contains zero or more patterns (such as
1206specific source and destination addresses or ports,
1207protocol options, incoming or outgoing interfaces, etc.)
1208that the packet must match in order to be recognised.
1209In general, the patterns are connected by (implicit)
1210.Cm and
1211operators -- i.e., all must match in order for the
1212rule to match.
1213Individual patterns can be prefixed by the
1214.Cm not
1215operator to reverse the result of the match, as in
1216.Pp
1217.Dl "ipfw add 100 allow ip from not 1.2.3.4 to any"
1218.Pp
1219Additionally, sets of alternative match patterns
1220.Pq Em or-blocks
1221can be constructed by putting the patterns in
1222lists enclosed between parentheses ( ) or braces { }, and
1223using the
1224.Cm or
1225operator as follows:
1226.Pp
1227.Dl "ipfw add 100 allow ip from { x or not y or z } to any"
1228.Pp
1229Only one level of parentheses is allowed.
1230Beware that most shells have special meanings for parentheses
1231or braces, so it is advisable to put a backslash \\ in front of them
1232to prevent such interpretations.
1233.Pp
1234The body of a rule must in general include a source and destination
1235address specifier.
1236The keyword
1237.Ar any
1238can be used in various places to specify that the content of
1239a required field is irrelevant.
1240.Pp
1241The rule body has the following format:
1242.Bd -ragged -offset indent
1243.Op Ar proto Cm from Ar src Cm to Ar dst
1244.Op Ar options
1245.Ed
1246.Pp
1247The first part (proto from src to dst) is for backward
1248compatibility with earlier versions of
1249.Fx .
1250In modern
1251.Fx
1252any match pattern (including MAC headers, IP protocols,
1253addresses and ports) can be specified in the
1254.Ar options
1255section.
1256.Pp
1257Rule fields have the following meaning:
1258.Bl -tag -width indent
1259.It Ar proto : protocol | Cm { Ar protocol Cm or ... }
1260.It Ar protocol : Oo Cm not Oc Ar protocol-name | protocol-number
1261An IP protocol specified by number or name
1262(for a complete list see
1263.Pa /etc/protocols ) ,
1264or one of the following keywords:
1265.Bl -tag -width indent
1266.It Cm ip4 | ipv4
1267Matches IPv4 packets.
1268.It Cm ip6 | ipv6
1269Matches IPv6 packets.
1270.It Cm ip | all
1271Matches any packet.
1272.El
1273.Pp
1274The
1275.Cm ipv6
1276in
1277.Cm proto
1278option will be treated as inner protocol.
1279And, the
1280.Cm ipv4
1281is not available in
1282.Cm proto
1283option.
1284.Pp
1285The
1286.Cm { Ar protocol Cm or ... }
1287format (an
1288.Em or-block )
1289is provided for convenience only but its use is deprecated.
1290.It Ar src No and Ar dst : Bro Cm addr | Cm { Ar addr Cm or ... } Brc Op Oo Cm not Oc Ar ports
1291An address (or a list, see below)
1292optionally followed by
1293.Ar ports
1294specifiers.
1295.Pp
1296The second format
1297.Em ( or-block
1298with multiple addresses) is provided for convenience only and
1299its use is discouraged.
1300.It Ar addr : Oo Cm not Oc Bro
1301.Cm any | me | me6 |
1302.Cm table Ns Pq Ar name Ns Op , Ns Ar value
1303.Ar | addr-list | addr-set
1304.Brc
1305.Bl -tag -width indent
1306.It Cm any
1307matches any IP address.
1308.It Cm me
1309matches any IP address configured on an interface in the system.
1310.It Cm me6
1311matches any IPv6 address configured on an interface in the system.
1312The address list is evaluated at the time the packet is
1313analysed.
1314.It Cm table Ns Pq Ar name Ns Op , Ns Ar value
1315Matches any IPv4 or IPv6 address for which an entry exists in the lookup table
1316.Ar number .
1317If an optional 32-bit unsigned
1318.Ar value
1319is also specified, an entry will match only if it has this value.
1320See the
1321.Sx LOOKUP TABLES
1322section below for more information on lookup tables.
1323.El
1324.It Ar addr-list : ip-addr Ns Op Ns , Ns Ar addr-list
1325.It Ar ip-addr :
1326A host or subnet address specified in one of the following ways:
1327.Bl -tag -width indent
1328.It Ar numeric-ip | hostname
1329Matches a single IPv4 address, specified as dotted-quad or a hostname.
1330Hostnames are resolved at the time the rule is added to the firewall list.
1331.It Ar addr Ns / Ns Ar masklen
1332Matches all addresses with base
1333.Ar addr
1334(specified as an IP address, a network number, or a hostname)
1335and mask width of
1336.Cm masklen
1337bits.
1338As an example, 1.2.3.4/25 or 1.2.3.0/25 will match
1339all IP numbers from 1.2.3.0 to 1.2.3.127 .
1340.It Ar addr Ns : Ns Ar mask
1341Matches all addresses with base
1342.Ar addr
1343(specified as an IP address, a network number, or a hostname)
1344and the mask of
1345.Ar mask ,
1346specified as a dotted quad.
1347As an example, 1.2.3.4:255.0.255.0 or 1.0.3.0:255.0.255.0 will match
13481.*.3.*.
1349This form is advised only for non-contiguous
1350masks.
1351It is better to resort to the
1352.Ar addr Ns / Ns Ar masklen
1353format for contiguous masks, which is more compact and less
1354error-prone.
1355.El
1356.It Ar addr-set : addr Ns Oo Ns / Ns Ar masklen Oc Ns Cm { Ns Ar list Ns Cm }
1357.It Ar list : Bro Ar num | num-num Brc Ns Op Ns , Ns Ar list
1358Matches all addresses with base address
1359.Ar addr
1360(specified as an IP address, a network number, or a hostname)
1361and whose last byte is in the list between braces { } .
1362Note that there must be no spaces between braces and
1363numbers (spaces after commas are allowed).
1364Elements of the list can be specified as single entries
1365or ranges.
1366The
1367.Ar masklen
1368field is used to limit the size of the set of addresses,
1369and can have any value between 24 and 32.
1370If not specified,
1371it will be assumed as 24.
1372.br
1373This format is particularly useful to handle sparse address sets
1374within a single rule.
1375Because the matching occurs using a
1376bitmask, it takes constant time and dramatically reduces
1377the complexity of rulesets.
1378.br
1379As an example, an address specified as 1.2.3.4/24{128,35-55,89}
1380or 1.2.3.0/24{128,35-55,89}
1381will match the following IP addresses:
1382.br
13831.2.3.128, 1.2.3.35 to 1.2.3.55, 1.2.3.89 .
1384.It Ar addr6-list : ip6-addr Ns Op Ns , Ns Ar addr6-list
1385.It Ar ip6-addr :
1386A host or subnet specified one of the following ways:
1387.Bl -tag -width indent
1388.It Ar numeric-ip | hostname
1389Matches a single IPv6 address as allowed by
1390.Xr inet_pton 3
1391or a hostname.
1392Hostnames are resolved at the time the rule is added to the firewall
1393list.
1394.It Ar addr Ns / Ns Ar masklen
1395Matches all IPv6 addresses with base
1396.Ar addr
1397(specified as allowed by
1398.Xr inet_pton
1399or a hostname)
1400and mask width of
1401.Cm masklen
1402bits.
1403.It Ar addr Ns / Ns Ar mask
1404Matches all IPv6 addresses with base
1405.Ar addr
1406(specified as allowed by
1407.Xr inet_pton
1408or a hostname)
1409and the mask of
1410.Ar mask ,
1411specified as allowed by
1412.Xr inet_pton.
1413As an example, fe::640:0:0/ffff::ffff:ffff:0:0 will match
1414fe:*:*:*:0:640:*:*.
1415This form is advised only for non-contiguous
1416masks.
1417It is better to resort to the
1418.Ar addr Ns / Ns Ar masklen
1419format for contiguous masks, which is more compact and less
1420error-prone.
1421.El
1422.Pp
1423No support for sets of IPv6 addresses is provided because IPv6 addresses
1424are typically random past the initial prefix.
1425.It Ar ports : Bro Ar port | port Ns \&- Ns Ar port Ns Brc Ns Op , Ns Ar ports
1426For protocols which support port numbers (such as SCTP, TCP and UDP), optional
1427.Cm ports
1428may be specified as one or more ports or port ranges, separated
1429by commas but no spaces, and an optional
1430.Cm not
1431operator.
1432The
1433.Ql \&-
1434notation specifies a range of ports (including boundaries).
1435.Pp
1436Service names (from
1437.Pa /etc/services )
1438may be used instead of numeric port values.
1439The length of the port list is limited to 30 ports or ranges,
1440though one can specify larger ranges by using an
1441.Em or-block
1442in the
1443.Cm options
1444section of the rule.
1445.Pp
1446A backslash
1447.Pq Ql \e
1448can be used to escape the dash
1449.Pq Ql -
1450character in a service name (from a shell, the backslash must be
1451typed twice to avoid the shell itself interpreting it as an escape
1452character).
1453.Pp
1454.Dl "ipfw add count tcp from any ftp\e\e-data-ftp to any"
1455.Pp
1456Fragmented packets which have a non-zero offset (i.e., not the first
1457fragment) will never match a rule which has one or more port
1458specifications.
1459See the
1460.Cm frag
1461option for details on matching fragmented packets.
1462.El
1463.Ss RULE OPTIONS (MATCH PATTERNS)
1464Additional match patterns can be used within
1465rules.
1466Zero or more of these so-called
1467.Em options
1468can be present in a rule, optionally prefixed by the
1469.Cm not
1470operand, and possibly grouped into
1471.Em or-blocks .
1472.Pp
1473The following match patterns can be used (listed in alphabetical order):
1474.Bl -tag -width indent
1475.It Cm // this is a comment.
1476Inserts the specified text as a comment in the rule.
1477Everything following // is considered as a comment and stored in the rule.
1478You can have comment-only rules, which are listed as having a
1479.Cm count
1480action followed by the comment.
1481.It Cm bridged
1482Alias for
1483.Cm layer2 .
1484.It Cm defer-immediate-action | defer-action
1485A rule with this option will not perform normal action
1486upon a match. This option is intended to be used with
1487.Cm record-state
1488or
1489.Cm keep-state
1490as the dynamic rule, created but ignored on match, will work
1491as intended.
1492Rules with both
1493.Cm record-state
1494and
1495.Cm defer-immediate-action
1496create a dynamic rule and continue with the next rule without actually
1497performing the action part of this rule. When the rule is later activated
1498via the state table, the action is performed as usual.
1499.It Cm diverted
1500Matches only packets generated by a divert socket.
1501.It Cm diverted-loopback
1502Matches only packets coming from a divert socket back into the IP stack
1503input for delivery.
1504.It Cm diverted-output
1505Matches only packets going from a divert socket back outward to the IP
1506stack output for delivery.
1507.It Cm dst-ip Ar ip-address
1508Matches IPv4 packets whose destination IP is one of the address(es)
1509specified as argument.
1510.It Bro Cm dst-ip6 | dst-ipv6 Brc Ar ip6-address
1511Matches IPv6 packets whose destination IP is one of the address(es)
1512specified as argument.
1513.It Cm dst-port Ar ports
1514Matches IP packets whose destination port is one of the port(s)
1515specified as argument.
1516.It Cm established
1517Matches TCP packets that have the RST or ACK bits set.
1518.It Cm ext6hdr Ar header
1519Matches IPv6 packets containing the extended header given by
1520.Ar header .
1521Supported headers are:
1522.Pp
1523Fragment,
1524.Pq Cm frag ,
1525Hop-to-hop options
1526.Pq Cm hopopt ,
1527any type of Routing Header
1528.Pq Cm route ,
1529Source routing Routing Header Type 0
1530.Pq Cm rthdr0 ,
1531Mobile IPv6 Routing Header Type 2
1532.Pq Cm rthdr2 ,
1533Destination options
1534.Pq Cm dstopt ,
1535IPSec authentication headers
1536.Pq Cm ah ,
1537and IPsec encapsulated security payload headers
1538.Pq Cm esp .
1539.It Cm fib Ar fibnum
1540Matches a packet that has been tagged to use
1541the given FIB (routing table) number.
1542.It Cm flow Ar table Ns Pq Ar name Ns Op , Ns Ar value
1543Search for the flow entry in lookup table
1544.Ar name .
1545If not found, the match fails.
1546Otherwise, the match succeeds and
1547.Cm tablearg
1548is set to the value extracted from the table.
1549.Pp
1550This option can be useful to quickly dispatch traffic based on
1551certain packet fields.
1552See the
1553.Sx LOOKUP TABLES
1554section below for more information on lookup tables.
1555.It Cm flow-id Ar labels
1556Matches IPv6 packets containing any of the flow labels given in
1557.Ar labels .
1558.Ar labels
1559is a comma separated list of numeric flow labels.
1560.It Cm frag
1561Matches packets that are fragments and not the first
1562fragment of an IP datagram.
1563Note that these packets will not have
1564the next protocol header (e.g.\& TCP, UDP) so options that look into
1565these headers cannot match.
1566.It Cm gid Ar group
1567Matches all TCP or UDP packets sent by or received for a
1568.Ar group .
1569A
1570.Ar group
1571may be specified by name or number.
1572.It Cm jail Ar jail
1573Matches all TCP or UDP packets sent by or received for the
1574jail whose ID or name is
1575.Ar jail .
1576.It Cm icmptypes Ar types
1577Matches ICMP packets whose ICMP type is in the list
1578.Ar types .
1579The list may be specified as any combination of
1580individual types (numeric) separated by commas.
1581.Em Ranges are not allowed .
1582The supported ICMP types are:
1583.Pp
1584echo reply
1585.Pq Cm 0 ,
1586destination unreachable
1587.Pq Cm 3 ,
1588source quench
1589.Pq Cm 4 ,
1590redirect
1591.Pq Cm 5 ,
1592echo request
1593.Pq Cm 8 ,
1594router advertisement
1595.Pq Cm 9 ,
1596router solicitation
1597.Pq Cm 10 ,
1598time-to-live exceeded
1599.Pq Cm 11 ,
1600IP header bad
1601.Pq Cm 12 ,
1602timestamp request
1603.Pq Cm 13 ,
1604timestamp reply
1605.Pq Cm 14 ,
1606information request
1607.Pq Cm 15 ,
1608information reply
1609.Pq Cm 16 ,
1610address mask request
1611.Pq Cm 17
1612and address mask reply
1613.Pq Cm 18 .
1614.It Cm icmp6types Ar types
1615Matches ICMP6 packets whose ICMP6 type is in the list of
1616.Ar types .
1617The list may be specified as any combination of
1618individual types (numeric) separated by commas.
1619.Em Ranges are not allowed .
1620.It Cm in | out
1621Matches incoming or outgoing packets, respectively.
1622.Cm in
1623and
1624.Cm out
1625are mutually exclusive (in fact,
1626.Cm out
1627is implemented as
1628.Cm not in Ns No ).
1629.It Cm ipid Ar id-list
1630Matches IPv4 packets whose
1631.Cm ip_id
1632field has value included in
1633.Ar id-list ,
1634which is either a single value or a list of values or ranges
1635specified in the same way as
1636.Ar ports .
1637.It Cm iplen Ar len-list
1638Matches IP packets whose total length, including header and data, is
1639in the set
1640.Ar len-list ,
1641which is either a single value or a list of values or ranges
1642specified in the same way as
1643.Ar ports .
1644.It Cm ipoptions Ar spec
1645Matches packets whose IPv4 header contains the comma separated list of
1646options specified in
1647.Ar spec .
1648The supported IP options are:
1649.Pp
1650.Cm ssrr
1651(strict source route),
1652.Cm lsrr
1653(loose source route),
1654.Cm rr
1655(record packet route) and
1656.Cm ts
1657(timestamp).
1658The absence of a particular option may be denoted
1659with a
1660.Ql \&! .
1661.It Cm ipprecedence Ar precedence
1662Matches IPv4 packets whose precedence field is equal to
1663.Ar precedence .
1664.It Cm ipsec
1665Matches packets that have IPSEC history associated with them
1666(i.e., the packet comes encapsulated in IPSEC, the kernel
1667has IPSEC support, and can correctly decapsulate it).
1668.Pp
1669Note that specifying
1670.Cm ipsec
1671is different from specifying
1672.Cm proto Ar ipsec
1673as the latter will only look at the specific IP protocol field,
1674irrespective of IPSEC kernel support and the validity of the IPSEC data.
1675.Pp
1676Further note that this flag is silently ignored in kernels without
1677IPSEC support.
1678It does not affect rule processing when given and the
1679rules are handled as if with no
1680.Cm ipsec
1681flag.
1682.It Cm iptos Ar spec
1683Matches IPv4 packets whose
1684.Cm tos
1685field contains the comma separated list of
1686service types specified in
1687.Ar spec .
1688The supported IP types of service are:
1689.Pp
1690.Cm lowdelay
1691.Pq Dv IPTOS_LOWDELAY ,
1692.Cm throughput
1693.Pq Dv IPTOS_THROUGHPUT ,
1694.Cm reliability
1695.Pq Dv IPTOS_RELIABILITY ,
1696.Cm mincost
1697.Pq Dv IPTOS_MINCOST ,
1698.Cm congestion
1699.Pq Dv IPTOS_ECN_CE .
1700The absence of a particular type may be denoted
1701with a
1702.Ql \&! .
1703.It Cm dscp spec Ns Op , Ns Ar spec
1704Matches IPv4/IPv6 packets whose
1705.Cm DS
1706field value is contained in
1707.Ar spec
1708mask.
1709Multiple values can be specified via
1710the comma separated list.
1711Value can be one of keywords used in
1712.Cm setdscp
1713action or exact number.
1714.It Cm ipttl Ar ttl-list
1715Matches IPv4 packets whose time to live is included in
1716.Ar ttl-list ,
1717which is either a single value or a list of values or ranges
1718specified in the same way as
1719.Ar ports .
1720.It Cm ipversion Ar ver
1721Matches IP packets whose IP version field is
1722.Ar ver .
1723.It Cm keep-state Op Ar :flowname
1724Upon a match, the firewall will create a dynamic rule, whose
1725default behaviour is to match bidirectional traffic between
1726source and destination IP/port using the same protocol.
1727The rule has a limited lifetime (controlled by a set of
1728.Xr sysctl 8
1729variables), and the lifetime is refreshed every time a matching
1730packet is found.
1731The
1732.Ar :flowname
1733is used to assign additional to addresses, ports and protocol parameter
1734to dynamic rule. It can be used for more accurate matching by
1735.Cm check-state
1736rule.
1737The
1738.Cm :default
1739keyword is special name used for compatibility with old rulesets.
1740.It Cm layer2
1741Matches only layer2 packets, i.e., those passed to
1742.Nm
1743from ether_demux() and ether_output_frame().
1744.It Cm limit Bro Cm src-addr | src-port | dst-addr | dst-port Brc Ar N Op Ar :flowname
1745The firewall will only allow
1746.Ar N
1747connections with the same
1748set of parameters as specified in the rule.
1749One or more
1750of source and destination addresses and ports can be
1751specified.
1752.It Cm lookup Bro Cm dst-ip | dst-port | src-ip | src-port | uid | jail Brc Ar name
1753Search an entry in lookup table
1754.Ar name
1755that matches the field specified as argument.
1756If not found, the match fails.
1757Otherwise, the match succeeds and
1758.Cm tablearg
1759is set to the value extracted from the table.
1760.Pp
1761This option can be useful to quickly dispatch traffic based on
1762certain packet fields.
1763See the
1764.Sx LOOKUP TABLES
1765section below for more information on lookup tables.
1766.It Cm { MAC | mac } Ar dst-mac src-mac
1767Match packets with a given
1768.Ar dst-mac
1769and
1770.Ar src-mac
1771addresses, specified as the
1772.Cm any
1773keyword (matching any MAC address), or six groups of hex digits
1774separated by colons,
1775and optionally followed by a mask indicating the significant bits.
1776The mask may be specified using either of the following methods:
1777.Bl -enum -width indent
1778.It
1779A slash
1780.Pq /
1781followed by the number of significant bits.
1782For example, an address with 33 significant bits could be specified as:
1783.Pp
1784.Dl "MAC 10:20:30:40:50:60/33 any"
1785.It
1786An ampersand
1787.Pq &
1788followed by a bitmask specified as six groups of hex digits separated
1789by colons.
1790For example, an address in which the last 16 bits are significant could
1791be specified as:
1792.Pp
1793.Dl "MAC 10:20:30:40:50:60&00:00:00:00:ff:ff any"
1794.Pp
1795Note that the ampersand character has a special meaning in many shells
1796and should generally be escaped.
1797.El
1798Note that the order of MAC addresses (destination first,
1799source second) is
1800the same as on the wire, but the opposite of the one used for
1801IP addresses.
1802.It Cm mac-type Ar mac-type
1803Matches packets whose Ethernet Type field
1804corresponds to one of those specified as argument.
1805.Ar mac-type
1806is specified in the same way as
1807.Cm port numbers
1808(i.e., one or more comma-separated single values or ranges).
1809You can use symbolic names for known values such as
1810.Em vlan , ipv4, ipv6 .
1811Values can be entered as decimal or hexadecimal (if prefixed by 0x),
1812and they are always printed as hexadecimal (unless the
1813.Cm -N
1814option is used, in which case symbolic resolution will be attempted).
1815.It Cm proto Ar protocol
1816Matches packets with the corresponding IP protocol.
1817.It Cm record-state
1818Upon a match, the firewall will create a dynamic rule as if
1819.Cm keep-state
1820was specified.
1821However, this option doesn't imply an implicit
1822.Cm check-state
1823in contrast to
1824.Cm keep-state .
1825.It Cm recv | xmit | via Brq Ar ifX | Ar if Ns Cm * | Ar table Ns Po Ar name Ns Oo , Ns Ar value Oc Pc | Ar ipno | Ar any
1826Matches packets received, transmitted or going through,
1827respectively, the interface specified by exact name
1828.Po Ar ifX Pc ,
1829by device name
1830.Po Ar if* Pc ,
1831by IP address, or through some interface.
1832Table
1833.Ar name
1834may be used to match interface by its kernel ifindex.
1835See the
1836.Sx LOOKUP TABLES
1837section below for more information on lookup tables.
1838.Pp
1839The
1840.Cm via
1841keyword causes the interface to always be checked.
1842If
1843.Cm recv
1844or
1845.Cm xmit
1846is used instead of
1847.Cm via ,
1848then only the receive or transmit interface (respectively)
1849is checked.
1850By specifying both, it is possible to match packets based on
1851both receive and transmit interface, e.g.:
1852.Pp
1853.Dl "ipfw add deny ip from any to any out recv ed0 xmit ed1"
1854.Pp
1855The
1856.Cm recv
1857interface can be tested on either incoming or outgoing packets,
1858while the
1859.Cm xmit
1860interface can only be tested on outgoing packets.
1861So
1862.Cm out
1863is required (and
1864.Cm in
1865is invalid) whenever
1866.Cm xmit
1867is used.
1868.Pp
1869A packet might not have a receive or transmit interface: packets
1870originating from the local host have no receive interface,
1871while packets destined for the local host have no transmit
1872interface.
1873.It Cm set-limit Bro Cm src-addr | src-port | dst-addr | dst-port Brc Ar N
1874Works like
1875.Cm limit
1876but does not have an implicit
1877.Cm check-state
1878attached to it.
1879.It Cm setup
1880Matches TCP packets that have the SYN bit set but no ACK bit.
1881This is the short form of
1882.Dq Li tcpflags\ syn,!ack .
1883.It Cm sockarg
1884Matches packets that are associated to a local socket and
1885for which the SO_USER_COOKIE socket option has been set
1886to a non-zero value.
1887As a side effect, the value of the
1888option is made available as
1889.Cm tablearg
1890value, which in turn can be used as
1891.Cm skipto
1892or
1893.Cm pipe
1894number.
1895.It Cm src-ip Ar ip-address
1896Matches IPv4 packets whose source IP is one of the address(es)
1897specified as an argument.
1898.It Cm src-ip6 Ar ip6-address
1899Matches IPv6 packets whose source IP is one of the address(es)
1900specified as an argument.
1901.It Cm src-port Ar ports
1902Matches IP packets whose source port is one of the port(s)
1903specified as argument.
1904.It Cm tagged Ar tag-list
1905Matches packets whose tags are included in
1906.Ar tag-list ,
1907which is either a single value or a list of values or ranges
1908specified in the same way as
1909.Ar ports .
1910Tags can be applied to the packet using
1911.Cm tag
1912rule action parameter (see it's description for details on tags).
1913.It Cm tcpack Ar ack
1914TCP packets only.
1915Match if the TCP header acknowledgment number field is set to
1916.Ar ack .
1917.It Cm tcpdatalen Ar tcpdatalen-list
1918Matches TCP packets whose length of TCP data is
1919.Ar tcpdatalen-list ,
1920which is either a single value or a list of values or ranges
1921specified in the same way as
1922.Ar ports .
1923.It Cm tcpflags Ar spec
1924TCP packets only.
1925Match if the TCP header contains the comma separated list of
1926flags specified in
1927.Ar spec .
1928The supported TCP flags are:
1929.Pp
1930.Cm fin ,
1931.Cm syn ,
1932.Cm rst ,
1933.Cm psh ,
1934.Cm ack
1935and
1936.Cm urg .
1937The absence of a particular flag may be denoted
1938with a
1939.Ql \&! .
1940A rule which contains a
1941.Cm tcpflags
1942specification can never match a fragmented packet which has
1943a non-zero offset.
1944See the
1945.Cm frag
1946option for details on matching fragmented packets.
1947.It Cm tcpseq Ar seq
1948TCP packets only.
1949Match if the TCP header sequence number field is set to
1950.Ar seq .
1951.It Cm tcpwin Ar tcpwin-list
1952Matches TCP packets whose  header window field is set to
1953.Ar tcpwin-list ,
1954which is either a single value or a list of values or ranges
1955specified in the same way as
1956.Ar ports .
1957.It Cm tcpoptions Ar spec
1958TCP packets only.
1959Match if the TCP header contains the comma separated list of
1960options specified in
1961.Ar spec .
1962The supported TCP options are:
1963.Pp
1964.Cm mss
1965(maximum segment size),
1966.Cm window
1967(tcp window advertisement),
1968.Cm sack
1969(selective ack),
1970.Cm ts
1971(rfc1323 timestamp) and
1972.Cm cc
1973(rfc1644 t/tcp connection count).
1974The absence of a particular option may be denoted
1975with a
1976.Ql \&! .
1977.It Cm uid Ar user
1978Match all TCP or UDP packets sent by or received for a
1979.Ar user .
1980A
1981.Ar user
1982may be matched by name or identification number.
1983.It Cm verrevpath
1984For incoming packets,
1985a routing table lookup is done on the packet's source address.
1986If the interface on which the packet entered the system matches the
1987outgoing interface for the route,
1988the packet matches.
1989If the interfaces do not match up,
1990the packet does not match.
1991All outgoing packets or packets with no incoming interface match.
1992.Pp
1993The name and functionality of the option is intentionally similar to
1994the Cisco IOS command:
1995.Pp
1996.Dl ip verify unicast reverse-path
1997.Pp
1998This option can be used to make anti-spoofing rules to reject all
1999packets with source addresses not from this interface.
2000See also the option
2001.Cm antispoof .
2002.It Cm versrcreach
2003For incoming packets,
2004a routing table lookup is done on the packet's source address.
2005If a route to the source address exists, but not the default route
2006or a blackhole/reject route, the packet matches.
2007Otherwise, the packet does not match.
2008All outgoing packets match.
2009.Pp
2010The name and functionality of the option is intentionally similar to
2011the Cisco IOS command:
2012.Pp
2013.Dl ip verify unicast source reachable-via any
2014.Pp
2015This option can be used to make anti-spoofing rules to reject all
2016packets whose source address is unreachable.
2017.It Cm antispoof
2018For incoming packets, the packet's source address is checked if it
2019belongs to a directly connected network.
2020If the network is directly connected, then the interface the packet
2021came on in is compared to the interface the network is connected to.
2022When incoming interface and directly connected interface are not the
2023same, the packet does not match.
2024Otherwise, the packet does match.
2025All outgoing packets match.
2026.Pp
2027This option can be used to make anti-spoofing rules to reject all
2028packets that pretend to be from a directly connected network but do
2029not come in through that interface.
2030This option is similar to but more restricted than
2031.Cm verrevpath
2032because it engages only on packets with source addresses of directly
2033connected networks instead of all source addresses.
2034.El
2035.Sh LOOKUP TABLES
2036Lookup tables are useful to handle large sparse sets of
2037addresses or other search keys (e.g., ports, jail IDs, interface names).
2038In the rest of this section we will use the term ``key''.
2039Table name needs to match the following spec:
2040.Ar table-name .
2041Tables with the same name can be created in different
2042.Ar sets .
2043However, rule links to the tables in
2044.Ar set 0
2045by default.
2046This behavior can be controlled by
2047.Va net.inet.ip.fw.tables_sets
2048variable.
2049See the
2050.Sx SETS OF RULES
2051section for more information.
2052There may be up to 65535 different lookup tables.
2053.Pp
2054The following table types are supported:
2055.Bl -tag -width indent
2056.It Ar table-type : Ar addr | iface | number | flow
2057.It Ar table-key : Ar addr Ns Oo / Ns Ar masklen Oc | iface-name | number | flow-spec
2058.It Ar flow-spec : Ar flow-field Ns Op , Ns Ar flow-spec
2059.It Ar flow-field : src-ip | proto | src-port | dst-ip | dst-port
2060.It Cm addr
2061matches IPv4 or IPv6 address.
2062Each entry is represented by an
2063.Ar addr Ns Op / Ns Ar masklen
2064and will match all addresses with base
2065.Ar addr
2066(specified as an IPv4/IPv6 address, or a hostname) and mask width of
2067.Ar masklen
2068bits.
2069If
2070.Ar masklen
2071is not specified, it defaults to 32 for IPv4 and 128 for IPv6.
2072When looking up an IP address in a table, the most specific
2073entry will match.
2074.It Cm iface
2075matches interface names.
2076Each entry is represented by string treated as interface name.
2077Wildcards are not supported.
2078.It Cm number
2079maches protocol ports, uids/gids or jail IDs.
2080Each entry is represented by 32-bit unsigned integer.
2081Ranges are not supported.
2082.It Cm flow
2083Matches packet fields specified by
2084.Ar flow
2085type suboptions with table entries.
2086.El
2087.Pp
2088Tables require explicit creation via
2089.Cm create
2090before use.
2091.Pp
2092The following creation options are supported:
2093.Bl -tag -width indent
2094.It Ar create-options : Ar create-option | create-options
2095.It Ar create-option : Cm type Ar table-type | Cm valtype Ar value-mask | Cm algo Ar algo-desc |
2096.Cm limit Ar number | Cm locked
2097.It Cm type
2098Table key type.
2099.It Cm valtype
2100Table value mask.
2101.It Cm algo
2102Table algorithm to use (see below).
2103.It Cm limit
2104Maximum number of items that may be inserted into table.
2105.It Cm locked
2106Restrict any table modifications.
2107.El
2108.Pp
2109Some of these options may be modified later via
2110.Cm modify
2111keyword.
2112The following options can be changed:
2113.Bl -tag -width indent
2114.It Ar modify-options : Ar modify-option | modify-options
2115.It Ar modify-option : Cm limit Ar number
2116.It Cm limit
2117Alter maximum number of items that may be inserted into table.
2118.El
2119.Pp
2120Additionally, table can be locked or unlocked using
2121.Cm lock
2122or
2123.Cm unlock
2124commands.
2125.Pp
2126Tables of the same
2127.Ar type
2128can be swapped with each other using
2129.Cm swap Ar name
2130command.
2131Swap may fail if tables limits are set and data exchange
2132would result in limits hit.
2133Operation is performed atomically.
2134.Pp
2135One or more entries can be added to a table at once using
2136.Cm add
2137command.
2138Addition of all items are performed atomically.
2139By default, error in addition of one entry does not influence
2140addition of other entries. However, non-zero error code is returned
2141in that case.
2142Special
2143.Cm atomic
2144keyword may be specified before
2145.Cm add
2146to indicate all-or-none add request.
2147.Pp
2148One or more entries can be removed from a table at once using
2149.Cm delete
2150command.
2151By default, error in removal of one entry does not influence
2152removing of other entries. However, non-zero error code is returned
2153in that case.
2154.Pp
2155It may be possible to check what entry will be found on particular
2156.Ar table-key
2157using
2158.Cm lookup
2159.Ar table-key
2160command.
2161This functionality is optional and may be unsupported in some algorithms.
2162.Pp
2163The following operations can be performed on
2164.Ar one
2165or
2166.Cm all
2167tables:
2168.Bl -tag -width indent
2169.It Cm list
2170List all entries.
2171.It Cm flush
2172Removes all entries.
2173.It Cm info
2174Shows generic table information.
2175.It Cm detail
2176Shows generic table information and algo-specific data.
2177.El
2178.Pp
2179The following lookup algorithms are supported:
2180.Bl -tag -width indent
2181.It Ar algo-desc : algo-name | "algo-name algo-data"
2182.It Ar algo-name: Ar addr:radix | addr:hash | iface:array | number:array | flow:hash
2183.It Cm addr:radix
2184Separate Radix trees for IPv4 and IPv6, the same way as the routing table (see
2185.Xr route 4 ) .
2186Default choice for
2187.Ar addr
2188type.
2189.It Cm addr:hash
2190Separate auto-growing hashes for IPv4 and IPv6.
2191Accepts entries with the same mask length specified initially via
2192.Cm "addr:hash masks=/v4,/v6"
2193algorithm creation options.
2194Assume /32 and /128 masks by default.
2195Search removes host bits (according to mask) from supplied address and checks
2196resulting key in appropriate hash.
2197Mostly optimized for /64 and byte-ranged IPv6 masks.
2198.It Cm iface:array
2199Array storing sorted indexes for entries which are presented in the system.
2200Optimized for very fast lookup.
2201.It Cm number:array
2202Array storing sorted u32 numbers.
2203.It Cm flow:hash
2204Auto-growing hash storing flow entries.
2205Search calculates hash on required packet fields and searches for matching
2206entries in selected bucket.
2207.El
2208.Pp
2209The
2210.Cm tablearg
2211feature provides the ability to use a value, looked up in the table, as
2212the argument for a rule action, action parameter or rule option.
2213This can significantly reduce number of rules in some configurations.
2214If two tables are used in a rule, the result of the second (destination)
2215is used.
2216.Pp
2217Each record may hold one or more values according to
2218.Ar value-mask .
2219This mask is set on table creation via
2220.Cm valtype
2221option.
2222The following value types are supported:
2223.Bl -tag -width indent
2224.It Ar value-mask : Ar value-type Ns Op , Ns Ar value-mask
2225.It Ar value-type : Ar skipto | pipe | fib | nat | dscp | tag | divert |
2226.Ar netgraph | limit | ipv4
2227.It Cm skipto
2228rule number to jump to.
2229.It Cm pipe
2230Pipe number to use.
2231.It Cm fib
2232fib number to match/set.
2233.It Cm nat
2234nat number to jump to.
2235.It Cm dscp
2236dscp value to match/set.
2237.It Cm tag
2238tag number to match/set.
2239.It Cm divert
2240port number to divert traffic to.
2241.It Cm netgraph
2242hook number to move packet to.
2243.It Cm limit
2244maximum number of connections.
2245.It Cm ipv4
2246IPv4 nexthop to fwd packets to.
2247.It Cm ipv6
2248IPv6 nexthop to fwd packets to.
2249.El
2250.Pp
2251The
2252.Cm tablearg
2253argument can be used with the following actions:
2254.Cm nat, pipe , queue, divert, tee, netgraph, ngtee, fwd, skipto, setfib,
2255action parameters:
2256.Cm tag, untag,
2257rule options:
2258.Cm limit, tagged.
2259.Pp
2260When used with the
2261.Cm skipto
2262action, the user should be aware that the code will walk the ruleset
2263up to a rule equal to, or past, the given number.
2264.Pp
2265See the
2266.Sx EXAMPLES
2267Section for example usage of tables and the tablearg keyword.
2268.Sh SETS OF RULES
2269Each rule or table belongs to one of 32 different
2270.Em sets
2271, numbered 0 to 31.
2272Set 31 is reserved for the default rule.
2273.Pp
2274By default, rules or tables are put in set 0, unless you use the
2275.Cm set N
2276attribute when adding a new rule or table.
2277Sets can be individually and atomically enabled or disabled,
2278so this mechanism permits an easy way to store multiple configurations
2279of the firewall and quickly (and atomically) switch between them.
2280.Pp
2281By default, tables from set 0 are referenced when adding rule with
2282table opcodes regardless of rule set.
2283This behavior can be changed by setting
2284.Va net.inet.ip.fw.tables_sets
2285variable to 1.
2286Rule's set will then be used for table references.
2287.Pp
2288The command to enable/disable sets is
2289.Bd -ragged -offset indent
2290.Nm
2291.Cm set Oo Cm disable Ar number ... Oc Op Cm enable Ar number ...
2292.Ed
2293.Pp
2294where multiple
2295.Cm enable
2296or
2297.Cm disable
2298sections can be specified.
2299Command execution is atomic on all the sets specified in the command.
2300By default, all sets are enabled.
2301.Pp
2302When you disable a set, its rules behave as if they do not exist
2303in the firewall configuration, with only one exception:
2304.Bd -ragged -offset indent
2305dynamic rules created from a rule before it had been disabled
2306will still be active until they expire.
2307In order to delete
2308dynamic rules you have to explicitly delete the parent rule
2309which generated them.
2310.Ed
2311.Pp
2312The set number of rules can be changed with the command
2313.Bd -ragged -offset indent
2314.Nm
2315.Cm set move
2316.Brq Cm rule Ar rule-number | old-set
2317.Cm to Ar new-set
2318.Ed
2319.Pp
2320Also, you can atomically swap two rulesets with the command
2321.Bd -ragged -offset indent
2322.Nm
2323.Cm set swap Ar first-set second-set
2324.Ed
2325.Pp
2326See the
2327.Sx EXAMPLES
2328Section on some possible uses of sets of rules.
2329.Sh STATEFUL FIREWALL
2330Stateful operation is a way for the firewall to dynamically
2331create rules for specific flows when packets that
2332match a given pattern are detected.
2333Support for stateful
2334operation comes through the
2335.Cm check-state , keep-state , record-state , limit
2336and
2337.Cm set-limit
2338options of
2339.Nm rules .
2340.Pp
2341Dynamic rules are created when a packet matches a
2342.Cm keep-state ,
2343.Cm record-state ,
2344.Cm limit
2345or
2346.Cm set-limit
2347rule, causing the creation of a
2348.Em dynamic
2349rule which will match all and only packets with
2350a given
2351.Em protocol
2352between a
2353.Em src-ip/src-port dst-ip/dst-port
2354pair of addresses
2355.Em ( src
2356and
2357.Em dst
2358are used here only to denote the initial match addresses, but they
2359are completely equivalent afterwards).
2360Rules created by
2361.Cm keep-state
2362option also have a
2363.Ar :flowname
2364taken from it.
2365This name is used in matching together with addresses, ports and protocol.
2366Dynamic rules will be checked at the first
2367.Cm check-state, keep-state
2368or
2369.Cm limit
2370occurrence, and the action performed upon a match will be the same
2371as in the parent rule.
2372.Pp
2373Note that no additional attributes other than protocol and IP addresses
2374and ports and :flowname are checked on dynamic rules.
2375.Pp
2376The typical use of dynamic rules is to keep a closed firewall configuration,
2377but let the first TCP SYN packet from the inside network install a
2378dynamic rule for the flow so that packets belonging to that session
2379will be allowed through the firewall:
2380.Pp
2381.Dl "ipfw add check-state :OUTBOUND"
2382.Dl "ipfw add allow tcp from my-subnet to any setup keep-state :OUTBOUND"
2383.Dl "ipfw add deny tcp from any to any"
2384.Pp
2385A similar approach can be used for UDP, where an UDP packet coming
2386from the inside will install a dynamic rule to let the response through
2387the firewall:
2388.Pp
2389.Dl "ipfw add check-state :OUTBOUND"
2390.Dl "ipfw add allow udp from my-subnet to any keep-state :OUTBOUND"
2391.Dl "ipfw add deny udp from any to any"
2392.Pp
2393Dynamic rules expire after some time, which depends on the status
2394of the flow and the setting of some
2395.Cm sysctl
2396variables.
2397See Section
2398.Sx SYSCTL VARIABLES
2399for more details.
2400For TCP sessions, dynamic rules can be instructed to periodically
2401send keepalive packets to refresh the state of the rule when it is
2402about to expire.
2403.Pp
2404See Section
2405.Sx EXAMPLES
2406for more examples on how to use dynamic rules.
2407.Sh TRAFFIC SHAPER (DUMMYNET) CONFIGURATION
2408.Nm
2409is also the user interface for the
2410.Nm dummynet
2411traffic shaper, packet scheduler and network emulator, a subsystem that
2412can artificially queue, delay or drop packets
2413emulating the behaviour of certain network links
2414or queueing systems.
2415.Pp
2416.Nm dummynet
2417operates by first using the firewall to select packets
2418using any match pattern that can be used in
2419.Nm
2420rules.
2421Matching packets are then passed to either of two
2422different objects, which implement the traffic regulation:
2423.Bl -hang -offset XXXX
2424.It Em pipe
2425A
2426.Em pipe
2427emulates a
2428.Em link
2429with given bandwidth and propagation delay,
2430driven by a FIFO scheduler and a single queue with programmable
2431queue size and packet loss rate.
2432Packets are appended to the queue as they come out from
2433.Nm ipfw ,
2434and then transferred in FIFO order to the link at the desired rate.
2435.It Em queue
2436A
2437.Em queue
2438is an abstraction used to implement packet scheduling
2439using one of several packet scheduling algorithms.
2440Packets sent to a
2441.Em queue
2442are first grouped into flows according to a mask on the 5-tuple.
2443Flows are then passed to the scheduler associated to the
2444.Em queue ,
2445and each flow uses scheduling parameters (weight and others)
2446as configured in the
2447.Em queue
2448itself.
2449A scheduler in turn is connected to an emulated link,
2450and arbitrates the link's bandwidth among backlogged flows according to
2451weights and to the features of the scheduling algorithm in use.
2452.El
2453.Pp
2454In practice,
2455.Em pipes
2456can be used to set hard limits to the bandwidth that a flow can use, whereas
2457.Em queues
2458can be used to determine how different flows share the available bandwidth.
2459.Pp
2460A graphical representation of the binding of queues,
2461flows, schedulers and links is below.
2462.Bd -literal -offset indent
2463                 (flow_mask|sched_mask)  sched_mask
2464         +---------+   weight Wx  +-------------+
2465         |         |->-[flow]-->--|             |-+
2466    -->--| QUEUE x |   ...        |             | |
2467         |         |->-[flow]-->--| SCHEDuler N | |
2468         +---------+              |             | |
2469             ...                  |             +--[LINK N]-->--
2470         +---------+   weight Wy  |             | +--[LINK N]-->--
2471         |         |->-[flow]-->--|             | |
2472    -->--| QUEUE y |   ...        |             | |
2473         |         |->-[flow]-->--|             | |
2474         +---------+              +-------------+ |
2475                                    +-------------+
2476.Ed
2477It is important to understand the role of the SCHED_MASK
2478and FLOW_MASK, which are configured through the commands
2479.Dl "ipfw sched N config mask SCHED_MASK ..."
2480and
2481.Dl "ipfw queue X config mask FLOW_MASK ..." .
2482.Pp
2483The SCHED_MASK is used to assign flows to one or more
2484scheduler instances, one for each
2485value of the packet's 5-tuple after applying SCHED_MASK.
2486As an example, using ``src-ip 0xffffff00'' creates one instance
2487for each /24 destination subnet.
2488.Pp
2489The FLOW_MASK, together with the SCHED_MASK, is used to split
2490packets into flows.
2491As an example, using
2492``src-ip 0x000000ff''
2493together with the previous SCHED_MASK makes a flow for
2494each individual source address.
2495In turn, flows for each /24
2496subnet will be sent to the same scheduler instance.
2497.Pp
2498The above diagram holds even for the
2499.Em pipe
2500case, with the only restriction that a
2501.Em pipe
2502only supports a SCHED_MASK, and forces the use of a FIFO
2503scheduler (these are for backward compatibility reasons;
2504in fact, internally, a
2505.Nm dummynet's
2506pipe is implemented exactly as above).
2507.Pp
2508There are two modes of
2509.Nm dummynet
2510operation:
2511.Dq normal
2512and
2513.Dq fast .
2514The
2515.Dq normal
2516mode tries to emulate a real link: the
2517.Nm dummynet
2518scheduler ensures that the packet will not leave the pipe faster than it
2519would on the real link with a given bandwidth.
2520The
2521.Dq fast
2522mode allows certain packets to bypass the
2523.Nm dummynet
2524scheduler (if packet flow does not exceed pipe's bandwidth).
2525This is the reason why the
2526.Dq fast
2527mode requires less CPU cycles per packet (on average) and packet latency
2528can be significantly lower in comparison to a real link with the same
2529bandwidth.
2530The default mode is
2531.Dq normal .
2532The
2533.Dq fast
2534mode can be enabled by setting the
2535.Va net.inet.ip.dummynet.io_fast
2536.Xr sysctl 8
2537variable to a non-zero value.
2538.Pp
2539.Ss PIPE, QUEUE AND SCHEDULER CONFIGURATION
2540The
2541.Em pipe ,
2542.Em queue
2543and
2544.Em scheduler
2545configuration commands are the following:
2546.Bd -ragged -offset indent
2547.Cm pipe Ar number Cm config Ar pipe-configuration
2548.Pp
2549.Cm queue Ar number Cm config Ar queue-configuration
2550.Pp
2551.Cm sched Ar number Cm config Ar sched-configuration
2552.Ed
2553.Pp
2554The following parameters can be configured for a pipe:
2555.Pp
2556.Bl -tag -width indent -compact
2557.It Cm bw Ar bandwidth | device
2558Bandwidth, measured in
2559.Sm off
2560.Op Cm K | M | G
2561.Brq Cm bit/s | Byte/s .
2562.Sm on
2563.Pp
2564A value of 0 (default) means unlimited bandwidth.
2565The unit must immediately follow the number, as in
2566.Pp
2567.Dl "ipfw pipe 1 config bw 300Kbit/s"
2568.Pp
2569If a device name is specified instead of a numeric value, as in
2570.Pp
2571.Dl "ipfw pipe 1 config bw tun0"
2572.Pp
2573then the transmit clock is supplied by the specified device.
2574At the moment only the
2575.Xr tun 4
2576device supports this
2577functionality, for use in conjunction with
2578.Xr ppp 8 .
2579.Pp
2580.It Cm delay Ar ms-delay
2581Propagation delay, measured in milliseconds.
2582The value is rounded to the next multiple of the clock tick
2583(typically 10ms, but it is a good practice to run kernels
2584with
2585.Dq "options HZ=1000"
2586to reduce
2587the granularity to 1ms or less).
2588The default value is 0, meaning no delay.
2589.Pp
2590.It Cm burst Ar size
2591If the data to be sent exceeds the pipe's bandwidth limit
2592(and the pipe was previously idle), up to
2593.Ar size
2594bytes of data are allowed to bypass the
2595.Nm dummynet
2596scheduler, and will be sent as fast as the physical link allows.
2597Any additional data will be transmitted at the rate specified
2598by the
2599.Nm pipe
2600bandwidth.
2601The burst size depends on how long the pipe has been idle;
2602the effective burst size is calculated as follows:
2603MAX(
2604.Ar size
2605,
2606.Nm bw
2607* pipe_idle_time).
2608.Pp
2609.It Cm profile Ar filename
2610A file specifying the additional overhead incurred in the transmission
2611of a packet on the link.
2612.Pp
2613Some link types introduce extra delays in the transmission
2614of a packet, e.g., because of MAC level framing, contention on
2615the use of the channel, MAC level retransmissions and so on.
2616From our point of view, the channel is effectively unavailable
2617for this extra time, which is constant or variable depending
2618on the link type.
2619Additionally, packets may be dropped after this
2620time (e.g., on a wireless link after too many retransmissions).
2621We can model the additional delay with an empirical curve
2622that represents its distribution.
2623.Bd -literal -offset indent
2624      cumulative probability
2625      1.0 ^
2626          |
2627      L   +-- loss-level          x
2628          |                 ******
2629          |                *
2630          |           *****
2631          |          *
2632          |        **
2633          |       *
2634          +-------*------------------->
2635                      delay
2636.Ed
2637The empirical curve may have both vertical and horizontal lines.
2638Vertical lines represent constant delay for a range of
2639probabilities.
2640Horizontal lines correspond to a discontinuity in the delay
2641distribution: the pipe will use the largest delay for a
2642given probability.
2643.Pp
2644The file format is the following, with whitespace acting as
2645a separator and '#' indicating the beginning a comment:
2646.Bl -tag -width indent
2647.It Cm name Ar identifier
2648optional name (listed by "ipfw pipe show")
2649to identify the delay distribution;
2650.It Cm bw Ar value
2651the bandwidth used for the pipe.
2652If not specified here, it must be present
2653explicitly as a configuration parameter for the pipe;
2654.It Cm loss-level Ar L
2655the probability above which packets are lost.
2656(0.0 <= L <= 1.0, default 1.0 i.e., no loss);
2657.It Cm samples Ar N
2658the number of samples used in the internal
2659representation of the curve (2..1024; default 100);
2660.It Cm "delay prob" | "prob delay"
2661One of these two lines is mandatory and defines
2662the format of the following lines with data points.
2663.It Ar XXX Ar YYY
26642 or more lines representing points in the curve,
2665with either delay or probability first, according
2666to the chosen format.
2667The unit for delay is milliseconds.
2668Data points do not need to be sorted.
2669Also, the number of actual lines can be different
2670from the value of the "samples" parameter:
2671.Nm
2672utility will sort and interpolate
2673the curve as needed.
2674.El
2675.Pp
2676Example of a profile file:
2677.Bd -literal -offset indent
2678name    bla_bla_bla
2679samples 100
2680loss-level    0.86
2681prob    delay
26820       200	# minimum overhead is 200ms
26830.5     200
26840.5     300
26850.8     1000
26860.9     1300
26871       1300
2688#configuration file end
2689.Ed
2690.El
2691.Pp
2692The following parameters can be configured for a queue:
2693.Pp
2694.Bl -tag -width indent -compact
2695.It Cm pipe Ar pipe_nr
2696Connects a queue to the specified pipe.
2697Multiple queues (with the same or different weights) can be connected to
2698the same pipe, which specifies the aggregate rate for the set of queues.
2699.Pp
2700.It Cm weight Ar weight
2701Specifies the weight to be used for flows matching this queue.
2702The weight must be in the range 1..100, and defaults to 1.
2703.El
2704.Pp
2705The following case-insensitive parameters can be configured for a
2706scheduler:
2707.Pp
2708.Bl -tag -width indent -compact
2709.It Cm type Ar {fifo | wf2q+ | rr | qfq}
2710specifies the scheduling algorithm to use.
2711.Bl -tag -width indent -compact
2712.It Cm fifo
2713is just a FIFO scheduler (which means that all packets
2714are stored in the same queue as they arrive to the scheduler).
2715FIFO has O(1) per-packet time complexity, with very low
2716constants (estimate 60-80ns on a 2GHz desktop machine)
2717but gives no service guarantees.
2718.It Cm wf2q+
2719implements the WF2Q+ algorithm, which is a Weighted Fair Queueing
2720algorithm which permits flows to share bandwidth according to
2721their weights.
2722Note that weights are not priorities; even a flow
2723with a minuscule weight will never starve.
2724WF2Q+ has O(log N) per-packet processing cost, where N is the number
2725of flows, and is the default algorithm used by previous versions
2726dummynet's queues.
2727.It Cm rr
2728implements the Deficit Round Robin algorithm, which has O(1) processing
2729costs (roughly, 100-150ns per packet)
2730and permits bandwidth allocation according to weights, but
2731with poor service guarantees.
2732.It Cm qfq
2733implements the QFQ algorithm, which is a very fast variant of
2734WF2Q+, with similar service guarantees and O(1) processing
2735costs (roughly, 200-250ns per packet).
2736.El
2737.El
2738.Pp
2739In addition to the type, all parameters allowed for a pipe can also
2740be specified for a scheduler.
2741.Pp
2742Finally, the following parameters can be configured for both
2743pipes and queues:
2744.Pp
2745.Bl -tag -width XXXX -compact
2746.It Cm buckets Ar hash-table-size
2747Specifies the size of the hash table used for storing the
2748various queues.
2749Default value is 64 controlled by the
2750.Xr sysctl 8
2751variable
2752.Va net.inet.ip.dummynet.hash_size ,
2753allowed range is 16 to 65536.
2754.Pp
2755.It Cm mask Ar mask-specifier
2756Packets sent to a given pipe or queue by an
2757.Nm
2758rule can be further classified into multiple flows, each of which is then
2759sent to a different
2760.Em dynamic
2761pipe or queue.
2762A flow identifier is constructed by masking the IP addresses,
2763ports and protocol types as specified with the
2764.Cm mask
2765options in the configuration of the pipe or queue.
2766For each different flow identifier, a new pipe or queue is created
2767with the same parameters as the original object, and matching packets
2768are sent to it.
2769.Pp
2770Thus, when
2771.Em dynamic pipes
2772are used, each flow will get the same bandwidth as defined by the pipe,
2773whereas when
2774.Em dynamic queues
2775are used, each flow will share the parent's pipe bandwidth evenly
2776with other flows generated by the same queue (note that other queues
2777with different weights might be connected to the same pipe).
2778.br
2779Available mask specifiers are a combination of one or more of the following:
2780.Pp
2781.Cm dst-ip Ar mask ,
2782.Cm dst-ip6 Ar mask ,
2783.Cm src-ip Ar mask ,
2784.Cm src-ip6 Ar mask ,
2785.Cm dst-port Ar mask ,
2786.Cm src-port Ar mask ,
2787.Cm flow-id Ar mask ,
2788.Cm proto Ar mask
2789or
2790.Cm all ,
2791.Pp
2792where the latter means all bits in all fields are significant.
2793.Pp
2794.It Cm noerror
2795When a packet is dropped by a
2796.Nm dummynet
2797queue or pipe, the error
2798is normally reported to the caller routine in the kernel, in the
2799same way as it happens when a device queue fills up.
2800Setting this
2801option reports the packet as successfully delivered, which can be
2802needed for some experimental setups where you want to simulate
2803loss or congestion at a remote router.
2804.Pp
2805.It Cm plr Ar packet-loss-rate
2806Packet loss rate.
2807Argument
2808.Ar packet-loss-rate
2809is a floating-point number between 0 and 1, with 0 meaning no
2810loss, 1 meaning 100% loss.
2811The loss rate is internally represented on 31 bits.
2812.Pp
2813.It Cm queue Brq Ar slots | size Ns Cm Kbytes
2814Queue size, in
2815.Ar slots
2816or
2817.Cm KBytes .
2818Default value is 50 slots, which
2819is the typical queue size for Ethernet devices.
2820Note that for slow speed links you should keep the queue
2821size short or your traffic might be affected by a significant
2822queueing delay.
2823E.g., 50 max-sized ethernet packets (1500 bytes) mean 600Kbit
2824or 20s of queue on a 30Kbit/s pipe.
2825Even worse effects can result if you get packets from an
2826interface with a much larger MTU, e.g.\& the loopback interface
2827with its 16KB packets.
2828The
2829.Xr sysctl 8
2830variables
2831.Em net.inet.ip.dummynet.pipe_byte_limit
2832and
2833.Em net.inet.ip.dummynet.pipe_slot_limit
2834control the maximum lengths that can be specified.
2835.Pp
2836.It Cm red | gred Ar w_q Ns / Ns Ar min_th Ns / Ns Ar max_th Ns / Ns Ar max_p
2837[ecn]
2838Make use of the RED (Random Early Detection) queue management algorithm.
2839.Ar w_q
2840and
2841.Ar max_p
2842are floating
2843point numbers between 0 and 1 (inclusive), while
2844.Ar min_th
2845and
2846.Ar max_th
2847are integer numbers specifying thresholds for queue management
2848(thresholds are computed in bytes if the queue has been defined
2849in bytes, in slots otherwise).
2850The two parameters can also be of the same value if needed. The
2851.Nm dummynet
2852also supports the gentle RED variant (gred) and ECN (Explicit Congestion
2853Notification) as optional. Three
2854.Xr sysctl 8
2855variables can be used to control the RED behaviour:
2856.Bl -tag -width indent
2857.It Va net.inet.ip.dummynet.red_lookup_depth
2858specifies the accuracy in computing the average queue
2859when the link is idle (defaults to 256, must be greater than zero)
2860.It Va net.inet.ip.dummynet.red_avg_pkt_size
2861specifies the expected average packet size (defaults to 512, must be
2862greater than zero)
2863.It Va net.inet.ip.dummynet.red_max_pkt_size
2864specifies the expected maximum packet size, only used when queue
2865thresholds are in bytes (defaults to 1500, must be greater than zero).
2866.El
2867.El
2868.Pp
2869When used with IPv6 data,
2870.Nm dummynet
2871currently has several limitations.
2872Information necessary to route link-local packets to an
2873interface is not available after processing by
2874.Nm dummynet
2875so those packets are dropped in the output path.
2876Care should be taken to ensure that link-local packets are not passed to
2877.Nm dummynet .
2878.Sh CHECKLIST
2879Here are some important points to consider when designing your
2880rules:
2881.Bl -bullet
2882.It
2883Remember that you filter both packets going
2884.Cm in
2885and
2886.Cm out .
2887Most connections need packets going in both directions.
2888.It
2889Remember to test very carefully.
2890It is a good idea to be near the console when doing this.
2891If you cannot be near the console,
2892use an auto-recovery script such as the one in
2893.Pa /usr/share/examples/ipfw/change_rules.sh .
2894.It
2895Do not forget the loopback interface.
2896.El
2897.Sh FINE POINTS
2898.Bl -bullet
2899.It
2900There are circumstances where fragmented datagrams are unconditionally
2901dropped.
2902TCP packets are dropped if they do not contain at least 20 bytes of
2903TCP header, UDP packets are dropped if they do not contain a full 8
2904byte UDP header, and ICMP packets are dropped if they do not contain
29054 bytes of ICMP header, enough to specify the ICMP type, code, and
2906checksum.
2907These packets are simply logged as
2908.Dq pullup failed
2909since there may not be enough good data in the packet to produce a
2910meaningful log entry.
2911.It
2912Another type of packet is unconditionally dropped, a TCP packet with a
2913fragment offset of one.
2914This is a valid packet, but it only has one use, to try
2915to circumvent firewalls.
2916When logging is enabled, these packets are
2917reported as being dropped by rule -1.
2918.It
2919If you are logged in over a network, loading the
2920.Xr kld 4
2921version of
2922.Nm
2923is probably not as straightforward as you would think.
2924The following command line is recommended:
2925.Bd -literal -offset indent
2926kldload ipfw && \e
2927ipfw add 32000 allow ip from any to any
2928.Ed
2929.Pp
2930Along the same lines, doing an
2931.Bd -literal -offset indent
2932ipfw flush
2933.Ed
2934.Pp
2935in similar surroundings is also a bad idea.
2936.It
2937The
2938.Nm
2939filter list may not be modified if the system security level
2940is set to 3 or higher
2941(see
2942.Xr init 8
2943for information on system security levels).
2944.El
2945.Sh PACKET DIVERSION
2946A
2947.Xr divert 4
2948socket bound to the specified port will receive all packets
2949diverted to that port.
2950If no socket is bound to the destination port, or if the divert module is
2951not loaded, or if the kernel was not compiled with divert socket support,
2952the packets are dropped.
2953.Sh NETWORK ADDRESS TRANSLATION (NAT)
2954.Nm
2955support in-kernel NAT using the kernel version of
2956.Xr libalias 3 .
2957The kernel module
2958.Cm ipfw_nat
2959should be loaded or kernel should have
2960.Cm options IPFIREWALL_NAT
2961to be able use NAT.
2962.Pp
2963The nat configuration command is the following:
2964.Bd -ragged -offset indent
2965.Bk -words
2966.Cm nat
2967.Ar nat_number
2968.Cm config
2969.Ar nat-configuration
2970.Ek
2971.Ed
2972.Pp
2973The following parameters can be configured:
2974.Bl -tag -width indent
2975.It Cm ip Ar ip_address
2976Define an ip address to use for aliasing.
2977.It Cm if Ar nic
2978Use ip address of NIC for aliasing, dynamically changing
2979it if NIC's ip address changes.
2980.It Cm log
2981Enable logging on this nat instance.
2982.It Cm deny_in
2983Deny any incoming connection from outside world.
2984.It Cm same_ports
2985Try to leave the alias port numbers unchanged from
2986the actual local port numbers.
2987.It Cm unreg_only
2988Traffic on the local network not originating from an
2989unregistered address spaces will be ignored.
2990.It Cm reset
2991Reset table of the packet aliasing engine on address change.
2992.It Cm reverse
2993Reverse the way libalias handles aliasing.
2994.It Cm proxy_only
2995Obey transparent proxy rules only, packet aliasing is not performed.
2996.It Cm skip_global
2997Skip instance in case of global state lookup (see below).
2998.El
2999.Pp
3000Some specials value can be supplied instead of
3001.Va nat_number:
3002.Bl -tag -width indent
3003.It Cm global
3004Looks up translation state in all configured nat instances.
3005If an entry is found, packet is aliased according to that entry.
3006If no entry was found in any of the instances, packet is passed unchanged,
3007and no new entry will be created.
3008See section
3009.Sx MULTIPLE INSTANCES
3010in
3011.Xr natd 8
3012for more information.
3013.It Cm tablearg
3014Uses argument supplied in lookup table.
3015See
3016.Sx LOOKUP TABLES
3017section below for more information on lookup tables.
3018.El
3019.Pp
3020To let the packet continue after being (de)aliased, set the sysctl variable
3021.Va net.inet.ip.fw.one_pass
3022to 0.
3023For more information about aliasing modes, refer to
3024.Xr libalias 3 .
3025See Section
3026.Sx EXAMPLES
3027for some examples about nat usage.
3028.Ss REDIRECT AND LSNAT SUPPORT IN IPFW
3029Redirect and LSNAT support follow closely the syntax used in
3030.Xr natd 8 .
3031See Section
3032.Sx EXAMPLES
3033for some examples on how to do redirect and lsnat.
3034.Ss SCTP NAT SUPPORT
3035SCTP nat can be configured in a similar manner to TCP through the
3036.Nm
3037command line tool.
3038The main difference is that
3039.Nm sctp nat
3040does not do port translation.
3041Since the local and global side ports will be the same,
3042there is no need to specify both.
3043Ports are redirected as follows:
3044.Bd -ragged -offset indent
3045.Bk -words
3046.Cm nat
3047.Ar nat_number
3048.Cm config if
3049.Ar nic
3050.Cm redirect_port sctp
3051.Ar ip_address [,addr_list] {[port | port-port] [,ports]}
3052.Ek
3053.Ed
3054.Pp
3055Most
3056.Nm sctp nat
3057configuration can be done in real-time through the
3058.Xr sysctl 8
3059interface.
3060All may be changed dynamically, though the hash_table size will only
3061change for new
3062.Nm nat
3063instances.
3064See
3065.Sx SYSCTL VARIABLES
3066for more info.
3067.Sh IPv6/IPv4 NETWORK ADDRESS AND PROTOCOL TRANSLATION
3068.Nm
3069supports in-kernel IPv6/IPv4 network address and protocol translation.
3070Stateful NAT64 translation allows IPv6-only clients to contact IPv4 servers
3071using unicast TCP, UDP or ICMP protocols.
3072One or more IPv4 addresses assigned to a stateful NAT64 translator are shared
3073among several IPv6-only clients.
3074When stateful NAT64 is used in conjunction with DNS64, no changes are usually
3075required in the IPv6 client or the IPv4 server.
3076The kernel module
3077.Cm ipfw_nat64
3078should be loaded or kernel should have
3079.Cm options IPFIREWALL_NAT64
3080to be able use stateful NAT64 translator.
3081.Pp
3082Stateful NAT64 uses a bunch of memory for several types of objects.
3083When IPv6 client initiates connection, NAT64 translator creates a host entry
3084in the states table.
3085Each host entry has a number of ports group entries allocated on demand.
3086Ports group entries contains connection state entries.
3087There are several options to control limits and lifetime for these objects.
3088.Pp
3089NAT64 translator follows RFC7915 when does ICMPv6/ICMP translation,
3090unsupported message types will be silently dropped.
3091IPv6 needs several ICMPv6 message types to be explicitly allowed for correct
3092operation.
3093Make sure that ND6 neighbor solicitation (ICMPv6 type 135) and neighbor
3094advertisement (ICMPv6 type 136) messages will not be handled by translation
3095rules.
3096.Pp
3097After translation NAT64 translator sends packets through corresponding netisr
3098queue.
3099Thus translator host should be configured as IPv4 and IPv6 router.
3100.Pp
3101The stateful NAT64 configuration command is the following:
3102.Bd -ragged -offset indent
3103.Bk -words
3104.Cm nat64lsn
3105.Ar name
3106.Cm create
3107.Ar create-options
3108.Ek
3109.Ed
3110.Pp
3111The following parameters can be configured:
3112.Bl -tag -width indent
3113.It Cm prefix4 Ar ipv4_prefix/plen
3114The IPv4 prefix with mask defines the pool of IPv4 addresses used as
3115source address after translation.
3116Stateful NAT64 module translates IPv6 source address of client to one
3117IPv4 address from this pool.
3118Note that incoming IPv4 packets that don't have corresponding state entry
3119in the states table will be dropped by translator.
3120Make sure that translation rules handle packets, destined to configured prefix.
3121.It Cm prefix6 Ar ipv6_prefix/length
3122The IPv6 prefix defines IPv4-embedded IPv6 addresses used by translator
3123to represent IPv4 addresses. This IPv6 prefix should be configured in DNS64.
3124The translator implementation follows RFC6052, that restricts the length of
3125prefixes to one of following: 32, 40, 48, 56, 64, or 96.
3126The Well-Known IPv6 Prefix 64:ff9b:: must be 96 bits long.
3127.It Cm max_ports Ar number
3128Maximum number of ports reserved for upper level protocols to one IPv6 client.
3129All reserved ports are divided into chunks between supported protocols.
3130The number of connections from one IPv6 client is limited by this option.
3131Note that closed TCP connections still remain in the list of connections until
3132.Cm tcp_close_age
3133interval will not expire.
3134Default value is
3135.Ar 2048 .
3136.It Cm host_del_age Ar seconds
3137The number of seconds until the host entry for a IPv6 client will be deleted
3138and all its resources will be released due to inactivity.
3139Default value is
3140.Ar 3600 .
3141.It Cm pg_del_age Ar seconds
3142The number of seconds until a ports group with unused state entries will
3143be released.
3144Default value is
3145.Ar 900 .
3146.It Cm tcp_syn_age Ar seconds
3147The number of seconds while a state entry for TCP connection with only SYN
3148sent will be kept.
3149If TCP connection establishing will not be finished,
3150state entry will be deleted.
3151Default value is
3152.Ar 10 .
3153.It Cm tcp_est_age Ar seconds
3154The number of seconds while a state entry for established TCP connection
3155will be kept.
3156Default value is
3157.Ar 7200 .
3158.It Cm tcp_close_age Ar seconds
3159The number of seconds while a state entry for closed TCP connection
3160will be kept.
3161Keeping state entries for closed connections is needed, because IPv4 servers
3162typically keep closed connections in a TIME_WAIT state for a several minutes.
3163Since translator's IPv4 addresses are shared among all IPv6 clients,
3164new connections from the same addresses and ports may be rejected by server,
3165because these connections are still in a TIME_WAIT state.
3166Keeping them in translator's state table protects from such rejects.
3167Default value is
3168.Ar 180 .
3169.It Cm udp_age Ar seconds
3170The number of seconds while translator keeps state entry in a waiting for
3171reply to the sent UDP datagram.
3172Default value is
3173.Ar 120 .
3174.It Cm icmp_age Ar seconds
3175The number of seconds while translator keeps state entry in a waiting for
3176reply to the sent ICMP message.
3177Default value is
3178.Ar 60 .
3179.It Cm log
3180Turn on logging of all handled packets via BPF through
3181.Ar ipfwlog0
3182interface.
3183.Ar ipfwlog0
3184is a pseudo interface and can be created after a boot manually with
3185.Cm ifconfig
3186command.
3187Note that it has different purpose than
3188.Ar ipfw0
3189interface.
3190Translators sends to BPF an additional information with each packet.
3191With
3192.Cm tcpdump
3193you are able to see each handled packet before and after translation.
3194.It Cm -log
3195Turn off logging of all handled packets via BPF.
3196.El
3197.Pp
3198To inspect a states table of stateful NAT64 the following command can be used:
3199.Bd -ragged -offset indent
3200.Bk -words
3201.Cm nat64lsn
3202.Ar name
3203.Cm show Cm states
3204.Ek
3205.Ed
3206.Pp
3207.Pp
3208Stateless NAT64 translator doesn't use a states table for translation
3209and converts IPv4 addresses to IPv6 and vice versa solely based on the
3210mappings taken from configured lookup tables.
3211Since a states table doesn't used by stateless translator,
3212it can be configured to pass IPv4 clients to IPv6-only servers.
3213.Pp
3214The stateless NAT64 configuration command is the following:
3215.Bd -ragged -offset indent
3216.Bk -words
3217.Cm nat64stl
3218.Ar name
3219.Cm create
3220.Ar create-options
3221.Ek
3222.Ed
3223.Pp
3224The following parameters can be configured:
3225.Bl -tag -width indent
3226.It Cm prefix6 Ar ipv6_prefix/length
3227The IPv6 prefix defines IPv4-embedded IPv6 addresses used by translator
3228to represent IPv4 addresses. This IPv6 prefix should be configured in DNS64.
3229.It Cm table4 Ar table46
3230The lookup table
3231.Ar table46
3232contains mapping how IPv4 addresses should be translated to IPv6 addresses.
3233.It Cm table6 Ar table64
3234The lookup table
3235.Ar table64
3236contains mapping how IPv6 addresses should be translated to IPv4 addresses.
3237.It Cm log
3238Turn on logging of all handled packets via BPF through
3239.Ar ipfwlog0
3240interface.
3241.It Cm -log
3242Turn off logging of all handled packets via BPF.
3243.El
3244.Pp
3245Note that the behavior of stateless translator with respect to not matched
3246packets differs from stateful translator.
3247If corresponding addresses was not found in the lookup tables, the packet
3248will not be dropped and the search continues.
3249.Sh IPv6-to-IPv6 NETWORK PREFIX TRANSLATION (NPTv6)
3250.Nm
3251supports in-kernel IPv6-to-IPv6 network prefix translation as described
3252in RFC6296.
3253The kernel module
3254.Cm ipfw_nptv6
3255should be loaded or kernel should has
3256.Cm options IPFIREWALL_NPTV6
3257to be able use NPTv6 translator.
3258.Pp
3259The NPTv6 configuration command is the following:
3260.Bd -ragged -offset indent
3261.Bk -words
3262.Cm nptv6
3263.Ar name
3264.Cm create
3265.Ar create-options
3266.Ek
3267.Ed
3268.Pp
3269The following parameters can be configured:
3270.Bl -tag -width indent
3271.It Cm int_prefix Ar ipv6_prefix
3272IPv6 prefix used in internal network.
3273NPTv6 module translates source address when it matches this prefix.
3274.It Cm ext_prefix Ar ipv6_prefix
3275IPv6 prefix used in external network.
3276NPTv6 module translates destination address when it matches this prefix.
3277.It Cm prefixlen Ar length
3278The length of specified IPv6 prefixes. It must be in range from 8 to 64.
3279.El
3280.Pp
3281Note that the prefix translation rules are silently ignored when IPv6 packet
3282forwarding is disabled.
3283To enable the packet forwarding, set the sysctl variable
3284.Va net.inet6.ip6.forwarding
3285to 1.
3286.Pp
3287To let the packet continue after being translated, set the sysctl variable
3288.Va net.inet.ip.fw.one_pass
3289to 0.
3290.Sh LOADER TUNABLES
3291Tunables can be set in
3292.Xr loader 8
3293prompt,
3294.Xr loader.conf 5
3295or
3296.Xr kenv 1
3297before ipfw module gets loaded.
3298.Bl -tag -width indent
3299.It Va net.inet.ip.fw.default_to_accept: No 0
3300Defines ipfw last rule behavior.
3301This value overrides
3302.Cd "options IPFW_DEFAULT_TO_(ACCEPT|DENY)"
3303from kernel configuration file.
3304.It Va net.inet.ip.fw.tables_max: No 128
3305Defines number of tables available in ipfw.
3306Number cannot exceed 65534.
3307.El
3308.Sh SYSCTL VARIABLES
3309A set of
3310.Xr sysctl 8
3311variables controls the behaviour of the firewall and
3312associated modules
3313.Pq Nm dummynet , bridge , sctp nat .
3314These are shown below together with their default value
3315(but always check with the
3316.Xr sysctl 8
3317command what value is actually in use) and meaning:
3318.Bl -tag -width indent
3319.It Va net.inet.ip.alias.sctp.accept_global_ootb_addip: No 0
3320Defines how the
3321.Nm nat
3322responds to receipt of global OOTB ASCONF-AddIP:
3323.Bl -tag -width indent
3324.It Cm 0
3325No response (unless a partially matching association exists -
3326ports and vtags match but global address does not)
3327.It Cm 1
3328.Nm nat
3329will accept and process all OOTB global AddIP messages.
3330.El
3331.Pp
3332Option 1 should never be selected as this forms a security risk.
3333An attacker can
3334establish multiple fake associations by sending AddIP messages.
3335.It Va net.inet.ip.alias.sctp.chunk_proc_limit: No 5
3336Defines the maximum number of chunks in an SCTP packet that will be
3337parsed for a
3338packet that matches an existing association.
3339This value is enforced to be greater or equal than
3340.Cm net.inet.ip.alias.sctp.initialising_chunk_proc_limit .
3341A high value is
3342a DoS risk yet setting too low a value may result in
3343important control chunks in
3344the packet not being located and parsed.
3345.It Va net.inet.ip.alias.sctp.error_on_ootb: No 1
3346Defines when the
3347.Nm nat
3348responds to any Out-of-the-Blue (OOTB) packets with ErrorM packets.
3349An OOTB packet is a packet that arrives with no existing association
3350registered in the
3351.Nm nat
3352and is not an INIT or ASCONF-AddIP packet:
3353.Bl -tag -width indent
3354.It Cm 0
3355ErrorM is never sent in response to OOTB packets.
3356.It Cm 1
3357ErrorM is only sent to OOTB packets received on the local side.
3358.It Cm 2
3359ErrorM is sent to the local side and on the global side ONLY if there is a
3360partial match (ports and vtags match but the source global IP does not).
3361This value is only useful if the
3362.Nm nat
3363is tracking global IP addresses.
3364.It Cm 3
3365ErrorM is sent in response to all OOTB packets on both
3366the local and global side
3367(DoS risk).
3368.El
3369.Pp
3370At the moment the default is 0, since the ErrorM packet is not yet
3371supported by most SCTP stacks.
3372When it is supported, and if not tracking
3373global addresses, we recommend setting this value to 1 to allow
3374multi-homed local hosts to function with the
3375.Nm nat .
3376To track global addresses, we recommend setting this value to 2 to
3377allow global hosts to be informed when they need to (re)send an
3378ASCONF-AddIP.
3379Value 3 should never be chosen (except for debugging) as the
3380.Nm nat
3381will respond to all OOTB global packets (a DoS risk).
3382.It Va net.inet.ip.alias.sctp.hashtable_size: No 2003
3383Size of hash tables used for
3384.Nm nat
3385lookups (100 < prime_number > 1000001).
3386This value sets the
3387.Nm hash table
3388size for any future created
3389.Nm nat
3390instance and therefore must be set prior to creating a
3391.Nm nat
3392instance.
3393The table sizes may be changed to suit specific needs.
3394If there will be few
3395concurrent associations, and memory is scarce, you may make these smaller.
3396If there will be many thousands (or millions) of concurrent associations, you
3397should make these larger.
3398A prime number is best for the table size.
3399The sysctl
3400update function will adjust your input value to the next highest prime number.
3401.It Va net.inet.ip.alias.sctp.holddown_time:  No 0
3402Hold association in table for this many seconds after receiving a
3403SHUTDOWN-COMPLETE.
3404This allows endpoints to correct shutdown gracefully if a
3405shutdown_complete is lost and retransmissions are required.
3406.It Va net.inet.ip.alias.sctp.init_timer: No 15
3407Timeout value while waiting for (INIT-ACK|AddIP-ACK).
3408This value cannot be 0.
3409.It Va net.inet.ip.alias.sctp.initialising_chunk_proc_limit: No 2
3410Defines the maximum number of chunks in an SCTP packet that will be parsed when
3411no existing association exists that matches that packet.
3412Ideally this packet
3413will only be an INIT or ASCONF-AddIP packet.
3414A higher value may become a DoS
3415risk as malformed packets can consume processing resources.
3416.It Va net.inet.ip.alias.sctp.param_proc_limit: No 25
3417Defines the maximum number of parameters within a chunk that will be
3418parsed in a
3419packet.
3420As for other similar sysctl variables, larger values pose a DoS risk.
3421.It Va net.inet.ip.alias.sctp.log_level: No 0
3422Level of detail in the system log messages (0 \- minimal, 1 \- event,
34232 \- info, 3 \- detail, 4 \- debug, 5 \- max debug).
3424May be a good
3425option in high loss environments.
3426.It Va net.inet.ip.alias.sctp.shutdown_time: No 15
3427Timeout value while waiting for SHUTDOWN-COMPLETE.
3428This value cannot be 0.
3429.It Va net.inet.ip.alias.sctp.track_global_addresses: No 0
3430Enables/disables global IP address tracking within the
3431.Nm nat
3432and places an
3433upper limit on the number of addresses tracked for each association:
3434.Bl -tag -width indent
3435.It Cm 0
3436Global tracking is disabled
3437.It Cm >1
3438Enables tracking, the maximum number of addresses tracked for each
3439association is limited to this value
3440.El
3441.Pp
3442This variable is fully dynamic, the new value will be adopted for all newly
3443arriving associations, existing associations are treated
3444as they were previously.
3445Global tracking will decrease the number of collisions within the
3446.Nm nat
3447at a cost
3448of increased processing load, memory usage, complexity, and possible
3449.Nm nat
3450state
3451problems in complex networks with multiple
3452.Nm nats .
3453We recommend not tracking
3454global IP addresses, this will still result in a fully functional
3455.Nm nat .
3456.It Va net.inet.ip.alias.sctp.up_timer: No 300
3457Timeout value to keep an association up with no traffic.
3458This value cannot be 0.
3459.It Va net.inet.ip.dummynet.expire : No 1
3460Lazily delete dynamic pipes/queue once they have no pending traffic.
3461You can disable this by setting the variable to 0, in which case
3462the pipes/queues will only be deleted when the threshold is reached.
3463.It Va net.inet.ip.dummynet.hash_size : No 64
3464Default size of the hash table used for dynamic pipes/queues.
3465This value is used when no
3466.Cm buckets
3467option is specified when configuring a pipe/queue.
3468.It Va net.inet.ip.dummynet.io_fast : No 0
3469If set to a non-zero value,
3470the
3471.Dq fast
3472mode of
3473.Nm dummynet
3474operation (see above) is enabled.
3475.It Va net.inet.ip.dummynet.io_pkt
3476Number of packets passed to
3477.Nm dummynet .
3478.It Va net.inet.ip.dummynet.io_pkt_drop
3479Number of packets dropped by
3480.Nm dummynet .
3481.It Va net.inet.ip.dummynet.io_pkt_fast
3482Number of packets bypassed by the
3483.Nm dummynet
3484scheduler.
3485.It Va net.inet.ip.dummynet.max_chain_len : No 16
3486Target value for the maximum number of pipes/queues in a hash bucket.
3487The product
3488.Cm max_chain_len*hash_size
3489is used to determine the threshold over which empty pipes/queues
3490will be expired even when
3491.Cm net.inet.ip.dummynet.expire=0 .
3492.It Va net.inet.ip.dummynet.red_lookup_depth : No 256
3493.It Va net.inet.ip.dummynet.red_avg_pkt_size : No 512
3494.It Va net.inet.ip.dummynet.red_max_pkt_size : No 1500
3495Parameters used in the computations of the drop probability
3496for the RED algorithm.
3497.It Va net.inet.ip.dummynet.pipe_byte_limit : No 1048576
3498.It Va net.inet.ip.dummynet.pipe_slot_limit : No 100
3499The maximum queue size that can be specified in bytes or packets.
3500These limits prevent accidental exhaustion of resources such as mbufs.
3501If you raise these limits,
3502you should make sure the system is configured so that sufficient resources
3503are available.
3504.It Va net.inet.ip.fw.autoinc_step : No 100
3505Delta between rule numbers when auto-generating them.
3506The value must be in the range 1..1000.
3507.It Va net.inet.ip.fw.curr_dyn_buckets : Va net.inet.ip.fw.dyn_buckets
3508The current number of buckets in the hash table for dynamic rules
3509(readonly).
3510.It Va net.inet.ip.fw.debug : No 1
3511Controls debugging messages produced by
3512.Nm .
3513.It Va net.inet.ip.fw.default_rule : No 65535
3514The default rule number (read-only).
3515By the design of
3516.Nm , the default rule is the last one, so its number
3517can also serve as the highest number allowed for a rule.
3518.It Va net.inet.ip.fw.dyn_buckets : No 256
3519The number of buckets in the hash table for dynamic rules.
3520Must be a power of 2, up to 65536.
3521It only takes effect when all dynamic rules have expired, so you
3522are advised to use a
3523.Cm flush
3524command to make sure that the hash table is resized.
3525.It Va net.inet.ip.fw.dyn_count : No 3
3526Current number of dynamic rules
3527(read-only).
3528.It Va net.inet.ip.fw.dyn_keepalive : No 1
3529Enables generation of keepalive packets for
3530.Cm keep-state
3531rules on TCP sessions.
3532A keepalive is generated to both
3533sides of the connection every 5 seconds for the last 20
3534seconds of the lifetime of the rule.
3535.It Va net.inet.ip.fw.dyn_max : No 8192
3536Maximum number of dynamic rules.
3537When you hit this limit, no more dynamic rules can be
3538installed until old ones expire.
3539.It Va net.inet.ip.fw.dyn_ack_lifetime : No 300
3540.It Va net.inet.ip.fw.dyn_syn_lifetime : No 20
3541.It Va net.inet.ip.fw.dyn_fin_lifetime : No 1
3542.It Va net.inet.ip.fw.dyn_rst_lifetime : No 1
3543.It Va net.inet.ip.fw.dyn_udp_lifetime : No 5
3544.It Va net.inet.ip.fw.dyn_short_lifetime : No 30
3545These variables control the lifetime, in seconds, of dynamic
3546rules.
3547Upon the initial SYN exchange the lifetime is kept short,
3548then increased after both SYN have been seen, then decreased
3549again during the final FIN exchange or when a RST is received.
3550Both
3551.Em dyn_fin_lifetime
3552and
3553.Em dyn_rst_lifetime
3554must be strictly lower than 5 seconds, the period of
3555repetition of keepalives.
3556The firewall enforces that.
3557.It Va net.inet.ip.fw.dyn_keep_states: No 0
3558Keep dynamic states on rule/set deletion.
3559States are relinked to default rule (65535).
3560This can be handly for ruleset reload.
3561Turned off by default.
3562.It Va net.inet.ip.fw.enable : No 1
3563Enables the firewall.
3564Setting this variable to 0 lets you run your machine without
3565firewall even if compiled in.
3566.It Va net.inet6.ip6.fw.enable : No 1
3567provides the same functionality as above for the IPv6 case.
3568.It Va net.inet.ip.fw.one_pass : No 1
3569When set, the packet exiting from the
3570.Nm dummynet
3571pipe or from
3572.Xr ng_ipfw 4
3573node is not passed though the firewall again.
3574Otherwise, after an action, the packet is
3575reinjected into the firewall at the next rule.
3576.It Va net.inet.ip.fw.tables_max : No 128
3577Maximum number of tables.
3578.It Va net.inet.ip.fw.verbose : No 1
3579Enables verbose messages.
3580.It Va net.inet.ip.fw.verbose_limit : No 0
3581Limits the number of messages produced by a verbose firewall.
3582.It Va net.inet6.ip6.fw.deny_unknown_exthdrs : No 1
3583If enabled packets with unknown IPv6 Extension Headers will be denied.
3584.It Va net.link.ether.ipfw : No 0
3585Controls whether layer-2 packets are passed to
3586.Nm .
3587Default is no.
3588.It Va net.link.bridge.ipfw : No 0
3589Controls whether bridged packets are passed to
3590.Nm .
3591Default is no.
3592.El
3593.Sh INTERNAL DIAGNOSTICS
3594There are some commands that may be useful to understand current state
3595of certain subsystems inside kernel module.
3596These commands provide debugging output which may change without notice.
3597.Pp
3598Currently the following commands are available as
3599.Cm internal
3600sub-options:
3601.Bl -tag -width indent
3602.It Cm iflist
3603Lists all interface which are currently tracked by
3604.Nm
3605with their in-kernel status.
3606.It Cm talist
3607List all table lookup algorithms currently available.
3608.El
3609.Sh EXAMPLES
3610There are far too many possible uses of
3611.Nm
3612so this Section will only give a small set of examples.
3613.Pp
3614.Ss BASIC PACKET FILTERING
3615This command adds an entry which denies all tcp packets from
3616.Em cracker.evil.org
3617to the telnet port of
3618.Em wolf.tambov.su
3619from being forwarded by the host:
3620.Pp
3621.Dl "ipfw add deny tcp from cracker.evil.org to wolf.tambov.su telnet"
3622.Pp
3623This one disallows any connection from the entire cracker's
3624network to my host:
3625.Pp
3626.Dl "ipfw add deny ip from 123.45.67.0/24 to my.host.org"
3627.Pp
3628A first and efficient way to limit access (not using dynamic rules)
3629is the use of the following rules:
3630.Pp
3631.Dl "ipfw add allow tcp from any to any established"
3632.Dl "ipfw add allow tcp from net1 portlist1 to net2 portlist2 setup"
3633.Dl "ipfw add allow tcp from net3 portlist3 to net3 portlist3 setup"
3634.Dl "..."
3635.Dl "ipfw add deny tcp from any to any"
3636.Pp
3637The first rule will be a quick match for normal TCP packets,
3638but it will not match the initial SYN packet, which will be
3639matched by the
3640.Cm setup
3641rules only for selected source/destination pairs.
3642All other SYN packets will be rejected by the final
3643.Cm deny
3644rule.
3645.Pp
3646If you administer one or more subnets, you can take advantage
3647of the address sets and or-blocks and write extremely
3648compact rulesets which selectively enable services to blocks
3649of clients, as below:
3650.Pp
3651.Dl "goodguys=\*q{ 10.1.2.0/24{20,35,66,18} or 10.2.3.0/28{6,3,11} }\*q"
3652.Dl "badguys=\*q10.1.2.0/24{8,38,60}\*q"
3653.Dl ""
3654.Dl "ipfw add allow ip from ${goodguys} to any"
3655.Dl "ipfw add deny ip from ${badguys} to any"
3656.Dl "... normal policies ..."
3657.Pp
3658The
3659.Cm verrevpath
3660option could be used to do automated anti-spoofing by adding the
3661following to the top of a ruleset:
3662.Pp
3663.Dl "ipfw add deny ip from any to any not verrevpath in"
3664.Pp
3665This rule drops all incoming packets that appear to be coming to the
3666system on the wrong interface.
3667For example, a packet with a source
3668address belonging to a host on a protected internal network would be
3669dropped if it tried to enter the system from an external interface.
3670.Pp
3671The
3672.Cm antispoof
3673option could be used to do similar but more restricted anti-spoofing
3674by adding the following to the top of a ruleset:
3675.Pp
3676.Dl "ipfw add deny ip from any to any not antispoof in"
3677.Pp
3678This rule drops all incoming packets that appear to be coming from another
3679directly connected system but on the wrong interface.
3680For example, a packet with a source address of
3681.Li 192.168.0.0/24 ,
3682configured on
3683.Li fxp0 ,
3684but coming in on
3685.Li fxp1
3686would be dropped.
3687.Pp
3688The
3689.Cm setdscp
3690option could be used to (re)mark user traffic,
3691by adding the following to the appropriate place in ruleset:
3692.Pp
3693.Dl "ipfw add setdscp be ip from any to any dscp af11,af21"
3694.Ss DYNAMIC RULES
3695In order to protect a site from flood attacks involving fake
3696TCP packets, it is safer to use dynamic rules:
3697.Pp
3698.Dl "ipfw add check-state"
3699.Dl "ipfw add deny tcp from any to any established"
3700.Dl "ipfw add allow tcp from my-net to any setup keep-state"
3701.Pp
3702This will let the firewall install dynamic rules only for
3703those connection which start with a regular SYN packet coming
3704from the inside of our network.
3705Dynamic rules are checked when encountering the first
3706occurrence of a
3707.Cm check-state ,
3708.Cm keep-state
3709or
3710.Cm limit
3711rule.
3712A
3713.Cm check-state
3714rule should usually be placed near the beginning of the
3715ruleset to minimize the amount of work scanning the ruleset.
3716Your mileage may vary.
3717.Pp
3718For more complex scenarios with dynamic rules
3719.Cm record-state
3720and
3721.Cm defer-action
3722can be used to precisely control creation and checking of dynamic rules.
3723Example of usage of these options are provided in
3724.Sx NETWORK ADDRESS TRANSLATION (NAT)
3725Section.
3726.Pp
3727To limit the number of connections a user can open
3728you can use the following type of rules:
3729.Pp
3730.Dl "ipfw add allow tcp from my-net/24 to any setup limit src-addr 10"
3731.Dl "ipfw add allow tcp from any to me setup limit src-addr 4"
3732.Pp
3733The former (assuming it runs on a gateway) will allow each host
3734on a /24 network to open at most 10 TCP connections.
3735The latter can be placed on a server to make sure that a single
3736client does not use more than 4 simultaneous connections.
3737.Pp
3738.Em BEWARE :
3739stateful rules can be subject to denial-of-service attacks
3740by a SYN-flood which opens a huge number of dynamic rules.
3741The effects of such attacks can be partially limited by
3742acting on a set of
3743.Xr sysctl 8
3744variables which control the operation of the firewall.
3745.Pp
3746Here is a good usage of the
3747.Cm list
3748command to see accounting records and timestamp information:
3749.Pp
3750.Dl ipfw -at list
3751.Pp
3752or in short form without timestamps:
3753.Pp
3754.Dl ipfw -a list
3755.Pp
3756which is equivalent to:
3757.Pp
3758.Dl ipfw show
3759.Pp
3760Next rule diverts all incoming packets from 192.168.2.0/24
3761to divert port 5000:
3762.Pp
3763.Dl ipfw divert 5000 ip from 192.168.2.0/24 to any in
3764.Ss TRAFFIC SHAPING
3765The following rules show some of the applications of
3766.Nm
3767and
3768.Nm dummynet
3769for simulations and the like.
3770.Pp
3771This rule drops random incoming packets with a probability
3772of 5%:
3773.Pp
3774.Dl "ipfw add prob 0.05 deny ip from any to any in"
3775.Pp
3776A similar effect can be achieved making use of
3777.Nm dummynet
3778pipes:
3779.Pp
3780.Dl "ipfw add pipe 10 ip from any to any"
3781.Dl "ipfw pipe 10 config plr 0.05"
3782.Pp
3783We can use pipes to artificially limit bandwidth, e.g.\& on a
3784machine acting as a router, if we want to limit traffic from
3785local clients on 192.168.2.0/24 we do:
3786.Pp
3787.Dl "ipfw add pipe 1 ip from 192.168.2.0/24 to any out"
3788.Dl "ipfw pipe 1 config bw 300Kbit/s queue 50KBytes"
3789.Pp
3790note that we use the
3791.Cm out
3792modifier so that the rule is not used twice.
3793Remember in fact that
3794.Nm
3795rules are checked both on incoming and outgoing packets.
3796.Pp
3797Should we want to simulate a bidirectional link with bandwidth
3798limitations, the correct way is the following:
3799.Pp
3800.Dl "ipfw add pipe 1 ip from any to any out"
3801.Dl "ipfw add pipe 2 ip from any to any in"
3802.Dl "ipfw pipe 1 config bw 64Kbit/s queue 10Kbytes"
3803.Dl "ipfw pipe 2 config bw 64Kbit/s queue 10Kbytes"
3804.Pp
3805The above can be very useful, e.g.\& if you want to see how
3806your fancy Web page will look for a residential user who
3807is connected only through a slow link.
3808You should not use only one pipe for both directions, unless
3809you want to simulate a half-duplex medium (e.g.\& AppleTalk,
3810Ethernet, IRDA).
3811It is not necessary that both pipes have the same configuration,
3812so we can also simulate asymmetric links.
3813.Pp
3814Should we want to verify network performance with the RED queue
3815management algorithm:
3816.Pp
3817.Dl "ipfw add pipe 1 ip from any to any"
3818.Dl "ipfw pipe 1 config bw 500Kbit/s queue 100 red 0.002/30/80/0.1"
3819.Pp
3820Another typical application of the traffic shaper is to
3821introduce some delay in the communication.
3822This can significantly affect applications which do a lot of Remote
3823Procedure Calls, and where the round-trip-time of the
3824connection often becomes a limiting factor much more than
3825bandwidth:
3826.Pp
3827.Dl "ipfw add pipe 1 ip from any to any out"
3828.Dl "ipfw add pipe 2 ip from any to any in"
3829.Dl "ipfw pipe 1 config delay 250ms bw 1Mbit/s"
3830.Dl "ipfw pipe 2 config delay 250ms bw 1Mbit/s"
3831.Pp
3832Per-flow queueing can be useful for a variety of purposes.
3833A very simple one is counting traffic:
3834.Pp
3835.Dl "ipfw add pipe 1 tcp from any to any"
3836.Dl "ipfw add pipe 1 udp from any to any"
3837.Dl "ipfw add pipe 1 ip from any to any"
3838.Dl "ipfw pipe 1 config mask all"
3839.Pp
3840The above set of rules will create queues (and collect
3841statistics) for all traffic.
3842Because the pipes have no limitations, the only effect is
3843collecting statistics.
3844Note that we need 3 rules, not just the last one, because
3845when
3846.Nm
3847tries to match IP packets it will not consider ports, so we
3848would not see connections on separate ports as different
3849ones.
3850.Pp
3851A more sophisticated example is limiting the outbound traffic
3852on a net with per-host limits, rather than per-network limits:
3853.Pp
3854.Dl "ipfw add pipe 1 ip from 192.168.2.0/24 to any out"
3855.Dl "ipfw add pipe 2 ip from any to 192.168.2.0/24 in"
3856.Dl "ipfw pipe 1 config mask src-ip 0x000000ff bw 200Kbit/s queue 20Kbytes"
3857.Dl "ipfw pipe 2 config mask dst-ip 0x000000ff bw 200Kbit/s queue 20Kbytes"
3858.Ss LOOKUP TABLES
3859In the following example, we need to create several traffic bandwidth
3860classes and we need different hosts/networks to fall into different classes.
3861We create one pipe for each class and configure them accordingly.
3862Then we create a single table and fill it with IP subnets and addresses.
3863For each subnet/host we set the argument equal to the number of the pipe
3864that it should use.
3865Then we classify traffic using a single rule:
3866.Pp
3867.Dl "ipfw pipe 1 config bw 1000Kbyte/s"
3868.Dl "ipfw pipe 4 config bw 4000Kbyte/s"
3869.Dl "..."
3870.Dl "ipfw table T1 create type addr"
3871.Dl "ipfw table T1 add 192.168.2.0/24 1"
3872.Dl "ipfw table T1 add 192.168.0.0/27 4"
3873.Dl "ipfw table T1 add 192.168.0.2 1"
3874.Dl "..."
3875.Dl "ipfw add pipe tablearg ip from 'table(T1)' to any"
3876.Pp
3877Using the
3878.Cm fwd
3879action, the table entries may include hostnames and IP addresses.
3880.Pp
3881.Dl "ipfw table T2 create type addr ftype ip"
3882.Dl "ipfw table T2 add 192.168.2.0/24 10.23.2.1"
3883.Dl "ipfw table T21 add 192.168.0.0/27 router1.dmz"
3884.Dl "..."
3885.Dl "ipfw add 100 fwd tablearg ip from any to table(1)"
3886.Pp
3887In the following example per-interface firewall is created:
3888.Pp
3889.Dl "ipfw table IN create type iface valtype skipto,fib"
3890.Dl "ipfw table IN add vlan20 12000,12"
3891.Dl "ipfw table IN add vlan30 13000,13"
3892.Dl "ipfw table OUT create type iface valtype skipto"
3893.Dl "ipfw table OUT add vlan20 22000"
3894.Dl "ipfw table OUT add vlan30 23000"
3895.Dl ".."
3896.Dl "ipfw add 100 setfib tablearg ip from any to any recv 'table(IN)' in"
3897.Dl "ipfw add 200 skipto tablearg ip from any to any recv 'table(IN)' in"
3898.Dl "ipfw add 300 skipto tablearg ip from any to any xmit 'table(OUT)' out"
3899.Pp
3900The following example illustrate usage of flow tables:
3901.Pp
3902.Dl "ipfw table fl create type flow:flow:src-ip,proto,dst-ip,dst-port"
3903.Dl "ipfw table fl add 2a02:6b8:77::88,tcp,2a02:6b8:77::99,80 11"
3904.Dl "ipfw table fl add 10.0.0.1,udp,10.0.0.2,53 12"
3905.Dl ".."
3906.Dl "ipfw add 100 allow ip from any to any flow 'table(fl,11)' recv ix0"
3907.Ss SETS OF RULES
3908To add a set of rules atomically, e.g.\& set 18:
3909.Pp
3910.Dl "ipfw set disable 18"
3911.Dl "ipfw add NN set 18 ...         # repeat as needed"
3912.Dl "ipfw set enable 18"
3913.Pp
3914To delete a set of rules atomically the command is simply:
3915.Pp
3916.Dl "ipfw delete set 18"
3917.Pp
3918To test a ruleset and disable it and regain control if something goes wrong:
3919.Pp
3920.Dl "ipfw set disable 18"
3921.Dl "ipfw add NN set 18 ...         # repeat as needed"
3922.Dl "ipfw set enable 18; echo done; sleep 30 && ipfw set disable 18"
3923.Pp
3924Here if everything goes well, you press control-C before the "sleep"
3925terminates, and your ruleset will be left active.
3926Otherwise, e.g.\& if
3927you cannot access your box, the ruleset will be disabled after
3928the sleep terminates thus restoring the previous situation.
3929.Pp
3930To show rules of the specific set:
3931.Pp
3932.Dl "ipfw set 18 show"
3933.Pp
3934To show rules of the disabled set:
3935.Pp
3936.Dl "ipfw -S set 18 show"
3937.Pp
3938To clear a specific rule counters of the specific set:
3939.Pp
3940.Dl "ipfw set 18 zero NN"
3941.Pp
3942To delete a specific rule of the specific set:
3943.Pp
3944.Dl "ipfw set 18 delete NN"
3945.Ss NAT, REDIRECT AND LSNAT
3946First redirect all the traffic to nat instance 123:
3947.Pp
3948.Dl "ipfw add nat 123 all from any to any"
3949.Pp
3950Then to configure nat instance 123 to alias all the outgoing traffic with ip
3951192.168.0.123, blocking all incoming connections, trying to keep
3952same ports on both sides, clearing aliasing table on address change
3953and keeping a log of traffic/link statistics:
3954.Pp
3955.Dl "ipfw nat 123 config ip 192.168.0.123 log deny_in reset same_ports"
3956.Pp
3957Or to change address of instance 123, aliasing table will be cleared (see
3958reset option):
3959.Pp
3960.Dl "ipfw nat 123 config ip 10.0.0.1"
3961.Pp
3962To see configuration of nat instance 123:
3963.Pp
3964.Dl "ipfw nat 123 show config"
3965.Pp
3966To show logs of all the instances in range 111-999:
3967.Pp
3968.Dl "ipfw nat 111-999 show"
3969.Pp
3970To see configurations of all instances:
3971.Pp
3972.Dl "ipfw nat show config"
3973.Pp
3974Or a redirect rule with mixed modes could looks like:
3975.Pp
3976.Dl "ipfw nat 123 config redirect_addr 10.0.0.1 10.0.0.66"
3977.Dl "			 redirect_port tcp 192.168.0.1:80 500"
3978.Dl "			 redirect_proto udp 192.168.1.43 192.168.1.1"
3979.Dl "			 redirect_addr 192.168.0.10,192.168.0.11"
3980.Dl "			 	    10.0.0.100	# LSNAT"
3981.Dl "			 redirect_port tcp 192.168.0.1:80,192.168.0.10:22"
3982.Dl "			 	    500		# LSNAT"
3983.Pp
3984or it could be split in:
3985.Pp
3986.Dl "ipfw nat 1 config redirect_addr 10.0.0.1 10.0.0.66"
3987.Dl "ipfw nat 2 config redirect_port tcp 192.168.0.1:80 500"
3988.Dl "ipfw nat 3 config redirect_proto udp 192.168.1.43 192.168.1.1"
3989.Dl "ipfw nat 4 config redirect_addr 192.168.0.10,192.168.0.11,192.168.0.12"
3990.Dl "				         10.0.0.100"
3991.Dl "ipfw nat 5 config redirect_port tcp"
3992.Dl "			192.168.0.1:80,192.168.0.10:22,192.168.0.20:25 500"
3993.Pp
3994Sometimes you may want to mix NAT and dynamic rules. It could be achived with
3995.Cm record-state
3996and
3997.Cm defer-action
3998options. Problem is, you need to create dynamic rule before NAT and check it
3999after NAT actions (or vice versa) to have consistent addresses and ports.
4000Rule with
4001.Cm keep-state
4002option will trigger activation of existing dynamic state, and action of such
4003rule will be performed as soon as rule is matched. In case of NAT and
4004.Cm allow
4005rule packet need to be passed to NAT, not allowed as soon is possible.
4006.Pp
4007There is example of set of rules to achive this. Bear in mind that this
4008is exmaple only and it is not very usefult by itself.
4009.Pp
4010On way out, after all checks place this rules:
4011.Pp
4012.Dl "ipfw add allow record-state skip-action"
4013.Dl "ipfw add nat 1"
4014.Pp
4015And on way in there should be something like this:
4016.Pp
4017.Dl "ipfw add nat 1"
4018.Dl "ipfw add check-state"
4019.Pp
4020Please note, that first rule on way out doesn't allow packet and doesn't
4021execute existing dynamic rules. All it does, create new dynamic rule with
4022.Cm allow
4023action, if it is not created yet. Later, this dynamic rule is used on way
4024in by
4025.Cm check-state
4026rule.
4027.Sh SEE ALSO
4028.Xr cpp 1 ,
4029.Xr m4 1 ,
4030.Xr altq 4 ,
4031.Xr divert 4 ,
4032.Xr dummynet 4 ,
4033.Xr if_bridge 4 ,
4034.Xr ip 4 ,
4035.Xr ipfirewall 4 ,
4036.Xr ng_ipfw 4 ,
4037.Xr protocols 5 ,
4038.Xr services 5 ,
4039.Xr init 8 ,
4040.Xr kldload 8 ,
4041.Xr reboot 8 ,
4042.Xr sysctl 8 ,
4043.Xr syslogd 8
4044.Sh HISTORY
4045The
4046.Nm
4047utility first appeared in
4048.Fx 2.0 .
4049.Nm dummynet
4050was introduced in
4051.Fx 2.2.8 .
4052Stateful extensions were introduced in
4053.Fx 4.0 .
4054.Nm ipfw2
4055was introduced in Summer 2002.
4056.Sh AUTHORS
4057.An Ugen J. S. Antsilevich ,
4058.An Poul-Henning Kamp ,
4059.An Alex Nash ,
4060.An Archie Cobbs ,
4061.An Luigi Rizzo .
4062.Pp
4063.An -nosplit
4064API based upon code written by
4065.An Daniel Boulet
4066for BSDI.
4067.Pp
4068Dummynet has been introduced by Luigi Rizzo in 1997-1998.
4069.Pp
4070Some early work (1999-2000) on the
4071.Nm dummynet
4072traffic shaper supported by Akamba Corp.
4073.Pp
4074The ipfw core (ipfw2) has been completely redesigned and
4075reimplemented by Luigi Rizzo in summer 2002.
4076Further
4077actions and
4078options have been added by various developer over the years.
4079.Pp
4080.An -nosplit
4081In-kernel NAT support written by
4082.An Paolo Pisati Aq Mt piso@FreeBSD.org
4083as part of a Summer of Code 2005 project.
4084.Pp
4085SCTP
4086.Nm nat
4087support has been developed by
4088.An The Centre for Advanced Internet Architectures (CAIA) Aq http://www.caia.swin.edu.au .
4089The primary developers and maintainers are David Hayes and Jason But.
4090For further information visit:
4091.Aq http://www.caia.swin.edu.au/urp/SONATA
4092.Pp
4093Delay profiles have been developed by Alessandro Cerri and
4094Luigi Rizzo, supported by the
4095European Commission within Projects Onelab and Onelab2.
4096.Sh BUGS
4097The syntax has grown over the years and sometimes it might be confusing.
4098Unfortunately, backward compatibility prevents cleaning up mistakes
4099made in the definition of the syntax.
4100.Pp
4101.Em !!! WARNING !!!
4102.Pp
4103Misconfiguring the firewall can put your computer in an unusable state,
4104possibly shutting down network services and requiring console access to
4105regain control of it.
4106.Pp
4107Incoming packet fragments diverted by
4108.Cm divert
4109are reassembled before delivery to the socket.
4110The action used on those packet is the one from the
4111rule which matches the first fragment of the packet.
4112.Pp
4113Packets diverted to userland, and then reinserted by a userland process
4114may lose various packet attributes.
4115The packet source interface name
4116will be preserved if it is shorter than 8 bytes and the userland process
4117saves and reuses the sockaddr_in
4118(as does
4119.Xr natd 8 ) ;
4120otherwise, it may be lost.
4121If a packet is reinserted in this manner, later rules may be incorrectly
4122applied, making the order of
4123.Cm divert
4124rules in the rule sequence very important.
4125.Pp
4126Dummynet drops all packets with IPv6 link-local addresses.
4127.Pp
4128Rules using
4129.Cm uid
4130or
4131.Cm gid
4132may not behave as expected.
4133In particular, incoming SYN packets may
4134have no uid or gid associated with them since they do not yet belong
4135to a TCP connection, and the uid/gid associated with a packet may not
4136be as expected if the associated process calls
4137.Xr setuid 2
4138or similar system calls.
4139.Pp
4140Rule syntax is subject to the command line environment and some patterns
4141may need to be escaped with the backslash character
4142or quoted appropriately.
4143.Pp
4144Due to the architecture of
4145.Xr libalias 3 ,
4146ipfw nat is not compatible with the TCP segmentation offloading (TSO).
4147Thus, to reliably nat your network traffic, please disable TSO
4148on your NICs using
4149.Xr ifconfig 8 .
4150.Pp
4151ICMP error messages are not implicitly matched by dynamic rules
4152for the respective conversations.
4153To avoid failures of network error detection and path MTU discovery,
4154ICMP error messages may need to be allowed explicitly through static
4155rules.
4156.Pp
4157Rules using
4158.Cm call
4159and
4160.Cm return
4161actions may lead to confusing behaviour if ruleset has mistakes,
4162and/or interaction with other subsystems (netgraph, dummynet, etc.) is used.
4163One possible case for this is packet leaving
4164.Nm
4165in subroutine on the input pass, while later on output encountering unpaired
4166.Cm return
4167first.
4168As the call stack is kept intact after input pass, packet will suddenly
4169return to the rule number used on input pass, not on output one.
4170Order of processing should be checked carefully to avoid such mistakes.
4171