xref: /freebsd/sbin/ipfw/ipfw.8 (revision 809a8352dd5bb83dca991ff6c82e509208afad26)
1.\"
2.\" $FreeBSD$
3.\"
4.Dd March 1, 2019
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.Ss STATEFUL IPv6/IPv4 NETWORK ADDRESS AND PROTOCOL TRANSLATION
109.Nm
110.Oo Cm set Ar N Oc Cm nat64lsn Ar name Cm create Ar create-options
111.Nm
112.Oo Cm set Ar N Oc Cm nat64lsn Ar name Cm config Ar config-options
113.Nm
114.Oo Cm set Ar N Oc Cm nat64lsn
115.Brq Ar name | all
116.Brq Cm list | show
117.Op Cm states
118.Nm
119.Oo Cm set Ar N Oc Cm nat64lsn
120.Brq Ar name | all
121.Cm destroy
122.Nm
123.Oo Cm set Ar N Oc Cm nat64lsn Ar name Cm stats Op Cm reset
124.Ss STATELESS IPv6/IPv4 NETWORK ADDRESS AND PROTOCOL TRANSLATION
125.Nm
126.Oo Cm set Ar N Oc Cm nat64stl Ar name Cm create Ar create-options
127.Nm
128.Oo Cm set Ar N Oc Cm nat64stl Ar name Cm config Ar config-options
129.Nm
130.Oo Cm set Ar N Oc Cm nat64stl
131.Brq Ar name | all
132.Brq Cm list | show
133.Nm
134.Oo Cm set Ar N Oc Cm nat64stl
135.Brq Ar name | all
136.Cm destroy
137.Nm
138.Oo Cm set Ar N Oc Cm nat64stl Ar name Cm stats Op Cm reset
139.Ss IPv6-to-IPv6 NETWORK PREFIX TRANSLATION
140.Nm
141.Oo Cm set Ar N Oc Cm nptv6 Ar name Cm create Ar create-options
142.Nm
143.Oo Cm set Ar N Oc Cm nptv6
144.Brq Ar name | all
145.Brq Cm list | show
146.Nm
147.Oo Cm set Ar N Oc Cm nptv6
148.Brq Ar name | all
149.Cm destroy
150.Nm
151.Oo Cm set Ar N Oc Cm nptv6 Ar name Cm stats Op Cm reset
152.Ss INTERNAL DIAGNOSTICS
153.Nm
154.Cm internal iflist
155.Nm
156.Cm internal talist
157.Nm
158.Cm internal vlist
159.Ss LIST OF RULES AND PREPROCESSING
160.Nm
161.Op Fl cfnNqS
162.Oo
163.Fl p Ar preproc
164.Oo
165.Ar preproc-flags
166.Oc
167.Oc
168.Ar pathname
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 D
314When listing, show only dynamic states.
315When deleting, delete only dynamic states.
316.It Fl f
317Run without prompting for confirmation for commands that can cause problems if misused,
318i.e.,
319.Cm flush .
320If there is no tty associated with the process, this is implied.
321The
322.Cm delete
323command with this flag ignores possible errors,
324i.e., nonexistent rule number.
325And for batched commands execution continues with the next command.
326.It Fl i
327When listing a table (see the
328.Sx LOOKUP TABLES
329section below for more information on lookup tables), format values
330as IP addresses.
331By default, values are shown as integers.
332.It Fl n
333Only check syntax of the command strings, without actually passing
334them to the kernel.
335.It Fl N
336Try to resolve addresses and service names in output.
337.It Fl q
338Be quiet when executing the
339.Cm add ,
340.Cm nat ,
341.Cm zero ,
342.Cm resetlog
343or
344.Cm flush
345commands;
346(implies
347.Fl f ) .
348This is useful when updating rulesets by executing multiple
349.Nm
350commands in a script
351(e.g.,
352.Ql sh\ /etc/rc.firewall ) ,
353or by processing a file with many
354.Nm
355rules across a remote login session.
356It also stops a table add or delete
357from failing if the entry already exists or is not present.
358.Pp
359The reason why this option may be important is that
360for some of these actions,
361.Nm
362may print a message; if the action results in blocking the
363traffic to the remote client,
364the remote login session will be closed
365and the rest of the ruleset will not be processed.
366Access to the console would then be required to recover.
367.It Fl S
368When listing rules, show the
369.Em set
370each rule belongs to.
371If this flag is not specified, disabled rules will not be
372listed.
373.It Fl s Op Ar field
374When listing pipes, sort according to one of the four
375counters (total or current packets or bytes).
376.It Fl t
377When listing, show last match timestamp converted with ctime().
378.It Fl T
379When listing, show last match timestamp as seconds from the epoch.
380This form can be more convenient for postprocessing by scripts.
381.El
382.Ss LIST OF RULES AND PREPROCESSING
383To ease configuration, rules can be put into a file which is
384processed using
385.Nm
386as shown in the last synopsis line.
387An absolute
388.Ar pathname
389must be used.
390The file will be read line by line and applied as arguments to the
391.Nm
392utility.
393.Pp
394Optionally, a preprocessor can be specified using
395.Fl p Ar preproc
396where
397.Ar pathname
398is to be piped through.
399Useful preprocessors include
400.Xr cpp 1
401and
402.Xr m4 1 .
403If
404.Ar preproc
405does not start with a slash
406.Pq Ql /
407as its first character, the usual
408.Ev PATH
409name search is performed.
410Care should be taken with this in environments where not all
411file systems are mounted (yet) by the time
412.Nm
413is being run (e.g.\& when they are mounted over NFS).
414Once
415.Fl p
416has been specified, any additional arguments are passed on to the preprocessor
417for interpretation.
418This allows for flexible configuration files (like conditionalizing
419them on the local hostname) and the use of macros to centralize
420frequently required arguments like IP addresses.
421.Ss TRAFFIC SHAPER CONFIGURATION
422The
423.Nm
424.Cm pipe , queue
425and
426.Cm sched
427commands are used to configure the traffic shaper and packet scheduler.
428See the
429.Sx TRAFFIC SHAPER (DUMMYNET) CONFIGURATION
430Section below for details.
431.Pp
432If the world and the kernel get out of sync the
433.Nm
434ABI may break, preventing you from being able to add any rules.
435This can adversely affect the booting process.
436You can use
437.Nm
438.Cm disable
439.Cm firewall
440to temporarily disable the firewall to regain access to the network,
441allowing you to fix the problem.
442.Sh PACKET FLOW
443A packet is checked against the active ruleset in multiple places
444in the protocol stack, under control of several sysctl variables.
445These places and variables are shown below, and it is important to
446have this picture in mind in order to design a correct ruleset.
447.Bd -literal -offset indent
448       ^    to upper layers    V
449       |                       |
450       +----------->-----------+
451       ^                       V
452 [ip(6)_input]           [ip(6)_output]     net.inet(6).ip(6).fw.enable=1
453       |                       |
454       ^                       V
455 [ether_demux]        [ether_output_frame]  net.link.ether.ipfw=1
456       |                       |
457       +-->--[bdg_forward]-->--+            net.link.bridge.ipfw=1
458       ^                       V
459       |      to devices       |
460.Ed
461.Pp
462The number of
463times the same packet goes through the firewall can
464vary between 0 and 4 depending on packet source and
465destination, and system configuration.
466.Pp
467Note that as packets flow through the stack, headers can be
468stripped or added to it, and so they may or may not be available
469for inspection.
470E.g., incoming packets will include the MAC header when
471.Nm
472is invoked from
473.Cm ether_demux() ,
474but the same packets will have the MAC header stripped off when
475.Nm
476is invoked from
477.Cm ip_input()
478or
479.Cm ip6_input() .
480.Pp
481Also note that each packet is always checked against the complete ruleset,
482irrespective of the place where the check occurs, or the source of the packet.
483If a rule contains some match patterns or actions which are not valid
484for the place of invocation (e.g.\& trying to match a MAC header within
485.Cm ip_input
486or
487.Cm ip6_input ),
488the match pattern will not match, but a
489.Cm not
490operator in front of such patterns
491.Em will
492cause the pattern to
493.Em always
494match on those packets.
495It is thus the responsibility of
496the programmer, if necessary, to write a suitable ruleset to
497differentiate among the possible places.
498.Cm skipto
499rules can be useful here, as an example:
500.Bd -literal -offset indent
501# packets from ether_demux or bdg_forward
502ipfw add 10 skipto 1000 all from any to any layer2 in
503# packets from ip_input
504ipfw add 10 skipto 2000 all from any to any not layer2 in
505# packets from ip_output
506ipfw add 10 skipto 3000 all from any to any not layer2 out
507# packets from ether_output_frame
508ipfw add 10 skipto 4000 all from any to any layer2 out
509.Ed
510.Pp
511(yes, at the moment there is no way to differentiate between
512ether_demux and bdg_forward).
513.Pp
514Also note that only actions
515.Cm allow,
516.Cm deny,
517.Cm netgraph,
518.Cm ngtee
519and related to
520.Cm dummynet
521are processed for
522.Cm layer2
523frames and all other actions act as if they were
524.Cm allow
525for such frames.
526Full set of actions is supported for IP packets without
527.Cm layer2
528headers only.
529For example,
530.Cm divert
531action does not divert
532.Cm layer2
533frames.
534.Sh SYNTAX
535In general, each keyword or argument must be provided as
536a separate command line argument, with no leading or trailing
537spaces.
538Keywords are case-sensitive, whereas arguments may
539or may not be case-sensitive depending on their nature
540(e.g.\& uid's are, hostnames are not).
541.Pp
542Some arguments (e.g., port or address lists) are comma-separated
543lists of values.
544In this case, spaces after commas ',' are allowed to make
545the line more readable.
546You can also put the entire
547command (including flags) into a single argument.
548E.g., the following forms are equivalent:
549.Bd -literal -offset indent
550ipfw -q add deny src-ip 10.0.0.0/24,127.0.0.1/8
551ipfw -q add deny src-ip 10.0.0.0/24, 127.0.0.1/8
552ipfw "-q add deny src-ip 10.0.0.0/24, 127.0.0.1/8"
553.Ed
554.Sh RULE FORMAT
555The format of firewall rules is the following:
556.Bd -ragged -offset indent
557.Bk -words
558.Op Ar rule_number
559.Op Cm set Ar set_number
560.Op Cm prob Ar match_probability
561.Ar action
562.Op Cm log Op Cm logamount Ar number
563.Op Cm altq Ar queue
564.Oo
565.Bro Cm tag | untag
566.Brc Ar number
567.Oc
568.Ar body
569.Ek
570.Ed
571.Pp
572where the body of the rule specifies which information is used
573for filtering packets, among the following:
574.Pp
575.Bl -tag -width "Source and dest. addresses and ports" -offset XXX -compact
576.It Layer-2 header fields
577When available
578.It IPv4 and IPv6 Protocol
579SCTP, TCP, UDP, ICMP, etc.
580.It Source and dest. addresses and ports
581.It Direction
582See Section
583.Sx PACKET FLOW
584.It Transmit and receive interface
585By name or address
586.It Misc. IP header fields
587Version, type of service, datagram length, identification,
588fragment flag (non-zero IP offset),
589Time To Live
590.It IP options
591.It IPv6 Extension headers
592Fragmentation, Hop-by-Hop options,
593Routing Headers, Source routing rthdr0, Mobile IPv6 rthdr2, IPSec options.
594.It IPv6 Flow-ID
595.It Misc. TCP header fields
596TCP flags (SYN, FIN, ACK, RST, etc.),
597sequence number, acknowledgment number,
598window
599.It TCP options
600.It ICMP types
601for ICMP packets
602.It ICMP6 types
603for ICMP6 packets
604.It User/group ID
605When the packet can be associated with a local socket.
606.It Divert status
607Whether a packet came from a divert socket (e.g.,
608.Xr natd 8 ) .
609.It Fib annotation state
610Whether a packet has been tagged for using a specific FIB (routing table)
611in future forwarding decisions.
612.El
613.Pp
614Note that some of the above information, e.g.\& source MAC or IP addresses and
615TCP/UDP ports, can be easily spoofed, so filtering on those fields
616alone might not guarantee the desired results.
617.Bl -tag -width indent
618.It Ar rule_number
619Each rule is associated with a
620.Ar rule_number
621in the range 1..65535, with the latter reserved for the
622.Em default
623rule.
624Rules are checked sequentially by rule number.
625Multiple rules can have the same number, in which case they are
626checked (and listed) according to the order in which they have
627been added.
628If a rule is entered without specifying a number, the kernel will
629assign one in such a way that the rule becomes the last one
630before the
631.Em default
632rule.
633Automatic rule numbers are assigned by incrementing the last
634non-default rule number by the value of the sysctl variable
635.Ar net.inet.ip.fw.autoinc_step
636which defaults to 100.
637If this is not possible (e.g.\& because we would go beyond the
638maximum allowed rule number), the number of the last
639non-default value is used instead.
640.It Cm set Ar set_number
641Each rule is associated with a
642.Ar set_number
643in the range 0..31.
644Sets can be individually disabled and enabled, so this parameter
645is of fundamental importance for atomic ruleset manipulation.
646It can be also used to simplify deletion of groups of rules.
647If a rule is entered without specifying a set number,
648set 0 will be used.
649.br
650Set 31 is special in that it cannot be disabled,
651and rules in set 31 are not deleted by the
652.Nm ipfw flush
653command (but you can delete them with the
654.Nm ipfw delete set 31
655command).
656Set 31 is also used for the
657.Em default
658rule.
659.It Cm prob Ar match_probability
660A match is only declared with the specified probability
661(floating point number between 0 and 1).
662This can be useful for a number of applications such as
663random packet drop or
664(in conjunction with
665.Nm dummynet )
666to simulate the effect of multiple paths leading to out-of-order
667packet delivery.
668.Pp
669Note: this condition is checked before any other condition, including
670ones such as
671.Cm keep-state
672or
673.Cm check-state
674which might have
675side effects.
676.It Cm log Op Cm logamount Ar number
677Packets matching a rule with the
678.Cm log
679keyword will be made available for logging in two ways:
680if the sysctl variable
681.Va net.inet.ip.fw.verbose
682is set to 0 (default), one can use
683.Xr bpf 4
684attached to the
685.Li ipfw0
686pseudo interface.
687This pseudo interface can be created manually after a system
688boot by using the following command:
689.Bd -literal -offset indent
690# ifconfig ipfw0 create
691.Ed
692.Pp
693Or, automatically at boot time by adding the following
694line to the
695.Xr rc.conf 5
696file:
697.Bd -literal -offset indent
698firewall_logif="YES"
699.Ed
700.Pp
701There is zero overhead when no
702.Xr bpf 4
703is attached to the pseudo interface.
704.Pp
705If
706.Va net.inet.ip.fw.verbose
707is set to 1, packets will be logged to
708.Xr syslogd 8
709with a
710.Dv LOG_SECURITY
711facility up to a maximum of
712.Cm logamount
713packets.
714If no
715.Cm logamount
716is specified, the limit is taken from the sysctl variable
717.Va net.inet.ip.fw.verbose_limit .
718In both cases, a value of 0 means unlimited logging.
719.Pp
720Once the limit is reached, logging can be re-enabled by
721clearing the logging counter or the packet counter for that entry, see the
722.Cm resetlog
723command.
724.Pp
725Note: logging is done after all other packet matching conditions
726have been successfully verified, and before performing the final
727action (accept, deny, etc.) on the packet.
728.It Cm tag Ar number
729When a packet matches a rule with the
730.Cm tag
731keyword, the numeric tag for the given
732.Ar number
733in the range 1..65534 will be attached to the packet.
734The tag acts as an internal marker (it is not sent out over
735the wire) that can be used to identify these packets later on.
736This can be used, for example, to provide trust between interfaces
737and to start doing policy-based filtering.
738A packet can have multiple tags at the same time.
739Tags are "sticky", meaning once a tag is applied to a packet by a
740matching rule it exists until explicit removal.
741Tags are kept with the packet everywhere within the kernel, but are
742lost when packet leaves the kernel, for example, on transmitting
743packet out to the network or sending packet to a
744.Xr divert 4
745socket.
746.Pp
747To check for previously applied tags, use the
748.Cm tagged
749rule option.
750To delete previously applied tag, use the
751.Cm untag
752keyword.
753.Pp
754Note: since tags are kept with the packet everywhere in kernelspace,
755they can be set and unset anywhere in the kernel network subsystem
756(using the
757.Xr mbuf_tags 9
758facility), not only by means of the
759.Xr ipfw 4
760.Cm tag
761and
762.Cm untag
763keywords.
764For example, there can be a specialized
765.Xr netgraph 4
766node doing traffic analyzing and tagging for later inspecting
767in firewall.
768.It Cm untag Ar number
769When a packet matches a rule with the
770.Cm untag
771keyword, the tag with the number
772.Ar number
773is searched among the tags attached to this packet and,
774if found, removed from it.
775Other tags bound to packet, if present, are left untouched.
776.It Cm altq Ar queue
777When a packet matches a rule with the
778.Cm altq
779keyword, the ALTQ identifier for the given
780.Ar queue
781(see
782.Xr altq 4 )
783will be attached.
784Note that this ALTQ tag is only meaningful for packets going "out" of IPFW,
785and not being rejected or going to divert sockets.
786Note that if there is insufficient memory at the time the packet is
787processed, it will not be tagged, so it is wise to make your ALTQ
788"default" queue policy account for this.
789If multiple
790.Cm altq
791rules match a single packet, only the first one adds the ALTQ classification
792tag.
793In doing so, traffic may be shaped by using
794.Cm count Cm altq Ar queue
795rules for classification early in the ruleset, then later applying
796the filtering decision.
797For example,
798.Cm check-state
799and
800.Cm keep-state
801rules may come later and provide the actual filtering decisions in
802addition to the fallback ALTQ tag.
803.Pp
804You must run
805.Xr pfctl 8
806to set up the queues before IPFW will be able to look them up by name,
807and if the ALTQ disciplines are rearranged, the rules in containing the
808queue identifiers in the kernel will likely have gone stale and need
809to be reloaded.
810Stale queue identifiers will probably result in misclassification.
811.Pp
812All system ALTQ processing can be turned on or off via
813.Nm
814.Cm enable Ar altq
815and
816.Nm
817.Cm disable Ar altq .
818The usage of
819.Va net.inet.ip.fw.one_pass
820is irrelevant to ALTQ traffic shaping, as the actual rule action is followed
821always after adding an ALTQ tag.
822.El
823.Ss RULE ACTIONS
824A rule can be associated with one of the following actions, which
825will be executed when the packet matches the body of the rule.
826.Bl -tag -width indent
827.It Cm allow | accept | pass | permit
828Allow packets that match rule.
829The search terminates.
830.It Cm check-state Op Ar :flowname | Cm :any
831Checks the packet against the dynamic ruleset.
832If a match is found, execute the action associated with
833the rule which generated this dynamic rule, otherwise
834move to the next rule.
835.br
836.Cm Check-state
837rules do not have a body.
838If no
839.Cm check-state
840rule is found, the dynamic ruleset is checked at the first
841.Cm keep-state
842or
843.Cm limit
844rule.
845The
846.Ar :flowname
847is symbolic name assigned to dynamic rule by
848.Cm keep-state
849opcode.
850The special flowname
851.Cm :any
852can be used to ignore states flowname when matching.
853The
854.Cm :default
855keyword is special name used for compatibility with old rulesets.
856.It Cm count
857Update counters for all packets that match rule.
858The search continues with the next rule.
859.It Cm deny | drop
860Discard packets that match this rule.
861The search terminates.
862.It Cm divert Ar port
863Divert packets that match this rule to the
864.Xr divert 4
865socket bound to port
866.Ar port .
867The search terminates.
868.It Cm fwd | forward Ar ipaddr | tablearg Ns Op , Ns Ar port
869Change the next-hop on matching packets to
870.Ar ipaddr ,
871which can be an IP address or a host name.
872The next hop can also be supplied by the last table
873looked up for the packet by using the
874.Cm tablearg
875keyword instead of an explicit address.
876The search terminates if this rule matches.
877.Pp
878If
879.Ar ipaddr
880is a local address, then matching packets will be forwarded to
881.Ar port
882(or the port number in the packet if one is not specified in the rule)
883on the local machine.
884.br
885If
886.Ar ipaddr
887is not a local address, then the port number
888(if specified) is ignored, and the packet will be
889forwarded to the remote address, using the route as found in
890the local routing table for that IP.
891.br
892A
893.Ar fwd
894rule will not match layer-2 packets (those received
895on ether_input, ether_output, or bridged).
896.br
897The
898.Cm fwd
899action does not change the contents of the packet at all.
900In particular, the destination address remains unmodified, so
901packets forwarded to another system will usually be rejected by that system
902unless there is a matching rule on that system to capture them.
903For packets forwarded locally,
904the local address of the socket will be
905set to the original destination address of the packet.
906This makes the
907.Xr netstat 1
908entry look rather weird but is intended for
909use with transparent proxy servers.
910.It Cm nat Ar nat_nr | tablearg
911Pass packet to a
912nat instance
913(for network address translation, address redirect, etc.):
914see the
915.Sx NETWORK ADDRESS TRANSLATION (NAT)
916Section for further information.
917.It Cm nat64lsn Ar name
918Pass packet to a stateful NAT64 instance (for IPv6/IPv4 network address and
919protocol translation): see the
920.Sx IPv6/IPv4 NETWORK ADDRESS AND PROTOCOL TRANSLATION
921Section for further information.
922.It Cm nat64stl Ar name
923Pass packet to a stateless NAT64 instance (for IPv6/IPv4 network address and
924protocol translation): see the
925.Sx IPv6/IPv4 NETWORK ADDRESS AND PROTOCOL TRANSLATION
926Section for further information.
927.It Cm nptv6 Ar name
928Pass packet to a NPTv6 instance (for IPv6-to-IPv6 network prefix translation):
929see the
930.Sx IPv6-to-IPv6 NETWORK PREFIX TRANSLATION (NPTv6)
931Section for further information.
932.It Cm pipe Ar pipe_nr
933Pass packet to a
934.Nm dummynet
935.Dq pipe
936(for bandwidth limitation, delay, etc.).
937See the
938.Sx TRAFFIC SHAPER (DUMMYNET) CONFIGURATION
939Section for further information.
940The search terminates; however, on exit from the pipe and if
941the
942.Xr sysctl 8
943variable
944.Va net.inet.ip.fw.one_pass
945is not set, the packet is passed again to the firewall code
946starting from the next rule.
947.It Cm queue Ar queue_nr
948Pass packet to a
949.Nm dummynet
950.Dq queue
951(for bandwidth limitation using WF2Q+).
952.It Cm reject
953(Deprecated).
954Synonym for
955.Cm unreach host .
956.It Cm reset
957Discard packets that match this rule, and if the
958packet is a TCP packet, try to send a TCP reset (RST) notice.
959The search terminates.
960.It Cm reset6
961Discard packets that match this rule, and if the
962packet is a TCP packet, try to send a TCP reset (RST) notice.
963The search terminates.
964.It Cm skipto Ar number | tablearg
965Skip all subsequent rules numbered less than
966.Ar number .
967The search continues with the first rule numbered
968.Ar number
969or higher.
970It is possible to use the
971.Cm tablearg
972keyword with a skipto for a
973.Em computed
974skipto. Skipto may work either in O(log(N)) or in O(1) depending
975on amount of memory and/or sysctl variables.
976See the
977.Sx SYSCTL VARIABLES
978section for more details.
979.It Cm call Ar number | tablearg
980The current rule number is saved in the internal stack and
981ruleset processing continues with the first rule numbered
982.Ar number
983or higher.
984If later a rule with the
985.Cm return
986action is encountered, the processing returns to the first rule
987with number of this
988.Cm call
989rule plus one or higher
990(the same behaviour as with packets returning from
991.Xr divert 4
992socket after a
993.Cm divert
994action).
995This could be used to make somewhat like an assembly language
996.Dq subroutine
997calls to rules with common checks for different interfaces, etc.
998.Pp
999Rule with any number could be called, not just forward jumps as with
1000.Cm skipto .
1001So, to prevent endless loops in case of mistakes, both
1002.Cm call
1003and
1004.Cm return
1005actions don't do any jumps and simply go to the next rule if memory
1006cannot be allocated or stack overflowed/underflowed.
1007.Pp
1008Internally stack for rule numbers is implemented using
1009.Xr mbuf_tags 9
1010facility and currently has size of 16 entries.
1011As mbuf tags are lost when packet leaves the kernel,
1012.Cm divert
1013should not be used in subroutines to avoid endless loops
1014and other undesired effects.
1015.It Cm return
1016Takes rule number saved to internal stack by the last
1017.Cm call
1018action and returns ruleset processing to the first rule
1019with number greater than number of corresponding
1020.Cm call
1021rule.
1022See description of the
1023.Cm call
1024action for more details.
1025.Pp
1026Note that
1027.Cm return
1028rules usually end a
1029.Dq subroutine
1030and thus are unconditional, but
1031.Nm
1032command-line utility currently requires every action except
1033.Cm check-state
1034to have body.
1035While it is sometimes useful to return only on some packets,
1036usually you want to print just
1037.Dq return
1038for readability.
1039A workaround for this is to use new syntax and
1040.Fl c
1041switch:
1042.Bd -literal -offset indent
1043# Add a rule without actual body
1044ipfw add 2999 return via any
1045
1046# List rules without "from any to any" part
1047ipfw -c list
1048.Ed
1049.Pp
1050This cosmetic annoyance may be fixed in future releases.
1051.It Cm tee Ar port
1052Send a copy of packets matching this rule to the
1053.Xr divert 4
1054socket bound to port
1055.Ar port .
1056The search continues with the next rule.
1057.It Cm unreach Ar code
1058Discard packets that match this rule, and try to send an ICMP
1059unreachable notice with code
1060.Ar code ,
1061where
1062.Ar code
1063is a number from 0 to 255, or one of these aliases:
1064.Cm net , host , protocol , port ,
1065.Cm needfrag , srcfail , net-unknown , host-unknown ,
1066.Cm isolated , net-prohib , host-prohib , tosnet ,
1067.Cm toshost , filter-prohib , host-precedence
1068or
1069.Cm precedence-cutoff .
1070The search terminates.
1071.It Cm unreach6 Ar code
1072Discard packets that match this rule, and try to send an ICMPv6
1073unreachable notice with code
1074.Ar code ,
1075where
1076.Ar code
1077is a number from 0, 1, 3 or 4, or one of these aliases:
1078.Cm no-route, admin-prohib, address
1079or
1080.Cm port .
1081The search terminates.
1082.It Cm netgraph Ar cookie
1083Divert packet into netgraph with given
1084.Ar cookie .
1085The search terminates.
1086If packet is later returned from netgraph it is either
1087accepted or continues with the next rule, depending on
1088.Va net.inet.ip.fw.one_pass
1089sysctl variable.
1090.It Cm ngtee Ar cookie
1091A copy of packet is diverted into netgraph, original
1092packet continues with the next rule.
1093See
1094.Xr ng_ipfw 4
1095for more information on
1096.Cm netgraph
1097and
1098.Cm ngtee
1099actions.
1100.It Cm setfib Ar fibnum | tablearg
1101The packet is tagged so as to use the FIB (routing table)
1102.Ar fibnum
1103in any subsequent forwarding decisions.
1104In the current implementation, this is limited to the values 0 through 15, see
1105.Xr setfib 2 .
1106Processing continues at the next rule.
1107It is possible to use the
1108.Cm tablearg
1109keyword with setfib.
1110If the tablearg value is not within the compiled range of fibs,
1111the packet's fib is set to 0.
1112.It Cm setdscp Ar DSCP | number | tablearg
1113Set specified DiffServ codepoint for an IPv4/IPv6 packet.
1114Processing continues at the next rule.
1115Supported values are:
1116.Pp
1117.Cm cs0
1118.Pq Dv 000000 ,
1119.Cm cs1
1120.Pq Dv 001000 ,
1121.Cm cs2
1122.Pq Dv 010000 ,
1123.Cm cs3
1124.Pq Dv 011000 ,
1125.Cm cs4
1126.Pq Dv 100000 ,
1127.Cm cs5
1128.Pq Dv 101000 ,
1129.Cm cs6
1130.Pq Dv 110000 ,
1131.Cm cs7
1132.Pq Dv 111000 ,
1133.Cm af11
1134.Pq Dv 001010 ,
1135.Cm af12
1136.Pq Dv 001100 ,
1137.Cm af13
1138.Pq Dv 001110 ,
1139.Cm af21
1140.Pq Dv 010010 ,
1141.Cm af22
1142.Pq Dv 010100 ,
1143.Cm af23
1144.Pq Dv 010110 ,
1145.Cm af31
1146.Pq Dv 011010 ,
1147.Cm af32
1148.Pq Dv 011100 ,
1149.Cm af33
1150.Pq Dv 011110 ,
1151.Cm af41
1152.Pq Dv 100010 ,
1153.Cm af42
1154.Pq Dv 100100 ,
1155.Cm af43
1156.Pq Dv 100110 ,
1157.Cm ef
1158.Pq Dv 101110 ,
1159.Cm be
1160.Pq Dv 000000 .
1161Additionally, DSCP value can be specified by number (0..63).
1162It is also possible to use the
1163.Cm tablearg
1164keyword with setdscp.
1165If the tablearg value is not within the 0..63 range, lower 6 bits of supplied
1166value are used.
1167.It Cm tcp-setmss Ar mss
1168Set the Maximum Segment Size (MSS) in the TCP segment to value
1169.Ar mss .
1170The kernel module
1171.Cm ipfw_pmod
1172should be loaded or kernel should have
1173.Cm options IPFIREWALL_PMOD
1174to be able use this action.
1175This command does not change a packet if original MSS value is lower than
1176specified value.
1177Both TCP over IPv4 and over IPv6 are supported.
1178Regardless of matched a packet or not by the
1179.Cm tcp-setmss
1180rule, the search continues with the next rule.
1181.It Cm reass
1182Queue and reassemble IPv4 fragments.
1183If the packet is not fragmented, counters are updated and
1184processing continues with the next rule.
1185If the packet is the last logical fragment, the packet is reassembled and, if
1186.Va net.inet.ip.fw.one_pass
1187is set to 0, processing continues with the next rule.
1188Otherwise, the packet is allowed to pass and the search terminates.
1189If the packet is a fragment in the middle of a logical group of fragments,
1190it is consumed and
1191processing stops immediately.
1192.Pp
1193Fragment handling can be tuned via
1194.Va net.inet.ip.maxfragpackets
1195and
1196.Va net.inet.ip.maxfragsperpacket
1197which limit, respectively, the maximum number of processable
1198fragments (default: 800) and
1199the maximum number of fragments per packet (default: 16).
1200.Pp
1201NOTA BENE: since fragments do not contain port numbers,
1202they should be avoided with the
1203.Nm reass
1204rule.
1205Alternatively, direction-based (like
1206.Nm in
1207/
1208.Nm out
1209) and source-based (like
1210.Nm via
1211) match patterns can be used to select fragments.
1212.Pp
1213Usually a simple rule like:
1214.Bd -literal -offset indent
1215# reassemble incoming fragments
1216ipfw add reass all from any to any in
1217.Ed
1218.Pp
1219is all you need at the beginning of your ruleset.
1220.It Cm abort
1221Discard packets that match this rule, and if the packet is an SCTP packet,
1222try to send an SCTP packet containing an ABORT chunk.
1223The search terminates.
1224.It Cm abort6
1225Discard packets that match this rule, and if the packet is an SCTP packet,
1226try to send an SCTP packet containing an ABORT chunk.
1227The search terminates.
1228.El
1229.Ss RULE BODY
1230The body of a rule contains zero or more patterns (such as
1231specific source and destination addresses or ports,
1232protocol options, incoming or outgoing interfaces, etc.)
1233that the packet must match in order to be recognised.
1234In general, the patterns are connected by (implicit)
1235.Cm and
1236operators -- i.e., all must match in order for the
1237rule to match.
1238Individual patterns can be prefixed by the
1239.Cm not
1240operator to reverse the result of the match, as in
1241.Pp
1242.Dl "ipfw add 100 allow ip from not 1.2.3.4 to any"
1243.Pp
1244Additionally, sets of alternative match patterns
1245.Pq Em or-blocks
1246can be constructed by putting the patterns in
1247lists enclosed between parentheses ( ) or braces { }, and
1248using the
1249.Cm or
1250operator as follows:
1251.Pp
1252.Dl "ipfw add 100 allow ip from { x or not y or z } to any"
1253.Pp
1254Only one level of parentheses is allowed.
1255Beware that most shells have special meanings for parentheses
1256or braces, so it is advisable to put a backslash \\ in front of them
1257to prevent such interpretations.
1258.Pp
1259The body of a rule must in general include a source and destination
1260address specifier.
1261The keyword
1262.Ar any
1263can be used in various places to specify that the content of
1264a required field is irrelevant.
1265.Pp
1266The rule body has the following format:
1267.Bd -ragged -offset indent
1268.Op Ar proto Cm from Ar src Cm to Ar dst
1269.Op Ar options
1270.Ed
1271.Pp
1272The first part (proto from src to dst) is for backward
1273compatibility with earlier versions of
1274.Fx .
1275In modern
1276.Fx
1277any match pattern (including MAC headers, IP protocols,
1278addresses and ports) can be specified in the
1279.Ar options
1280section.
1281.Pp
1282Rule fields have the following meaning:
1283.Bl -tag -width indent
1284.It Ar proto : protocol | Cm { Ar protocol Cm or ... }
1285.It Ar protocol : Oo Cm not Oc Ar protocol-name | protocol-number
1286An IP protocol specified by number or name
1287(for a complete list see
1288.Pa /etc/protocols ) ,
1289or one of the following keywords:
1290.Bl -tag -width indent
1291.It Cm ip4 | ipv4
1292Matches IPv4 packets.
1293.It Cm ip6 | ipv6
1294Matches IPv6 packets.
1295.It Cm ip | all
1296Matches any packet.
1297.El
1298.Pp
1299The
1300.Cm ipv6
1301in
1302.Cm proto
1303option will be treated as inner protocol.
1304And, the
1305.Cm ipv4
1306is not available in
1307.Cm proto
1308option.
1309.Pp
1310The
1311.Cm { Ar protocol Cm or ... }
1312format (an
1313.Em or-block )
1314is provided for convenience only but its use is deprecated.
1315.It Ar src No and Ar dst : Bro Cm addr | Cm { Ar addr Cm or ... } Brc Op Oo Cm not Oc Ar ports
1316An address (or a list, see below)
1317optionally followed by
1318.Ar ports
1319specifiers.
1320.Pp
1321The second format
1322.Em ( or-block
1323with multiple addresses) is provided for convenience only and
1324its use is discouraged.
1325.It Ar addr : Oo Cm not Oc Bro
1326.Cm any | me | me6 |
1327.Cm table Ns Pq Ar name Ns Op , Ns Ar value
1328.Ar | addr-list | addr-set
1329.Brc
1330.Bl -tag -width indent
1331.It Cm any
1332Matches any IP address.
1333.It Cm me
1334Matches any IP address configured on an interface in the system.
1335.It Cm me6
1336Matches any IPv6 address configured on an interface in the system.
1337The address list is evaluated at the time the packet is
1338analysed.
1339.It Cm table Ns Pq Ar name Ns Op , Ns Ar value
1340Matches any IPv4 or IPv6 address for which an entry exists in the lookup table
1341.Ar number .
1342If an optional 32-bit unsigned
1343.Ar value
1344is also specified, an entry will match only if it has this value.
1345See the
1346.Sx LOOKUP TABLES
1347section below for more information on lookup tables.
1348.El
1349.It Ar addr-list : ip-addr Ns Op Ns , Ns Ar addr-list
1350.It Ar ip-addr :
1351A host or subnet address specified in one of the following ways:
1352.Bl -tag -width indent
1353.It Ar numeric-ip | hostname
1354Matches a single IPv4 address, specified as dotted-quad or a hostname.
1355Hostnames are resolved at the time the rule is added to the firewall list.
1356.It Ar addr Ns / Ns Ar masklen
1357Matches all addresses with base
1358.Ar addr
1359(specified as an IP address, a network number, or a hostname)
1360and mask width of
1361.Cm masklen
1362bits.
1363As an example, 1.2.3.4/25 or 1.2.3.0/25 will match
1364all IP numbers from 1.2.3.0 to 1.2.3.127 .
1365.It Ar addr Ns : Ns Ar mask
1366Matches all addresses with base
1367.Ar addr
1368(specified as an IP address, a network number, or a hostname)
1369and the mask of
1370.Ar mask ,
1371specified as a dotted quad.
1372As an example, 1.2.3.4:255.0.255.0 or 1.0.3.0:255.0.255.0 will match
13731.*.3.*.
1374This form is advised only for non-contiguous
1375masks.
1376It is better to resort to the
1377.Ar addr Ns / Ns Ar masklen
1378format for contiguous masks, which is more compact and less
1379error-prone.
1380.El
1381.It Ar addr-set : addr Ns Oo Ns / Ns Ar masklen Oc Ns Cm { Ns Ar list Ns Cm }
1382.It Ar list : Bro Ar num | num-num Brc Ns Op Ns , Ns Ar list
1383Matches all addresses with base address
1384.Ar addr
1385(specified as an IP address, a network number, or a hostname)
1386and whose last byte is in the list between braces { } .
1387Note that there must be no spaces between braces and
1388numbers (spaces after commas are allowed).
1389Elements of the list can be specified as single entries
1390or ranges.
1391The
1392.Ar masklen
1393field is used to limit the size of the set of addresses,
1394and can have any value between 24 and 32.
1395If not specified,
1396it will be assumed as 24.
1397.br
1398This format is particularly useful to handle sparse address sets
1399within a single rule.
1400Because the matching occurs using a
1401bitmask, it takes constant time and dramatically reduces
1402the complexity of rulesets.
1403.br
1404As an example, an address specified as 1.2.3.4/24{128,35-55,89}
1405or 1.2.3.0/24{128,35-55,89}
1406will match the following IP addresses:
1407.br
14081.2.3.128, 1.2.3.35 to 1.2.3.55, 1.2.3.89 .
1409.It Ar addr6-list : ip6-addr Ns Op Ns , Ns Ar addr6-list
1410.It Ar ip6-addr :
1411A host or subnet specified one of the following ways:
1412.Bl -tag -width indent
1413.It Ar numeric-ip | hostname
1414Matches a single IPv6 address as allowed by
1415.Xr inet_pton 3
1416or a hostname.
1417Hostnames are resolved at the time the rule is added to the firewall
1418list.
1419.It Ar addr Ns / Ns Ar masklen
1420Matches all IPv6 addresses with base
1421.Ar addr
1422(specified as allowed by
1423.Xr inet_pton
1424or a hostname)
1425and mask width of
1426.Cm masklen
1427bits.
1428.It Ar addr Ns / Ns Ar mask
1429Matches all IPv6 addresses with base
1430.Ar addr
1431(specified as allowed by
1432.Xr inet_pton
1433or a hostname)
1434and the mask of
1435.Ar mask ,
1436specified as allowed by
1437.Xr inet_pton.
1438As an example, fe::640:0:0/ffff::ffff:ffff:0:0 will match
1439fe:*:*:*:0:640:*:*.
1440This form is advised only for non-contiguous
1441masks.
1442It is better to resort to the
1443.Ar addr Ns / Ns Ar masklen
1444format for contiguous masks, which is more compact and less
1445error-prone.
1446.El
1447.Pp
1448No support for sets of IPv6 addresses is provided because IPv6 addresses
1449are typically random past the initial prefix.
1450.It Ar ports : Bro Ar port | port Ns \&- Ns Ar port Ns Brc Ns Op , Ns Ar ports
1451For protocols which support port numbers (such as SCTP, TCP and UDP), optional
1452.Cm ports
1453may be specified as one or more ports or port ranges, separated
1454by commas but no spaces, and an optional
1455.Cm not
1456operator.
1457The
1458.Ql \&-
1459notation specifies a range of ports (including boundaries).
1460.Pp
1461Service names (from
1462.Pa /etc/services )
1463may be used instead of numeric port values.
1464The length of the port list is limited to 30 ports or ranges,
1465though one can specify larger ranges by using an
1466.Em or-block
1467in the
1468.Cm options
1469section of the rule.
1470.Pp
1471A backslash
1472.Pq Ql \e
1473can be used to escape the dash
1474.Pq Ql -
1475character in a service name (from a shell, the backslash must be
1476typed twice to avoid the shell itself interpreting it as an escape
1477character).
1478.Pp
1479.Dl "ipfw add count tcp from any ftp\e\e-data-ftp to any"
1480.Pp
1481Fragmented packets which have a non-zero offset (i.e., not the first
1482fragment) will never match a rule which has one or more port
1483specifications.
1484See the
1485.Cm frag
1486option for details on matching fragmented packets.
1487.El
1488.Ss RULE OPTIONS (MATCH PATTERNS)
1489Additional match patterns can be used within
1490rules.
1491Zero or more of these so-called
1492.Em options
1493can be present in a rule, optionally prefixed by the
1494.Cm not
1495operand, and possibly grouped into
1496.Em or-blocks .
1497.Pp
1498The following match patterns can be used (listed in alphabetical order):
1499.Bl -tag -width indent
1500.It Cm // this is a comment.
1501Inserts the specified text as a comment in the rule.
1502Everything following // is considered as a comment and stored in the rule.
1503You can have comment-only rules, which are listed as having a
1504.Cm count
1505action followed by the comment.
1506.It Cm bridged
1507Alias for
1508.Cm layer2 .
1509.It Cm defer-immediate-action | defer-action
1510A rule with this option will not perform normal action
1511upon a match. This option is intended to be used with
1512.Cm record-state
1513or
1514.Cm keep-state
1515as the dynamic rule, created but ignored on match, will work
1516as intended.
1517Rules with both
1518.Cm record-state
1519and
1520.Cm defer-immediate-action
1521create a dynamic rule and continue with the next rule without actually
1522performing the action part of this rule. When the rule is later activated
1523via the state table, the action is performed as usual.
1524.It Cm diverted
1525Matches only packets generated by a divert socket.
1526.It Cm diverted-loopback
1527Matches only packets coming from a divert socket back into the IP stack
1528input for delivery.
1529.It Cm diverted-output
1530Matches only packets going from a divert socket back outward to the IP
1531stack output for delivery.
1532.It Cm dst-ip Ar ip-address
1533Matches IPv4 packets whose destination IP is one of the address(es)
1534specified as argument.
1535.It Bro Cm dst-ip6 | dst-ipv6 Brc Ar ip6-address
1536Matches IPv6 packets whose destination IP is one of the address(es)
1537specified as argument.
1538.It Cm dst-port Ar ports
1539Matches IP packets whose destination port is one of the port(s)
1540specified as argument.
1541.It Cm established
1542Matches TCP packets that have the RST or ACK bits set.
1543.It Cm ext6hdr Ar header
1544Matches IPv6 packets containing the extended header given by
1545.Ar header .
1546Supported headers are:
1547.Pp
1548Fragment,
1549.Pq Cm frag ,
1550Hop-to-hop options
1551.Pq Cm hopopt ,
1552any type of Routing Header
1553.Pq Cm route ,
1554Source routing Routing Header Type 0
1555.Pq Cm rthdr0 ,
1556Mobile IPv6 Routing Header Type 2
1557.Pq Cm rthdr2 ,
1558Destination options
1559.Pq Cm dstopt ,
1560IPSec authentication headers
1561.Pq Cm ah ,
1562and IPsec encapsulated security payload headers
1563.Pq Cm esp .
1564.It Cm fib Ar fibnum
1565Matches a packet that has been tagged to use
1566the given FIB (routing table) number.
1567.It Cm flow Ar table Ns Pq Ar name Ns Op , Ns Ar value
1568Search for the flow entry in lookup table
1569.Ar name .
1570If not found, the match fails.
1571Otherwise, the match succeeds and
1572.Cm tablearg
1573is set to the value extracted from the table.
1574.Pp
1575This option can be useful to quickly dispatch traffic based on
1576certain packet fields.
1577See the
1578.Sx LOOKUP TABLES
1579section below for more information on lookup tables.
1580.It Cm flow-id Ar labels
1581Matches IPv6 packets containing any of the flow labels given in
1582.Ar labels .
1583.Ar labels
1584is a comma separated list of numeric flow labels.
1585.It Cm frag
1586Matches packets that are fragments and not the first
1587fragment of an IP datagram.
1588Note that these packets will not have
1589the next protocol header (e.g.\& TCP, UDP) so options that look into
1590these headers cannot match.
1591.It Cm gid Ar group
1592Matches all TCP or UDP packets sent by or received for a
1593.Ar group .
1594A
1595.Ar group
1596may be specified by name or number.
1597.It Cm jail Ar jail
1598Matches all TCP or UDP packets sent by or received for the
1599jail whose ID or name is
1600.Ar jail .
1601.It Cm icmptypes Ar types
1602Matches ICMP packets whose ICMP type is in the list
1603.Ar types .
1604The list may be specified as any combination of
1605individual types (numeric) separated by commas.
1606.Em Ranges are not allowed .
1607The supported ICMP types are:
1608.Pp
1609echo reply
1610.Pq Cm 0 ,
1611destination unreachable
1612.Pq Cm 3 ,
1613source quench
1614.Pq Cm 4 ,
1615redirect
1616.Pq Cm 5 ,
1617echo request
1618.Pq Cm 8 ,
1619router advertisement
1620.Pq Cm 9 ,
1621router solicitation
1622.Pq Cm 10 ,
1623time-to-live exceeded
1624.Pq Cm 11 ,
1625IP header bad
1626.Pq Cm 12 ,
1627timestamp request
1628.Pq Cm 13 ,
1629timestamp reply
1630.Pq Cm 14 ,
1631information request
1632.Pq Cm 15 ,
1633information reply
1634.Pq Cm 16 ,
1635address mask request
1636.Pq Cm 17
1637and address mask reply
1638.Pq Cm 18 .
1639.It Cm icmp6types Ar types
1640Matches ICMP6 packets whose ICMP6 type is in the list of
1641.Ar types .
1642The list may be specified as any combination of
1643individual types (numeric) separated by commas.
1644.Em Ranges are not allowed .
1645.It Cm in | out
1646Matches incoming or outgoing packets, respectively.
1647.Cm in
1648and
1649.Cm out
1650are mutually exclusive (in fact,
1651.Cm out
1652is implemented as
1653.Cm not in Ns No ).
1654.It Cm ipid Ar id-list
1655Matches IPv4 packets whose
1656.Cm ip_id
1657field has value included in
1658.Ar id-list ,
1659which is either a single value or a list of values or ranges
1660specified in the same way as
1661.Ar ports .
1662.It Cm iplen Ar len-list
1663Matches IP packets whose total length, including header and data, is
1664in the set
1665.Ar len-list ,
1666which is either a single value or a list of values or ranges
1667specified in the same way as
1668.Ar ports .
1669.It Cm ipoptions Ar spec
1670Matches packets whose IPv4 header contains the comma separated list of
1671options specified in
1672.Ar spec .
1673The supported IP options are:
1674.Pp
1675.Cm ssrr
1676(strict source route),
1677.Cm lsrr
1678(loose source route),
1679.Cm rr
1680(record packet route) and
1681.Cm ts
1682(timestamp).
1683The absence of a particular option may be denoted
1684with a
1685.Ql \&! .
1686.It Cm ipprecedence Ar precedence
1687Matches IPv4 packets whose precedence field is equal to
1688.Ar precedence .
1689.It Cm ipsec
1690Matches packets that have IPSEC history associated with them
1691(i.e., the packet comes encapsulated in IPSEC, the kernel
1692has IPSEC support, and can correctly decapsulate it).
1693.Pp
1694Note that specifying
1695.Cm ipsec
1696is different from specifying
1697.Cm proto Ar ipsec
1698as the latter will only look at the specific IP protocol field,
1699irrespective of IPSEC kernel support and the validity of the IPSEC data.
1700.Pp
1701Further note that this flag is silently ignored in kernels without
1702IPSEC support.
1703It does not affect rule processing when given and the
1704rules are handled as if with no
1705.Cm ipsec
1706flag.
1707.It Cm iptos Ar spec
1708Matches IPv4 packets whose
1709.Cm tos
1710field contains the comma separated list of
1711service types specified in
1712.Ar spec .
1713The supported IP types of service are:
1714.Pp
1715.Cm lowdelay
1716.Pq Dv IPTOS_LOWDELAY ,
1717.Cm throughput
1718.Pq Dv IPTOS_THROUGHPUT ,
1719.Cm reliability
1720.Pq Dv IPTOS_RELIABILITY ,
1721.Cm mincost
1722.Pq Dv IPTOS_MINCOST ,
1723.Cm congestion
1724.Pq Dv IPTOS_ECN_CE .
1725The absence of a particular type may be denoted
1726with a
1727.Ql \&! .
1728.It Cm dscp spec Ns Op , Ns Ar spec
1729Matches IPv4/IPv6 packets whose
1730.Cm DS
1731field value is contained in
1732.Ar spec
1733mask.
1734Multiple values can be specified via
1735the comma separated list.
1736Value can be one of keywords used in
1737.Cm setdscp
1738action or exact number.
1739.It Cm ipttl Ar ttl-list
1740Matches IPv4 packets whose time to live is included in
1741.Ar ttl-list ,
1742which is either a single value or a list of values or ranges
1743specified in the same way as
1744.Ar ports .
1745.It Cm ipversion Ar ver
1746Matches IP packets whose IP version field is
1747.Ar ver .
1748.It Cm keep-state Op Ar :flowname
1749Upon a match, the firewall will create a dynamic rule, whose
1750default behaviour is to match bidirectional traffic between
1751source and destination IP/port using the same protocol.
1752The rule has a limited lifetime (controlled by a set of
1753.Xr sysctl 8
1754variables), and the lifetime is refreshed every time a matching
1755packet is found.
1756The
1757.Ar :flowname
1758is used to assign additional to addresses, ports and protocol parameter
1759to dynamic rule. It can be used for more accurate matching by
1760.Cm check-state
1761rule.
1762The
1763.Cm :default
1764keyword is special name used for compatibility with old rulesets.
1765.It Cm layer2
1766Matches only layer2 packets, i.e., those passed to
1767.Nm
1768from ether_demux() and ether_output_frame().
1769.It Cm limit Bro Cm src-addr | src-port | dst-addr | dst-port Brc Ar N Op Ar :flowname
1770The firewall will only allow
1771.Ar N
1772connections with the same
1773set of parameters as specified in the rule.
1774One or more
1775of source and destination addresses and ports can be
1776specified.
1777.It Cm lookup Bro Cm dst-ip | dst-port | src-ip | src-port | uid | jail Brc Ar name
1778Search an entry in lookup table
1779.Ar name
1780that matches the field specified as argument.
1781If not found, the match fails.
1782Otherwise, the match succeeds and
1783.Cm tablearg
1784is set to the value extracted from the table.
1785.Pp
1786This option can be useful to quickly dispatch traffic based on
1787certain packet fields.
1788See the
1789.Sx LOOKUP TABLES
1790section below for more information on lookup tables.
1791.It Cm { MAC | mac } Ar dst-mac src-mac
1792Match packets with a given
1793.Ar dst-mac
1794and
1795.Ar src-mac
1796addresses, specified as the
1797.Cm any
1798keyword (matching any MAC address), or six groups of hex digits
1799separated by colons,
1800and optionally followed by a mask indicating the significant bits.
1801The mask may be specified using either of the following methods:
1802.Bl -enum -width indent
1803.It
1804A slash
1805.Pq /
1806followed by the number of significant bits.
1807For example, an address with 33 significant bits could be specified as:
1808.Pp
1809.Dl "MAC 10:20:30:40:50:60/33 any"
1810.It
1811An ampersand
1812.Pq &
1813followed by a bitmask specified as six groups of hex digits separated
1814by colons.
1815For example, an address in which the last 16 bits are significant could
1816be specified as:
1817.Pp
1818.Dl "MAC 10:20:30:40:50:60&00:00:00:00:ff:ff any"
1819.Pp
1820Note that the ampersand character has a special meaning in many shells
1821and should generally be escaped.
1822.El
1823Note that the order of MAC addresses (destination first,
1824source second) is
1825the same as on the wire, but the opposite of the one used for
1826IP addresses.
1827.It Cm mac-type Ar mac-type
1828Matches packets whose Ethernet Type field
1829corresponds to one of those specified as argument.
1830.Ar mac-type
1831is specified in the same way as
1832.Cm port numbers
1833(i.e., one or more comma-separated single values or ranges).
1834You can use symbolic names for known values such as
1835.Em vlan , ipv4, ipv6 .
1836Values can be entered as decimal or hexadecimal (if prefixed by 0x),
1837and they are always printed as hexadecimal (unless the
1838.Cm -N
1839option is used, in which case symbolic resolution will be attempted).
1840.It Cm proto Ar protocol
1841Matches packets with the corresponding IP protocol.
1842.It Cm record-state
1843Upon a match, the firewall will create a dynamic rule as if
1844.Cm keep-state
1845was specified.
1846However, this option doesn't imply an implicit
1847.Cm check-state
1848in contrast to
1849.Cm keep-state .
1850.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
1851Matches packets received, transmitted or going through,
1852respectively, the interface specified by exact name
1853.Po Ar ifX Pc ,
1854by device name
1855.Po Ar if* Pc ,
1856by IP address, or through some interface.
1857Table
1858.Ar name
1859may be used to match interface by its kernel ifindex.
1860See the
1861.Sx LOOKUP TABLES
1862section below for more information on lookup tables.
1863.Pp
1864The
1865.Cm via
1866keyword causes the interface to always be checked.
1867If
1868.Cm recv
1869or
1870.Cm xmit
1871is used instead of
1872.Cm via ,
1873then only the receive or transmit interface (respectively)
1874is checked.
1875By specifying both, it is possible to match packets based on
1876both receive and transmit interface, e.g.:
1877.Pp
1878.Dl "ipfw add deny ip from any to any out recv ed0 xmit ed1"
1879.Pp
1880The
1881.Cm recv
1882interface can be tested on either incoming or outgoing packets,
1883while the
1884.Cm xmit
1885interface can only be tested on outgoing packets.
1886So
1887.Cm out
1888is required (and
1889.Cm in
1890is invalid) whenever
1891.Cm xmit
1892is used.
1893.Pp
1894A packet might not have a receive or transmit interface: packets
1895originating from the local host have no receive interface,
1896while packets destined for the local host have no transmit
1897interface.
1898.It Cm set-limit Bro Cm src-addr | src-port | dst-addr | dst-port Brc Ar N
1899Works like
1900.Cm limit
1901but does not have an implicit
1902.Cm check-state
1903attached to it.
1904.It Cm setup
1905Matches TCP packets that have the SYN bit set but no ACK bit.
1906This is the short form of
1907.Dq Li tcpflags\ syn,!ack .
1908.It Cm sockarg
1909Matches packets that are associated to a local socket and
1910for which the SO_USER_COOKIE socket option has been set
1911to a non-zero value.
1912As a side effect, the value of the
1913option is made available as
1914.Cm tablearg
1915value, which in turn can be used as
1916.Cm skipto
1917or
1918.Cm pipe
1919number.
1920.It Cm src-ip Ar ip-address
1921Matches IPv4 packets whose source IP is one of the address(es)
1922specified as an argument.
1923.It Cm src-ip6 Ar ip6-address
1924Matches IPv6 packets whose source IP is one of the address(es)
1925specified as an argument.
1926.It Cm src-port Ar ports
1927Matches IP packets whose source port is one of the port(s)
1928specified as argument.
1929.It Cm tagged Ar tag-list
1930Matches packets whose tags are included in
1931.Ar tag-list ,
1932which is either a single value or a list of values or ranges
1933specified in the same way as
1934.Ar ports .
1935Tags can be applied to the packet using
1936.Cm tag
1937rule action parameter (see it's description for details on tags).
1938.It Cm tcpack Ar ack
1939TCP packets only.
1940Match if the TCP header acknowledgment number field is set to
1941.Ar ack .
1942.It Cm tcpdatalen Ar tcpdatalen-list
1943Matches TCP packets whose length of TCP data is
1944.Ar tcpdatalen-list ,
1945which is either a single value or a list of values or ranges
1946specified in the same way as
1947.Ar ports .
1948.It Cm tcpflags Ar spec
1949TCP packets only.
1950Match if the TCP header contains the comma separated list of
1951flags specified in
1952.Ar spec .
1953The supported TCP flags are:
1954.Pp
1955.Cm fin ,
1956.Cm syn ,
1957.Cm rst ,
1958.Cm psh ,
1959.Cm ack
1960and
1961.Cm urg .
1962The absence of a particular flag may be denoted
1963with a
1964.Ql \&! .
1965A rule which contains a
1966.Cm tcpflags
1967specification can never match a fragmented packet which has
1968a non-zero offset.
1969See the
1970.Cm frag
1971option for details on matching fragmented packets.
1972.It Cm tcpseq Ar seq
1973TCP packets only.
1974Match if the TCP header sequence number field is set to
1975.Ar seq .
1976.It Cm tcpwin Ar tcpwin-list
1977Matches TCP packets whose  header window field is set to
1978.Ar tcpwin-list ,
1979which is either a single value or a list of values or ranges
1980specified in the same way as
1981.Ar ports .
1982.It Cm tcpoptions Ar spec
1983TCP packets only.
1984Match if the TCP header contains the comma separated list of
1985options specified in
1986.Ar spec .
1987The supported TCP options are:
1988.Pp
1989.Cm mss
1990(maximum segment size),
1991.Cm window
1992(tcp window advertisement),
1993.Cm sack
1994(selective ack),
1995.Cm ts
1996(rfc1323 timestamp) and
1997.Cm cc
1998(rfc1644 t/tcp connection count).
1999The absence of a particular option may be denoted
2000with a
2001.Ql \&! .
2002.It Cm uid Ar user
2003Match all TCP or UDP packets sent by or received for a
2004.Ar user .
2005A
2006.Ar user
2007may be matched by name or identification number.
2008.It Cm verrevpath
2009For incoming packets,
2010a routing table lookup is done on the packet's source address.
2011If the interface on which the packet entered the system matches the
2012outgoing interface for the route,
2013the packet matches.
2014If the interfaces do not match up,
2015the packet does not match.
2016All outgoing packets or packets with no incoming interface match.
2017.Pp
2018The name and functionality of the option is intentionally similar to
2019the Cisco IOS command:
2020.Pp
2021.Dl ip verify unicast reverse-path
2022.Pp
2023This option can be used to make anti-spoofing rules to reject all
2024packets with source addresses not from this interface.
2025See also the option
2026.Cm antispoof .
2027.It Cm versrcreach
2028For incoming packets,
2029a routing table lookup is done on the packet's source address.
2030If a route to the source address exists, but not the default route
2031or a blackhole/reject route, the packet matches.
2032Otherwise, the packet does not match.
2033All outgoing packets match.
2034.Pp
2035The name and functionality of the option is intentionally similar to
2036the Cisco IOS command:
2037.Pp
2038.Dl ip verify unicast source reachable-via any
2039.Pp
2040This option can be used to make anti-spoofing rules to reject all
2041packets whose source address is unreachable.
2042.It Cm antispoof
2043For incoming packets, the packet's source address is checked if it
2044belongs to a directly connected network.
2045If the network is directly connected, then the interface the packet
2046came on in is compared to the interface the network is connected to.
2047When incoming interface and directly connected interface are not the
2048same, the packet does not match.
2049Otherwise, the packet does match.
2050All outgoing packets match.
2051.Pp
2052This option can be used to make anti-spoofing rules to reject all
2053packets that pretend to be from a directly connected network but do
2054not come in through that interface.
2055This option is similar to but more restricted than
2056.Cm verrevpath
2057because it engages only on packets with source addresses of directly
2058connected networks instead of all source addresses.
2059.El
2060.Sh LOOKUP TABLES
2061Lookup tables are useful to handle large sparse sets of
2062addresses or other search keys (e.g., ports, jail IDs, interface names).
2063In the rest of this section we will use the term ``key''.
2064Table name needs to match the following spec:
2065.Ar table-name .
2066Tables with the same name can be created in different
2067.Ar sets .
2068However, rule links to the tables in
2069.Ar set 0
2070by default.
2071This behavior can be controlled by
2072.Va net.inet.ip.fw.tables_sets
2073variable.
2074See the
2075.Sx SETS OF RULES
2076section for more information.
2077There may be up to 65535 different lookup tables.
2078.Pp
2079The following table types are supported:
2080.Bl -tag -width indent
2081.It Ar table-type : Ar addr | iface | number | flow
2082.It Ar table-key : Ar addr Ns Oo / Ns Ar masklen Oc | iface-name | number | flow-spec
2083.It Ar flow-spec : Ar flow-field Ns Op , Ns Ar flow-spec
2084.It Ar flow-field : src-ip | proto | src-port | dst-ip | dst-port
2085.It Cm addr
2086Matches IPv4 or IPv6 address.
2087Each entry is represented by an
2088.Ar addr Ns Op / Ns Ar masklen
2089and will match all addresses with base
2090.Ar addr
2091(specified as an IPv4/IPv6 address, or a hostname) and mask width of
2092.Ar masklen
2093bits.
2094If
2095.Ar masklen
2096is not specified, it defaults to 32 for IPv4 and 128 for IPv6.
2097When looking up an IP address in a table, the most specific
2098entry will match.
2099.It Cm iface
2100Matches interface names.
2101Each entry is represented by string treated as interface name.
2102Wildcards are not supported.
2103.It Cm number
2104Matches protocol ports, uids/gids or jail IDs.
2105Each entry is represented by 32-bit unsigned integer.
2106Ranges are not supported.
2107.It Cm flow
2108Matches packet fields specified by
2109.Ar flow
2110type suboptions with table entries.
2111.El
2112.Pp
2113Tables require explicit creation via
2114.Cm create
2115before use.
2116.Pp
2117The following creation options are supported:
2118.Bl -tag -width indent
2119.It Ar create-options : Ar create-option | create-options
2120.It Ar create-option : Cm type Ar table-type | Cm valtype Ar value-mask | Cm algo Ar algo-desc |
2121.Cm limit Ar number | Cm locked
2122.It Cm type
2123Table key type.
2124.It Cm valtype
2125Table value mask.
2126.It Cm algo
2127Table algorithm to use (see below).
2128.It Cm limit
2129Maximum number of items that may be inserted into table.
2130.It Cm locked
2131Restrict any table modifications.
2132.El
2133.Pp
2134Some of these options may be modified later via
2135.Cm modify
2136keyword.
2137The following options can be changed:
2138.Bl -tag -width indent
2139.It Ar modify-options : Ar modify-option | modify-options
2140.It Ar modify-option : Cm limit Ar number
2141.It Cm limit
2142Alter maximum number of items that may be inserted into table.
2143.El
2144.Pp
2145Additionally, table can be locked or unlocked using
2146.Cm lock
2147or
2148.Cm unlock
2149commands.
2150.Pp
2151Tables of the same
2152.Ar type
2153can be swapped with each other using
2154.Cm swap Ar name
2155command.
2156Swap may fail if tables limits are set and data exchange
2157would result in limits hit.
2158Operation is performed atomically.
2159.Pp
2160One or more entries can be added to a table at once using
2161.Cm add
2162command.
2163Addition of all items are performed atomically.
2164By default, error in addition of one entry does not influence
2165addition of other entries. However, non-zero error code is returned
2166in that case.
2167Special
2168.Cm atomic
2169keyword may be specified before
2170.Cm add
2171to indicate all-or-none add request.
2172.Pp
2173One or more entries can be removed from a table at once using
2174.Cm delete
2175command.
2176By default, error in removal of one entry does not influence
2177removing of other entries. However, non-zero error code is returned
2178in that case.
2179.Pp
2180It may be possible to check what entry will be found on particular
2181.Ar table-key
2182using
2183.Cm lookup
2184.Ar table-key
2185command.
2186This functionality is optional and may be unsupported in some algorithms.
2187.Pp
2188The following operations can be performed on
2189.Ar one
2190or
2191.Cm all
2192tables:
2193.Bl -tag -width indent
2194.It Cm list
2195List all entries.
2196.It Cm flush
2197Removes all entries.
2198.It Cm info
2199Shows generic table information.
2200.It Cm detail
2201Shows generic table information and algo-specific data.
2202.El
2203.Pp
2204The following lookup algorithms are supported:
2205.Bl -tag -width indent
2206.It Ar algo-desc : algo-name | "algo-name algo-data"
2207.It Ar algo-name: Ar addr:radix | addr:hash | iface:array | number:array | flow:hash
2208.It Cm addr:radix
2209Separate Radix trees for IPv4 and IPv6, the same way as the routing table (see
2210.Xr route 4 ) .
2211Default choice for
2212.Ar addr
2213type.
2214.It Cm addr:hash
2215Separate auto-growing hashes for IPv4 and IPv6.
2216Accepts entries with the same mask length specified initially via
2217.Cm "addr:hash masks=/v4,/v6"
2218algorithm creation options.
2219Assume /32 and /128 masks by default.
2220Search removes host bits (according to mask) from supplied address and checks
2221resulting key in appropriate hash.
2222Mostly optimized for /64 and byte-ranged IPv6 masks.
2223.It Cm iface:array
2224Array storing sorted indexes for entries which are presented in the system.
2225Optimized for very fast lookup.
2226.It Cm number:array
2227Array storing sorted u32 numbers.
2228.It Cm flow:hash
2229Auto-growing hash storing flow entries.
2230Search calculates hash on required packet fields and searches for matching
2231entries in selected bucket.
2232.El
2233.Pp
2234The
2235.Cm tablearg
2236feature provides the ability to use a value, looked up in the table, as
2237the argument for a rule action, action parameter or rule option.
2238This can significantly reduce number of rules in some configurations.
2239If two tables are used in a rule, the result of the second (destination)
2240is used.
2241.Pp
2242Each record may hold one or more values according to
2243.Ar value-mask .
2244This mask is set on table creation via
2245.Cm valtype
2246option.
2247The following value types are supported:
2248.Bl -tag -width indent
2249.It Ar value-mask : Ar value-type Ns Op , Ns Ar value-mask
2250.It Ar value-type : Ar skipto | pipe | fib | nat | dscp | tag | divert |
2251.Ar netgraph | limit | ipv4
2252.It Cm skipto
2253rule number to jump to.
2254.It Cm pipe
2255Pipe number to use.
2256.It Cm fib
2257fib number to match/set.
2258.It Cm nat
2259nat number to jump to.
2260.It Cm dscp
2261dscp value to match/set.
2262.It Cm tag
2263tag number to match/set.
2264.It Cm divert
2265port number to divert traffic to.
2266.It Cm netgraph
2267hook number to move packet to.
2268.It Cm limit
2269maximum number of connections.
2270.It Cm ipv4
2271IPv4 nexthop to fwd packets to.
2272.It Cm ipv6
2273IPv6 nexthop to fwd packets to.
2274.El
2275.Pp
2276The
2277.Cm tablearg
2278argument can be used with the following actions:
2279.Cm nat, pipe , queue, divert, tee, netgraph, ngtee, fwd, skipto, setfib,
2280action parameters:
2281.Cm tag, untag,
2282rule options:
2283.Cm limit, tagged.
2284.Pp
2285When used with the
2286.Cm skipto
2287action, the user should be aware that the code will walk the ruleset
2288up to a rule equal to, or past, the given number.
2289.Pp
2290See the
2291.Sx EXAMPLES
2292Section for example usage of tables and the tablearg keyword.
2293.Sh SETS OF RULES
2294Each rule or table belongs to one of 32 different
2295.Em sets
2296, numbered 0 to 31.
2297Set 31 is reserved for the default rule.
2298.Pp
2299By default, rules or tables are put in set 0, unless you use the
2300.Cm set N
2301attribute when adding a new rule or table.
2302Sets can be individually and atomically enabled or disabled,
2303so this mechanism permits an easy way to store multiple configurations
2304of the firewall and quickly (and atomically) switch between them.
2305.Pp
2306By default, tables from set 0 are referenced when adding rule with
2307table opcodes regardless of rule set.
2308This behavior can be changed by setting
2309.Va net.inet.ip.fw.tables_sets
2310variable to 1.
2311Rule's set will then be used for table references.
2312.Pp
2313The command to enable/disable sets is
2314.Bd -ragged -offset indent
2315.Nm
2316.Cm set Oo Cm disable Ar number ... Oc Op Cm enable Ar number ...
2317.Ed
2318.Pp
2319where multiple
2320.Cm enable
2321or
2322.Cm disable
2323sections can be specified.
2324Command execution is atomic on all the sets specified in the command.
2325By default, all sets are enabled.
2326.Pp
2327When you disable a set, its rules behave as if they do not exist
2328in the firewall configuration, with only one exception:
2329.Bd -ragged -offset indent
2330dynamic rules created from a rule before it had been disabled
2331will still be active until they expire.
2332In order to delete
2333dynamic rules you have to explicitly delete the parent rule
2334which generated them.
2335.Ed
2336.Pp
2337The set number of rules can be changed with the command
2338.Bd -ragged -offset indent
2339.Nm
2340.Cm set move
2341.Brq Cm rule Ar rule-number | old-set
2342.Cm to Ar new-set
2343.Ed
2344.Pp
2345Also, you can atomically swap two rulesets with the command
2346.Bd -ragged -offset indent
2347.Nm
2348.Cm set swap Ar first-set second-set
2349.Ed
2350.Pp
2351See the
2352.Sx EXAMPLES
2353Section on some possible uses of sets of rules.
2354.Sh STATEFUL FIREWALL
2355Stateful operation is a way for the firewall to dynamically
2356create rules for specific flows when packets that
2357match a given pattern are detected.
2358Support for stateful
2359operation comes through the
2360.Cm check-state , keep-state , record-state , limit
2361and
2362.Cm set-limit
2363options of
2364.Nm rules .
2365.Pp
2366Dynamic rules are created when a packet matches a
2367.Cm keep-state ,
2368.Cm record-state ,
2369.Cm limit
2370or
2371.Cm set-limit
2372rule, causing the creation of a
2373.Em dynamic
2374rule which will match all and only packets with
2375a given
2376.Em protocol
2377between a
2378.Em src-ip/src-port dst-ip/dst-port
2379pair of addresses
2380.Em ( src
2381and
2382.Em dst
2383are used here only to denote the initial match addresses, but they
2384are completely equivalent afterwards).
2385Rules created by
2386.Cm keep-state
2387option also have a
2388.Ar :flowname
2389taken from it.
2390This name is used in matching together with addresses, ports and protocol.
2391Dynamic rules will be checked at the first
2392.Cm check-state, keep-state
2393or
2394.Cm limit
2395occurrence, and the action performed upon a match will be the same
2396as in the parent rule.
2397.Pp
2398Note that no additional attributes other than protocol and IP addresses
2399and ports and :flowname are checked on dynamic rules.
2400.Pp
2401The typical use of dynamic rules is to keep a closed firewall configuration,
2402but let the first TCP SYN packet from the inside network install a
2403dynamic rule for the flow so that packets belonging to that session
2404will be allowed through the firewall:
2405.Pp
2406.Dl "ipfw add check-state :OUTBOUND"
2407.Dl "ipfw add allow tcp from my-subnet to any setup keep-state :OUTBOUND"
2408.Dl "ipfw add deny tcp from any to any"
2409.Pp
2410A similar approach can be used for UDP, where an UDP packet coming
2411from the inside will install a dynamic rule to let the response through
2412the firewall:
2413.Pp
2414.Dl "ipfw add check-state :OUTBOUND"
2415.Dl "ipfw add allow udp from my-subnet to any keep-state :OUTBOUND"
2416.Dl "ipfw add deny udp from any to any"
2417.Pp
2418Dynamic rules expire after some time, which depends on the status
2419of the flow and the setting of some
2420.Cm sysctl
2421variables.
2422See Section
2423.Sx SYSCTL VARIABLES
2424for more details.
2425For TCP sessions, dynamic rules can be instructed to periodically
2426send keepalive packets to refresh the state of the rule when it is
2427about to expire.
2428.Pp
2429See Section
2430.Sx EXAMPLES
2431for more examples on how to use dynamic rules.
2432.Sh TRAFFIC SHAPER (DUMMYNET) CONFIGURATION
2433.Nm
2434is also the user interface for the
2435.Nm dummynet
2436traffic shaper, packet scheduler and network emulator, a subsystem that
2437can artificially queue, delay or drop packets
2438emulating the behaviour of certain network links
2439or queueing systems.
2440.Pp
2441.Nm dummynet
2442operates by first using the firewall to select packets
2443using any match pattern that can be used in
2444.Nm
2445rules.
2446Matching packets are then passed to either of two
2447different objects, which implement the traffic regulation:
2448.Bl -hang -offset XXXX
2449.It Em pipe
2450A
2451.Em pipe
2452emulates a
2453.Em link
2454with given bandwidth and propagation delay,
2455driven by a FIFO scheduler and a single queue with programmable
2456queue size and packet loss rate.
2457Packets are appended to the queue as they come out from
2458.Nm ipfw ,
2459and then transferred in FIFO order to the link at the desired rate.
2460.It Em queue
2461A
2462.Em queue
2463is an abstraction used to implement packet scheduling
2464using one of several packet scheduling algorithms.
2465Packets sent to a
2466.Em queue
2467are first grouped into flows according to a mask on the 5-tuple.
2468Flows are then passed to the scheduler associated to the
2469.Em queue ,
2470and each flow uses scheduling parameters (weight and others)
2471as configured in the
2472.Em queue
2473itself.
2474A scheduler in turn is connected to an emulated link,
2475and arbitrates the link's bandwidth among backlogged flows according to
2476weights and to the features of the scheduling algorithm in use.
2477.El
2478.Pp
2479In practice,
2480.Em pipes
2481can be used to set hard limits to the bandwidth that a flow can use, whereas
2482.Em queues
2483can be used to determine how different flows share the available bandwidth.
2484.Pp
2485A graphical representation of the binding of queues,
2486flows, schedulers and links is below.
2487.Bd -literal -offset indent
2488                 (flow_mask|sched_mask)  sched_mask
2489         +---------+   weight Wx  +-------------+
2490         |         |->-[flow]-->--|             |-+
2491    -->--| QUEUE x |   ...        |             | |
2492         |         |->-[flow]-->--| SCHEDuler N | |
2493         +---------+              |             | |
2494             ...                  |             +--[LINK N]-->--
2495         +---------+   weight Wy  |             | +--[LINK N]-->--
2496         |         |->-[flow]-->--|             | |
2497    -->--| QUEUE y |   ...        |             | |
2498         |         |->-[flow]-->--|             | |
2499         +---------+              +-------------+ |
2500                                    +-------------+
2501.Ed
2502It is important to understand the role of the SCHED_MASK
2503and FLOW_MASK, which are configured through the commands
2504.Dl "ipfw sched N config mask SCHED_MASK ..."
2505and
2506.Dl "ipfw queue X config mask FLOW_MASK ..." .
2507.Pp
2508The SCHED_MASK is used to assign flows to one or more
2509scheduler instances, one for each
2510value of the packet's 5-tuple after applying SCHED_MASK.
2511As an example, using ``src-ip 0xffffff00'' creates one instance
2512for each /24 destination subnet.
2513.Pp
2514The FLOW_MASK, together with the SCHED_MASK, is used to split
2515packets into flows.
2516As an example, using
2517``src-ip 0x000000ff''
2518together with the previous SCHED_MASK makes a flow for
2519each individual source address.
2520In turn, flows for each /24
2521subnet will be sent to the same scheduler instance.
2522.Pp
2523The above diagram holds even for the
2524.Em pipe
2525case, with the only restriction that a
2526.Em pipe
2527only supports a SCHED_MASK, and forces the use of a FIFO
2528scheduler (these are for backward compatibility reasons;
2529in fact, internally, a
2530.Nm dummynet's
2531pipe is implemented exactly as above).
2532.Pp
2533There are two modes of
2534.Nm dummynet
2535operation:
2536.Dq normal
2537and
2538.Dq fast .
2539The
2540.Dq normal
2541mode tries to emulate a real link: the
2542.Nm dummynet
2543scheduler ensures that the packet will not leave the pipe faster than it
2544would on the real link with a given bandwidth.
2545The
2546.Dq fast
2547mode allows certain packets to bypass the
2548.Nm dummynet
2549scheduler (if packet flow does not exceed pipe's bandwidth).
2550This is the reason why the
2551.Dq fast
2552mode requires less CPU cycles per packet (on average) and packet latency
2553can be significantly lower in comparison to a real link with the same
2554bandwidth.
2555The default mode is
2556.Dq normal .
2557The
2558.Dq fast
2559mode can be enabled by setting the
2560.Va net.inet.ip.dummynet.io_fast
2561.Xr sysctl 8
2562variable to a non-zero value.
2563.Pp
2564.Ss PIPE, QUEUE AND SCHEDULER CONFIGURATION
2565The
2566.Em pipe ,
2567.Em queue
2568and
2569.Em scheduler
2570configuration commands are the following:
2571.Bd -ragged -offset indent
2572.Cm pipe Ar number Cm config Ar pipe-configuration
2573.Pp
2574.Cm queue Ar number Cm config Ar queue-configuration
2575.Pp
2576.Cm sched Ar number Cm config Ar sched-configuration
2577.Ed
2578.Pp
2579The following parameters can be configured for a pipe:
2580.Pp
2581.Bl -tag -width indent -compact
2582.It Cm bw Ar bandwidth | device
2583Bandwidth, measured in
2584.Sm off
2585.Op Cm K | M | G
2586.Brq Cm bit/s | Byte/s .
2587.Sm on
2588.Pp
2589A value of 0 (default) means unlimited bandwidth.
2590The unit must immediately follow the number, as in
2591.Pp
2592.Dl "ipfw pipe 1 config bw 300Kbit/s"
2593.Pp
2594If a device name is specified instead of a numeric value, as in
2595.Pp
2596.Dl "ipfw pipe 1 config bw tun0"
2597.Pp
2598then the transmit clock is supplied by the specified device.
2599At the moment only the
2600.Xr tun 4
2601device supports this
2602functionality, for use in conjunction with
2603.Xr ppp 8 .
2604.Pp
2605.It Cm delay Ar ms-delay
2606Propagation delay, measured in milliseconds.
2607The value is rounded to the next multiple of the clock tick
2608(typically 10ms, but it is a good practice to run kernels
2609with
2610.Dq "options HZ=1000"
2611to reduce
2612the granularity to 1ms or less).
2613The default value is 0, meaning no delay.
2614.Pp
2615.It Cm burst Ar size
2616If the data to be sent exceeds the pipe's bandwidth limit
2617(and the pipe was previously idle), up to
2618.Ar size
2619bytes of data are allowed to bypass the
2620.Nm dummynet
2621scheduler, and will be sent as fast as the physical link allows.
2622Any additional data will be transmitted at the rate specified
2623by the
2624.Nm pipe
2625bandwidth.
2626The burst size depends on how long the pipe has been idle;
2627the effective burst size is calculated as follows:
2628MAX(
2629.Ar size
2630,
2631.Nm bw
2632* pipe_idle_time).
2633.Pp
2634.It Cm profile Ar filename
2635A file specifying the additional overhead incurred in the transmission
2636of a packet on the link.
2637.Pp
2638Some link types introduce extra delays in the transmission
2639of a packet, e.g., because of MAC level framing, contention on
2640the use of the channel, MAC level retransmissions and so on.
2641From our point of view, the channel is effectively unavailable
2642for this extra time, which is constant or variable depending
2643on the link type.
2644Additionally, packets may be dropped after this
2645time (e.g., on a wireless link after too many retransmissions).
2646We can model the additional delay with an empirical curve
2647that represents its distribution.
2648.Bd -literal -offset indent
2649      cumulative probability
2650      1.0 ^
2651          |
2652      L   +-- loss-level          x
2653          |                 ******
2654          |                *
2655          |           *****
2656          |          *
2657          |        **
2658          |       *
2659          +-------*------------------->
2660                      delay
2661.Ed
2662The empirical curve may have both vertical and horizontal lines.
2663Vertical lines represent constant delay for a range of
2664probabilities.
2665Horizontal lines correspond to a discontinuity in the delay
2666distribution: the pipe will use the largest delay for a
2667given probability.
2668.Pp
2669The file format is the following, with whitespace acting as
2670a separator and '#' indicating the beginning a comment:
2671.Bl -tag -width indent
2672.It Cm name Ar identifier
2673optional name (listed by "ipfw pipe show")
2674to identify the delay distribution;
2675.It Cm bw Ar value
2676the bandwidth used for the pipe.
2677If not specified here, it must be present
2678explicitly as a configuration parameter for the pipe;
2679.It Cm loss-level Ar L
2680the probability above which packets are lost.
2681(0.0 <= L <= 1.0, default 1.0 i.e., no loss);
2682.It Cm samples Ar N
2683the number of samples used in the internal
2684representation of the curve (2..1024; default 100);
2685.It Cm "delay prob" | "prob delay"
2686One of these two lines is mandatory and defines
2687the format of the following lines with data points.
2688.It Ar XXX Ar YYY
26892 or more lines representing points in the curve,
2690with either delay or probability first, according
2691to the chosen format.
2692The unit for delay is milliseconds.
2693Data points do not need to be sorted.
2694Also, the number of actual lines can be different
2695from the value of the "samples" parameter:
2696.Nm
2697utility will sort and interpolate
2698the curve as needed.
2699.El
2700.Pp
2701Example of a profile file:
2702.Bd -literal -offset indent
2703name    bla_bla_bla
2704samples 100
2705loss-level    0.86
2706prob    delay
27070       200	# minimum overhead is 200ms
27080.5     200
27090.5     300
27100.8     1000
27110.9     1300
27121       1300
2713#configuration file end
2714.Ed
2715.El
2716.Pp
2717The following parameters can be configured for a queue:
2718.Pp
2719.Bl -tag -width indent -compact
2720.It Cm pipe Ar pipe_nr
2721Connects a queue to the specified pipe.
2722Multiple queues (with the same or different weights) can be connected to
2723the same pipe, which specifies the aggregate rate for the set of queues.
2724.Pp
2725.It Cm weight Ar weight
2726Specifies the weight to be used for flows matching this queue.
2727The weight must be in the range 1..100, and defaults to 1.
2728.El
2729.Pp
2730The following case-insensitive parameters can be configured for a
2731scheduler:
2732.Pp
2733.Bl -tag -width indent -compact
2734.It Cm type Ar {fifo | wf2q+ | rr | qfq | fq_codel | fq_pie}
2735specifies the scheduling algorithm to use.
2736.Bl -tag -width indent -compact
2737.It Cm fifo
2738is just a FIFO scheduler (which means that all packets
2739are stored in the same queue as they arrive to the scheduler).
2740FIFO has O(1) per-packet time complexity, with very low
2741constants (estimate 60-80ns on a 2GHz desktop machine)
2742but gives no service guarantees.
2743.It Cm wf2q+
2744implements the WF2Q+ algorithm, which is a Weighted Fair Queueing
2745algorithm which permits flows to share bandwidth according to
2746their weights.
2747Note that weights are not priorities; even a flow
2748with a minuscule weight will never starve.
2749WF2Q+ has O(log N) per-packet processing cost, where N is the number
2750of flows, and is the default algorithm used by previous versions
2751dummynet's queues.
2752.It Cm rr
2753implements the Deficit Round Robin algorithm, which has O(1) processing
2754costs (roughly, 100-150ns per packet)
2755and permits bandwidth allocation according to weights, but
2756with poor service guarantees.
2757.It Cm qfq
2758implements the QFQ algorithm, which is a very fast variant of
2759WF2Q+, with similar service guarantees and O(1) processing
2760costs (roughly, 200-250ns per packet).
2761.It Cm fq_codel
2762implements the FQ-CoDel (FlowQueue-CoDel) scheduler/AQM algorithm, which
2763uses a modified Deficit Round Robin scheduler to manage two lists of sub-queues
2764(old sub-queues and new sub-queues) for providing brief periods of priority to
2765lightweight or short burst flows.
2766By default, the total number of sub-queues is 1024.
2767FQ-CoDel's internal, dynamically
2768created sub-queues are controlled by separate instances of CoDel AQM.
2769.It Cm fq_pie
2770implements the FQ-PIE (FlowQueue-PIE) scheduler/AQM algorithm, which similar to
2771.Cm fq_codel
2772but uses per sub-queue PIE AQM instance to control the queue delay.
2773.El
2774.Pp
2775.Cm fq_codel
2776inherits AQM parameters and options from
2777.Cm codel
2778(see below), and
2779.Cm fq_pie
2780inherits AQM parameters and options from
2781.Cm pie
2782(see below).
2783Additionally, both of
2784.Cm fq_codel
2785and
2786.Cm fq_pie
2787have shared scheduler parameters which are:
2788.Bl -tag -width indent
2789.It Cm quantum
2790.Ar m
2791specifies the quantum (credit) of the scheduler.
2792.Ar m
2793is the number of bytes a queue can serve before being moved to the tail
2794of old queues list.
2795The default is 1514 bytes, and the maximum acceptable value
2796is 9000 bytes.
2797.It Cm limit
2798.Ar m
2799specifies the hard size limit (in unit of packets) of all queues managed by an
2800instance of the scheduler.
2801The default value of
2802.Ar m
2803is 10240 packets, and the maximum acceptable value is 20480 packets.
2804.It Cm flows
2805.Ar m
2806specifies the total number of flow queues (sub-queues) that fq_*
2807creates and manages.
2808By default, 1024 sub-queues are created when an instance
2809of the fq_{codel/pie} scheduler is created.
2810The maximum acceptable value is
281165536.
2812.El
2813.Pp
2814Note that any token after
2815.Cm fq_codel
2816or
2817.Cm fq_pie
2818is considered a parameter for fq_{codel/pie}.
2819So, ensure all scheduler
2820configuration options not related to fq_{codel/pie} are written before
2821.Cm fq_codel/fq_pie
2822tokens.
2823.El
2824.Pp
2825In addition to the type, all parameters allowed for a pipe can also
2826be specified for a scheduler.
2827.Pp
2828Finally, the following parameters can be configured for both
2829pipes and queues:
2830.Pp
2831.Bl -tag -width XXXX -compact
2832.It Cm buckets Ar hash-table-size
2833Specifies the size of the hash table used for storing the
2834various queues.
2835Default value is 64 controlled by the
2836.Xr sysctl 8
2837variable
2838.Va net.inet.ip.dummynet.hash_size ,
2839allowed range is 16 to 65536.
2840.Pp
2841.It Cm mask Ar mask-specifier
2842Packets sent to a given pipe or queue by an
2843.Nm
2844rule can be further classified into multiple flows, each of which is then
2845sent to a different
2846.Em dynamic
2847pipe or queue.
2848A flow identifier is constructed by masking the IP addresses,
2849ports and protocol types as specified with the
2850.Cm mask
2851options in the configuration of the pipe or queue.
2852For each different flow identifier, a new pipe or queue is created
2853with the same parameters as the original object, and matching packets
2854are sent to it.
2855.Pp
2856Thus, when
2857.Em dynamic pipes
2858are used, each flow will get the same bandwidth as defined by the pipe,
2859whereas when
2860.Em dynamic queues
2861are used, each flow will share the parent's pipe bandwidth evenly
2862with other flows generated by the same queue (note that other queues
2863with different weights might be connected to the same pipe).
2864.br
2865Available mask specifiers are a combination of one or more of the following:
2866.Pp
2867.Cm dst-ip Ar mask ,
2868.Cm dst-ip6 Ar mask ,
2869.Cm src-ip Ar mask ,
2870.Cm src-ip6 Ar mask ,
2871.Cm dst-port Ar mask ,
2872.Cm src-port Ar mask ,
2873.Cm flow-id Ar mask ,
2874.Cm proto Ar mask
2875or
2876.Cm all ,
2877.Pp
2878where the latter means all bits in all fields are significant.
2879.Pp
2880.It Cm noerror
2881When a packet is dropped by a
2882.Nm dummynet
2883queue or pipe, the error
2884is normally reported to the caller routine in the kernel, in the
2885same way as it happens when a device queue fills up.
2886Setting this
2887option reports the packet as successfully delivered, which can be
2888needed for some experimental setups where you want to simulate
2889loss or congestion at a remote router.
2890.Pp
2891.It Cm plr Ar packet-loss-rate
2892Packet loss rate.
2893Argument
2894.Ar packet-loss-rate
2895is a floating-point number between 0 and 1, with 0 meaning no
2896loss, 1 meaning 100% loss.
2897The loss rate is internally represented on 31 bits.
2898.Pp
2899.It Cm queue Brq Ar slots | size Ns Cm Kbytes
2900Queue size, in
2901.Ar slots
2902or
2903.Cm KBytes .
2904Default value is 50 slots, which
2905is the typical queue size for Ethernet devices.
2906Note that for slow speed links you should keep the queue
2907size short or your traffic might be affected by a significant
2908queueing delay.
2909E.g., 50 max-sized Ethernet packets (1500 bytes) mean 600Kbit
2910or 20s of queue on a 30Kbit/s pipe.
2911Even worse effects can result if you get packets from an
2912interface with a much larger MTU, e.g.\& the loopback interface
2913with its 16KB packets.
2914The
2915.Xr sysctl 8
2916variables
2917.Em net.inet.ip.dummynet.pipe_byte_limit
2918and
2919.Em net.inet.ip.dummynet.pipe_slot_limit
2920control the maximum lengths that can be specified.
2921.Pp
2922.It Cm red | gred Ar w_q Ns / Ns Ar min_th Ns / Ns Ar max_th Ns / Ns Ar max_p
2923[ecn]
2924Make use of the RED (Random Early Detection) queue management algorithm.
2925.Ar w_q
2926and
2927.Ar max_p
2928are floating
2929point numbers between 0 and 1 (inclusive), while
2930.Ar min_th
2931and
2932.Ar max_th
2933are integer numbers specifying thresholds for queue management
2934(thresholds are computed in bytes if the queue has been defined
2935in bytes, in slots otherwise).
2936The two parameters can also be of the same value if needed. The
2937.Nm dummynet
2938also supports the gentle RED variant (gred) and ECN (Explicit Congestion
2939Notification) as optional. Three
2940.Xr sysctl 8
2941variables can be used to control the RED behaviour:
2942.Bl -tag -width indent
2943.It Va net.inet.ip.dummynet.red_lookup_depth
2944specifies the accuracy in computing the average queue
2945when the link is idle (defaults to 256, must be greater than zero)
2946.It Va net.inet.ip.dummynet.red_avg_pkt_size
2947specifies the expected average packet size (defaults to 512, must be
2948greater than zero)
2949.It Va net.inet.ip.dummynet.red_max_pkt_size
2950specifies the expected maximum packet size, only used when queue
2951thresholds are in bytes (defaults to 1500, must be greater than zero).
2952.El
2953.Pp
2954.It Cm codel Oo Cm target Ar time Oc Oo Cm interval Ar time Oc Oo Cm ecn |
2955.Cm noecn Oc
2956Make use of the CoDel (Controlled-Delay) queue management algorithm.
2957.Ar time
2958is interpreted as milliseconds by default but seconds (s), milliseconds (ms) or
2959microseconds (us) can be specified instead.
2960CoDel drops or marks (ECN) packets
2961depending on packet sojourn time in the queue.
2962.Cm target
2963.Ar time
2964(5ms by default) is the minimum acceptable persistent queue delay that CoDel
2965allows.
2966CoDel does not drop packets directly after packets sojourn time becomes
2967higher than
2968.Cm target
2969.Ar time
2970but waits for
2971.Cm interval
2972.Ar time
2973(100ms default) before dropping.
2974.Cm interval
2975.Ar time
2976should be set to maximum RTT for all expected connections.
2977.Cm ecn
2978enables (disabled by default) packet marking (instead of dropping) for
2979ECN-enabled TCP flows when queue delay becomes high.
2980.Pp
2981Note that any token after
2982.Cm codel
2983is considered a parameter for CoDel.
2984So, ensure all pipe/queue
2985configuration options are written before
2986.Cm codel
2987token.
2988.Pp
2989The
2990.Xr sysctl 8
2991variables
2992.Va net.inet.ip.dummynet.codel.target
2993and
2994.Va net.inet.ip.dummynet.codel.interval
2995can be used to set CoDel default parameters.
2996.Pp
2997.It Cm pie Oo Cm target Ar time Oc Oo Cm tupdate Ar time Oc Oo
2998.Cm alpha Ar n Oc Oo Cm beta Ar n Oc Oo Cm max_burst Ar time Oc Oo
2999.Cm max_ecnth Ar n Oc Oo Cm ecn | Cm noecn Oc Oo Cm capdrop |
3000.Cm nocapdrop Oc Oo Cm drand | Cm nodrand Oc Oo Cm onoff
3001.Oc Oo Cm dre | Cm ts Oc
3002Make use of the PIE (Proportional Integral controller Enhanced) queue management
3003algorithm.
3004PIE drops or marks packets depending on a calculated drop probability during
3005en-queue process, with the aim of achieving high throughput while keeping queue
3006delay low.
3007At regular time intervals of
3008.Cm tupdate
3009.Ar time
3010(15ms by default) a background process (re)calculates the probability based on queue delay
3011deviations from
3012.Cm target
3013.Ar time
3014(15ms by default) and queue delay trends.
3015PIE approximates current queue
3016delay by using a departure rate estimation method, or (optionally) by using a
3017packet timestamp method similar to CoDel.
3018.Ar time
3019is interpreted as milliseconds by default but seconds (s), milliseconds (ms) or
3020microseconds (us) can be specified instead.
3021The other PIE parameters and options are as follows:
3022.Bl -tag -width indent
3023.It Cm alpha Ar n
3024.Ar n
3025is a floating point number between 0 and 7 which specifies the weight of queue
3026delay deviations that is used in drop probability calculation.
30270.125 is the default.
3028.It Cm beta Ar n
3029.Ar n
3030is a floating point number between 0 and 7 which specifies is the weight of queue
3031delay trend that is used in drop probability calculation.
30321.25 is the default.
3033.It Cm max_burst Ar time
3034The maximum period of time that PIE does not drop/mark packets.
3035150ms is the
3036default and 10s is the maximum value.
3037.It Cm max_ecnth Ar n
3038Even when ECN is enabled, PIE drops packets instead of marking them when drop
3039probability becomes higher than ECN probability threshold
3040.Cm max_ecnth Ar n
3041, the default is 0.1 (i.e 10%) and 1 is the maximum value.
3042.It Cm ecn | noecn
3043enable or disable ECN marking for ECN-enabled TCP flows.
3044Disabled by default.
3045.It Cm capdrop | nocapdrop
3046enable or disable cap drop adjustment.
3047Cap drop adjustment is enabled by default.
3048.It Cm drand | nodrand
3049enable or disable drop probability de-randomisation.
3050De-randomisation eliminates
3051the problem of dropping packets too close or too far.
3052De-randomisation is enabled by default.
3053.It Cm onoff
3054enable turning PIE on and off depending on queue load.
3055If this option is enabled,
3056PIE turns on when over 1/3 of queue becomes full.
3057This option is disabled by
3058default.
3059.It Cm dre | ts
3060Calculate queue delay using departure rate estimation
3061.Cm dre
3062or timestamps
3063.Cm ts .
3064.Cm dre
3065is used by default.
3066.El
3067.Pp
3068Note that any token after
3069.Cm pie
3070is considered a parameter for PIE.
3071So ensure all pipe/queue
3072the configuration options are written before
3073.Cm pie
3074token.
3075.Xr sysctl 8
3076variables can be used to control the
3077.Cm pie
3078default parameters.
3079See the
3080.Sx SYSCTL VARIABLES
3081section for more details.
3082.El
3083.Pp
3084When used with IPv6 data,
3085.Nm dummynet
3086currently has several limitations.
3087Information necessary to route link-local packets to an
3088interface is not available after processing by
3089.Nm dummynet
3090so those packets are dropped in the output path.
3091Care should be taken to ensure that link-local packets are not passed to
3092.Nm dummynet .
3093.Sh CHECKLIST
3094Here are some important points to consider when designing your
3095rules:
3096.Bl -bullet
3097.It
3098Remember that you filter both packets going
3099.Cm in
3100and
3101.Cm out .
3102Most connections need packets going in both directions.
3103.It
3104Remember to test very carefully.
3105It is a good idea to be near the console when doing this.
3106If you cannot be near the console,
3107use an auto-recovery script such as the one in
3108.Pa /usr/share/examples/ipfw/change_rules.sh .
3109.It
3110Do not forget the loopback interface.
3111.El
3112.Sh FINE POINTS
3113.Bl -bullet
3114.It
3115There are circumstances where fragmented datagrams are unconditionally
3116dropped.
3117TCP packets are dropped if they do not contain at least 20 bytes of
3118TCP header, UDP packets are dropped if they do not contain a full 8
3119byte UDP header, and ICMP packets are dropped if they do not contain
31204 bytes of ICMP header, enough to specify the ICMP type, code, and
3121checksum.
3122These packets are simply logged as
3123.Dq pullup failed
3124since there may not be enough good data in the packet to produce a
3125meaningful log entry.
3126.It
3127Another type of packet is unconditionally dropped, a TCP packet with a
3128fragment offset of one.
3129This is a valid packet, but it only has one use, to try
3130to circumvent firewalls.
3131When logging is enabled, these packets are
3132reported as being dropped by rule -1.
3133.It
3134If you are logged in over a network, loading the
3135.Xr kld 4
3136version of
3137.Nm
3138is probably not as straightforward as you would think.
3139The following command line is recommended:
3140.Bd -literal -offset indent
3141kldload ipfw && \e
3142ipfw add 32000 allow ip from any to any
3143.Ed
3144.Pp
3145Along the same lines, doing an
3146.Bd -literal -offset indent
3147ipfw flush
3148.Ed
3149.Pp
3150in similar surroundings is also a bad idea.
3151.It
3152The
3153.Nm
3154filter list may not be modified if the system security level
3155is set to 3 or higher
3156(see
3157.Xr init 8
3158for information on system security levels).
3159.El
3160.Sh PACKET DIVERSION
3161A
3162.Xr divert 4
3163socket bound to the specified port will receive all packets
3164diverted to that port.
3165If no socket is bound to the destination port, or if the divert module is
3166not loaded, or if the kernel was not compiled with divert socket support,
3167the packets are dropped.
3168.Sh NETWORK ADDRESS TRANSLATION (NAT)
3169.Nm
3170support in-kernel NAT using the kernel version of
3171.Xr libalias 3 .
3172The kernel module
3173.Cm ipfw_nat
3174should be loaded or kernel should have
3175.Cm options IPFIREWALL_NAT
3176to be able use NAT.
3177.Pp
3178The nat configuration command is the following:
3179.Bd -ragged -offset indent
3180.Bk -words
3181.Cm nat
3182.Ar nat_number
3183.Cm config
3184.Ar nat-configuration
3185.Ek
3186.Ed
3187.Pp
3188The following parameters can be configured:
3189.Bl -tag -width indent
3190.It Cm ip Ar ip_address
3191Define an ip address to use for aliasing.
3192.It Cm if Ar nic
3193Use ip address of NIC for aliasing, dynamically changing
3194it if NIC's ip address changes.
3195.It Cm log
3196Enable logging on this nat instance.
3197.It Cm deny_in
3198Deny any incoming connection from outside world.
3199.It Cm same_ports
3200Try to leave the alias port numbers unchanged from
3201the actual local port numbers.
3202.It Cm unreg_only
3203Traffic on the local network not originating from an
3204unregistered address spaces will be ignored.
3205.It Cm reset
3206Reset table of the packet aliasing engine on address change.
3207.It Cm reverse
3208Reverse the way libalias handles aliasing.
3209.It Cm proxy_only
3210Obey transparent proxy rules only, packet aliasing is not performed.
3211.It Cm skip_global
3212Skip instance in case of global state lookup (see below).
3213.El
3214.Pp
3215Some specials value can be supplied instead of
3216.Va nat_number:
3217.Bl -tag -width indent
3218.It Cm global
3219Looks up translation state in all configured nat instances.
3220If an entry is found, packet is aliased according to that entry.
3221If no entry was found in any of the instances, packet is passed unchanged,
3222and no new entry will be created.
3223See section
3224.Sx MULTIPLE INSTANCES
3225in
3226.Xr natd 8
3227for more information.
3228.It Cm tablearg
3229Uses argument supplied in lookup table.
3230See
3231.Sx LOOKUP TABLES
3232section below for more information on lookup tables.
3233.El
3234.Pp
3235To let the packet continue after being (de)aliased, set the sysctl variable
3236.Va net.inet.ip.fw.one_pass
3237to 0.
3238For more information about aliasing modes, refer to
3239.Xr libalias 3 .
3240See Section
3241.Sx EXAMPLES
3242for some examples about nat usage.
3243.Ss REDIRECT AND LSNAT SUPPORT IN IPFW
3244Redirect and LSNAT support follow closely the syntax used in
3245.Xr natd 8 .
3246See Section
3247.Sx EXAMPLES
3248for some examples on how to do redirect and lsnat.
3249.Ss SCTP NAT SUPPORT
3250SCTP nat can be configured in a similar manner to TCP through the
3251.Nm
3252command line tool.
3253The main difference is that
3254.Nm sctp nat
3255does not do port translation.
3256Since the local and global side ports will be the same,
3257there is no need to specify both.
3258Ports are redirected as follows:
3259.Bd -ragged -offset indent
3260.Bk -words
3261.Cm nat
3262.Ar nat_number
3263.Cm config if
3264.Ar nic
3265.Cm redirect_port sctp
3266.Ar ip_address [,addr_list] {[port | port-port] [,ports]}
3267.Ek
3268.Ed
3269.Pp
3270Most
3271.Nm sctp nat
3272configuration can be done in real-time through the
3273.Xr sysctl 8
3274interface.
3275All may be changed dynamically, though the hash_table size will only
3276change for new
3277.Nm nat
3278instances.
3279See
3280.Sx SYSCTL VARIABLES
3281for more info.
3282.Sh IPv6/IPv4 NETWORK ADDRESS AND PROTOCOL TRANSLATION
3283.Nm
3284supports in-kernel IPv6/IPv4 network address and protocol translation.
3285Stateful NAT64 translation allows IPv6-only clients to contact IPv4 servers
3286using unicast TCP, UDP or ICMP protocols.
3287One or more IPv4 addresses assigned to a stateful NAT64 translator are shared
3288among several IPv6-only clients.
3289When stateful NAT64 is used in conjunction with DNS64, no changes are usually
3290required in the IPv6 client or the IPv4 server.
3291The kernel module
3292.Cm ipfw_nat64
3293should be loaded or kernel should have
3294.Cm options IPFIREWALL_NAT64
3295to be able use stateful NAT64 translator.
3296.Pp
3297Stateful NAT64 uses a bunch of memory for several types of objects.
3298When IPv6 client initiates connection, NAT64 translator creates a host entry
3299in the states table.
3300Each host entry has a number of ports group entries allocated on demand.
3301Ports group entries contains connection state entries.
3302There are several options to control limits and lifetime for these objects.
3303.Pp
3304NAT64 translator follows RFC7915 when does ICMPv6/ICMP translation,
3305unsupported message types will be silently dropped.
3306IPv6 needs several ICMPv6 message types to be explicitly allowed for correct
3307operation.
3308Make sure that ND6 neighbor solicitation (ICMPv6 type 135) and neighbor
3309advertisement (ICMPv6 type 136) messages will not be handled by translation
3310rules.
3311.Pp
3312After translation NAT64 translator by default sends packets through
3313corresponding netisr queue.
3314Thus translator host should be configured as IPv4 and IPv6 router.
3315Also this means, that a packet is handled by firewall twice.
3316First time an original packet is handled and consumed by translator,
3317and then it is handled again as translated packet.
3318This behavior can be changed by sysctl variable
3319.Va net.inet.ip.fw.nat64_direct_output .
3320.Pp
3321The stateful NAT64 configuration command is the following:
3322.Bd -ragged -offset indent
3323.Bk -words
3324.Cm nat64lsn
3325.Ar name
3326.Cm create
3327.Ar create-options
3328.Ek
3329.Ed
3330.Pp
3331The following parameters can be configured:
3332.Bl -tag -width indent
3333.It Cm prefix4 Ar ipv4_prefix/plen
3334The IPv4 prefix with mask defines the pool of IPv4 addresses used as
3335source address after translation.
3336Stateful NAT64 module translates IPv6 source address of client to one
3337IPv4 address from this pool.
3338Note that incoming IPv4 packets that don't have corresponding state entry
3339in the states table will be dropped by translator.
3340Make sure that translation rules handle packets, destined to configured prefix.
3341.It Cm prefix6 Ar ipv6_prefix/length
3342The IPv6 prefix defines IPv4-embedded IPv6 addresses used by translator
3343to represent IPv4 addresses. This IPv6 prefix should be configured in DNS64.
3344The translator implementation follows RFC6052, that restricts the length of
3345prefixes to one of following: 32, 40, 48, 56, 64, or 96.
3346The Well-Known IPv6 Prefix 64:ff9b:: must be 96 bits long.
3347.It Cm max_ports Ar number
3348Maximum number of ports reserved for upper level protocols to one IPv6 client.
3349All reserved ports are divided into chunks between supported protocols.
3350The number of connections from one IPv6 client is limited by this option.
3351Note that closed TCP connections still remain in the list of connections until
3352.Cm tcp_close_age
3353interval will not expire.
3354Default value is
3355.Ar 2048 .
3356.It Cm host_del_age Ar seconds
3357The number of seconds until the host entry for a IPv6 client will be deleted
3358and all its resources will be released due to inactivity.
3359Default value is
3360.Ar 3600 .
3361.It Cm pg_del_age Ar seconds
3362The number of seconds until a ports group with unused state entries will
3363be released.
3364Default value is
3365.Ar 900 .
3366.It Cm tcp_syn_age Ar seconds
3367The number of seconds while a state entry for TCP connection with only SYN
3368sent will be kept.
3369If TCP connection establishing will not be finished,
3370state entry will be deleted.
3371Default value is
3372.Ar 10 .
3373.It Cm tcp_est_age Ar seconds
3374The number of seconds while a state entry for established TCP connection
3375will be kept.
3376Default value is
3377.Ar 7200 .
3378.It Cm tcp_close_age Ar seconds
3379The number of seconds while a state entry for closed TCP connection
3380will be kept.
3381Keeping state entries for closed connections is needed, because IPv4 servers
3382typically keep closed connections in a TIME_WAIT state for a several minutes.
3383Since translator's IPv4 addresses are shared among all IPv6 clients,
3384new connections from the same addresses and ports may be rejected by server,
3385because these connections are still in a TIME_WAIT state.
3386Keeping them in translator's state table protects from such rejects.
3387Default value is
3388.Ar 180 .
3389.It Cm udp_age Ar seconds
3390The number of seconds while translator keeps state entry in a waiting for
3391reply to the sent UDP datagram.
3392Default value is
3393.Ar 120 .
3394.It Cm icmp_age Ar seconds
3395The number of seconds while translator keeps state entry in a waiting for
3396reply to the sent ICMP message.
3397Default value is
3398.Ar 60 .
3399.It Cm log
3400Turn on logging of all handled packets via BPF through
3401.Ar ipfwlog0
3402interface.
3403.Ar ipfwlog0
3404is a pseudo interface and can be created after a boot manually with
3405.Cm ifconfig
3406command.
3407Note that it has different purpose than
3408.Ar ipfw0
3409interface.
3410Translators sends to BPF an additional information with each packet.
3411With
3412.Cm tcpdump
3413you are able to see each handled packet before and after translation.
3414.It Cm -log
3415Turn off logging of all handled packets via BPF.
3416.El
3417.Pp
3418To inspect a states table of stateful NAT64 the following command can be used:
3419.Bd -ragged -offset indent
3420.Bk -words
3421.Cm nat64lsn
3422.Ar name
3423.Cm show Cm states
3424.Ek
3425.Ed
3426.Pp
3427.Pp
3428Stateless NAT64 translator doesn't use a states table for translation
3429and converts IPv4 addresses to IPv6 and vice versa solely based on the
3430mappings taken from configured lookup tables.
3431Since a states table doesn't used by stateless translator,
3432it can be configured to pass IPv4 clients to IPv6-only servers.
3433.Pp
3434The stateless NAT64 configuration command is the following:
3435.Bd -ragged -offset indent
3436.Bk -words
3437.Cm nat64stl
3438.Ar name
3439.Cm create
3440.Ar create-options
3441.Ek
3442.Ed
3443.Pp
3444The following parameters can be configured:
3445.Bl -tag -width indent
3446.It Cm prefix6 Ar ipv6_prefix/length
3447The IPv6 prefix defines IPv4-embedded IPv6 addresses used by translator
3448to represent IPv4 addresses. This IPv6 prefix should be configured in DNS64.
3449.It Cm table4 Ar table46
3450The lookup table
3451.Ar table46
3452contains mapping how IPv4 addresses should be translated to IPv6 addresses.
3453.It Cm table6 Ar table64
3454The lookup table
3455.Ar table64
3456contains mapping how IPv6 addresses should be translated to IPv4 addresses.
3457.It Cm log
3458Turn on logging of all handled packets via BPF through
3459.Ar ipfwlog0
3460interface.
3461.It Cm -log
3462Turn off logging of all handled packets via BPF.
3463.El
3464.Pp
3465Note that the behavior of stateless translator with respect to not matched
3466packets differs from stateful translator.
3467If corresponding addresses was not found in the lookup tables, the packet
3468will not be dropped and the search continues.
3469.Sh IPv6-to-IPv6 NETWORK PREFIX TRANSLATION (NPTv6)
3470.Nm
3471supports in-kernel IPv6-to-IPv6 network prefix translation as described
3472in RFC6296.
3473The kernel module
3474.Cm ipfw_nptv6
3475should be loaded or kernel should has
3476.Cm options IPFIREWALL_NPTV6
3477to be able use NPTv6 translator.
3478.Pp
3479The NPTv6 configuration command is the following:
3480.Bd -ragged -offset indent
3481.Bk -words
3482.Cm nptv6
3483.Ar name
3484.Cm create
3485.Ar create-options
3486.Ek
3487.Ed
3488.Pp
3489The following parameters can be configured:
3490.Bl -tag -width indent
3491.It Cm int_prefix Ar ipv6_prefix
3492IPv6 prefix used in internal network.
3493NPTv6 module translates source address when it matches this prefix.
3494.It Cm ext_prefix Ar ipv6_prefix
3495IPv6 prefix used in external network.
3496NPTv6 module translates destination address when it matches this prefix.
3497.It Cm ext_if Ar nic
3498The NPTv6 module will use first global IPv6 address from interface
3499.Ar nic
3500as external prefix.
3501It can be useful when IPv6 prefix of external network is dynamically obtained.
3502.Cm ext_prefix
3503and
3504.Cm ext_if
3505options are mutually exclusive.
3506.It Cm prefixlen Ar length
3507The length of specified IPv6 prefixes. It must be in range from 8 to 64.
3508.El
3509.Pp
3510Note that the prefix translation rules are silently ignored when IPv6 packet
3511forwarding is disabled.
3512To enable the packet forwarding, set the sysctl variable
3513.Va net.inet6.ip6.forwarding
3514to 1.
3515.Pp
3516To let the packet continue after being translated, set the sysctl variable
3517.Va net.inet.ip.fw.one_pass
3518to 0.
3519.Sh LOADER TUNABLES
3520Tunables can be set in
3521.Xr loader 8
3522prompt,
3523.Xr loader.conf 5
3524or
3525.Xr kenv 1
3526before ipfw module gets loaded.
3527.Bl -tag -width indent
3528.It Va net.inet.ip.fw.default_to_accept: No 0
3529Defines ipfw last rule behavior.
3530This value overrides
3531.Cd "options IPFW_DEFAULT_TO_(ACCEPT|DENY)"
3532from kernel configuration file.
3533.It Va net.inet.ip.fw.tables_max: No 128
3534Defines number of tables available in ipfw.
3535Number cannot exceed 65534.
3536.El
3537.Sh SYSCTL VARIABLES
3538A set of
3539.Xr sysctl 8
3540variables controls the behaviour of the firewall and
3541associated modules
3542.Pq Nm dummynet , bridge , sctp nat .
3543These are shown below together with their default value
3544(but always check with the
3545.Xr sysctl 8
3546command what value is actually in use) and meaning:
3547.Bl -tag -width indent
3548.It Va net.inet.ip.alias.sctp.accept_global_ootb_addip: No 0
3549Defines how the
3550.Nm nat
3551responds to receipt of global OOTB ASCONF-AddIP:
3552.Bl -tag -width indent
3553.It Cm 0
3554No response (unless a partially matching association exists -
3555ports and vtags match but global address does not)
3556.It Cm 1
3557.Nm nat
3558will accept and process all OOTB global AddIP messages.
3559.El
3560.Pp
3561Option 1 should never be selected as this forms a security risk.
3562An attacker can
3563establish multiple fake associations by sending AddIP messages.
3564.It Va net.inet.ip.alias.sctp.chunk_proc_limit: No 5
3565Defines the maximum number of chunks in an SCTP packet that will be
3566parsed for a
3567packet that matches an existing association.
3568This value is enforced to be greater or equal than
3569.Cm net.inet.ip.alias.sctp.initialising_chunk_proc_limit .
3570A high value is
3571a DoS risk yet setting too low a value may result in
3572important control chunks in
3573the packet not being located and parsed.
3574.It Va net.inet.ip.alias.sctp.error_on_ootb: No 1
3575Defines when the
3576.Nm nat
3577responds to any Out-of-the-Blue (OOTB) packets with ErrorM packets.
3578An OOTB packet is a packet that arrives with no existing association
3579registered in the
3580.Nm nat
3581and is not an INIT or ASCONF-AddIP packet:
3582.Bl -tag -width indent
3583.It Cm 0
3584ErrorM is never sent in response to OOTB packets.
3585.It Cm 1
3586ErrorM is only sent to OOTB packets received on the local side.
3587.It Cm 2
3588ErrorM is sent to the local side and on the global side ONLY if there is a
3589partial match (ports and vtags match but the source global IP does not).
3590This value is only useful if the
3591.Nm nat
3592is tracking global IP addresses.
3593.It Cm 3
3594ErrorM is sent in response to all OOTB packets on both
3595the local and global side
3596(DoS risk).
3597.El
3598.Pp
3599At the moment the default is 0, since the ErrorM packet is not yet
3600supported by most SCTP stacks.
3601When it is supported, and if not tracking
3602global addresses, we recommend setting this value to 1 to allow
3603multi-homed local hosts to function with the
3604.Nm nat .
3605To track global addresses, we recommend setting this value to 2 to
3606allow global hosts to be informed when they need to (re)send an
3607ASCONF-AddIP.
3608Value 3 should never be chosen (except for debugging) as the
3609.Nm nat
3610will respond to all OOTB global packets (a DoS risk).
3611.It Va net.inet.ip.alias.sctp.hashtable_size: No 2003
3612Size of hash tables used for
3613.Nm nat
3614lookups (100 < prime_number > 1000001).
3615This value sets the
3616.Nm hash table
3617size for any future created
3618.Nm nat
3619instance and therefore must be set prior to creating a
3620.Nm nat
3621instance.
3622The table sizes may be changed to suit specific needs.
3623If there will be few
3624concurrent associations, and memory is scarce, you may make these smaller.
3625If there will be many thousands (or millions) of concurrent associations, you
3626should make these larger.
3627A prime number is best for the table size.
3628The sysctl
3629update function will adjust your input value to the next highest prime number.
3630.It Va net.inet.ip.alias.sctp.holddown_time:  No 0
3631Hold association in table for this many seconds after receiving a
3632SHUTDOWN-COMPLETE.
3633This allows endpoints to correct shutdown gracefully if a
3634shutdown_complete is lost and retransmissions are required.
3635.It Va net.inet.ip.alias.sctp.init_timer: No 15
3636Timeout value while waiting for (INIT-ACK|AddIP-ACK).
3637This value cannot be 0.
3638.It Va net.inet.ip.alias.sctp.initialising_chunk_proc_limit: No 2
3639Defines the maximum number of chunks in an SCTP packet that will be parsed when
3640no existing association exists that matches that packet.
3641Ideally this packet
3642will only be an INIT or ASCONF-AddIP packet.
3643A higher value may become a DoS
3644risk as malformed packets can consume processing resources.
3645.It Va net.inet.ip.alias.sctp.param_proc_limit: No 25
3646Defines the maximum number of parameters within a chunk that will be
3647parsed in a
3648packet.
3649As for other similar sysctl variables, larger values pose a DoS risk.
3650.It Va net.inet.ip.alias.sctp.log_level: No 0
3651Level of detail in the system log messages (0 \- minimal, 1 \- event,
36522 \- info, 3 \- detail, 4 \- debug, 5 \- max debug).
3653May be a good
3654option in high loss environments.
3655.It Va net.inet.ip.alias.sctp.shutdown_time: No 15
3656Timeout value while waiting for SHUTDOWN-COMPLETE.
3657This value cannot be 0.
3658.It Va net.inet.ip.alias.sctp.track_global_addresses: No 0
3659Enables/disables global IP address tracking within the
3660.Nm nat
3661and places an
3662upper limit on the number of addresses tracked for each association:
3663.Bl -tag -width indent
3664.It Cm 0
3665Global tracking is disabled
3666.It Cm >1
3667Enables tracking, the maximum number of addresses tracked for each
3668association is limited to this value
3669.El
3670.Pp
3671This variable is fully dynamic, the new value will be adopted for all newly
3672arriving associations, existing associations are treated
3673as they were previously.
3674Global tracking will decrease the number of collisions within the
3675.Nm nat
3676at a cost
3677of increased processing load, memory usage, complexity, and possible
3678.Nm nat
3679state
3680problems in complex networks with multiple
3681.Nm nats .
3682We recommend not tracking
3683global IP addresses, this will still result in a fully functional
3684.Nm nat .
3685.It Va net.inet.ip.alias.sctp.up_timer: No 300
3686Timeout value to keep an association up with no traffic.
3687This value cannot be 0.
3688.It Va net.inet.ip.dummynet.codel.interval : No 100000
3689Default
3690.Cm codel
3691AQM interval in microseconds.
3692The value must be in the range 1..5000000.
3693.It Va net.inet.ip.dummynet.codel.target : No 5000
3694Default
3695.Cm codel
3696AQM target delay time in microseconds (the minimum acceptable persistent queue
3697delay).
3698The value must be in the range 1..5000000.
3699.It Va net.inet.ip.dummynet.expire : No 1
3700Lazily delete dynamic pipes/queue once they have no pending traffic.
3701You can disable this by setting the variable to 0, in which case
3702the pipes/queues will only be deleted when the threshold is reached.
3703.It Va net.inet.ip.dummynet.fqcodel.flows : No 1024
3704Defines the default total number of flow queues (sub-queues) that
3705.Cm fq_codel
3706creates and manages.
3707The value must be in the range 1..65536.
3708.It Va net.inet.ip.dummynet.fqcodel.interval : No 100000
3709Default
3710.Cm fq_codel
3711scheduler/AQM interval in microseconds.
3712The value must be in the range 1..5000000.
3713.It Va net.inet.ip.dummynet.fqcodel.limit : No 10240
3714The default hard size limit (in unit of packet) of all queues managed by an
3715instance of the
3716.Cm fq_codel
3717scheduler.
3718The value must be in the range 1..20480.
3719.It Va net.inet.ip.dummynet.fqcodel.quantum : No 1514
3720The default quantum (credit) of the
3721.Cm fq_codel
3722in unit of byte.
3723The value must be in the range 1..9000.
3724.It Va net.inet.ip.dummynet.fqcodel.target : No 5000
3725Default
3726.Cm fq_codel
3727scheduler/AQM target delay time in microseconds (the minimum acceptable
3728persistent queue delay).
3729The value must be in the range 1..5000000.
3730.It Va net.inet.ip.dummynet.fqpie.alpha : No 125
3731The default
3732.Ar alpha
3733parameter (scaled by 1000) for
3734.Cm fq_pie
3735scheduler/AQM.
3736The value must be in the range 1..7000.
3737.It Va net.inet.ip.dummynet.fqpie.beta : No 1250
3738The default
3739.Ar beta
3740parameter (scaled by 1000) for
3741.Cm fq_pie
3742scheduler/AQM.
3743The value must be in the range 1..7000.
3744.It Va net.inet.ip.dummynet.fqpie.flows : No 1024
3745Defines the default total number of flow queues (sub-queues) that
3746.Cm fq_pie
3747creates and manages.
3748The value must be in the range 1..65536.
3749.It Va net.inet.ip.dummynet.fqpie.limit : No 10240
3750The default hard size limit (in unit of packet) of all queues managed by an
3751instance of the
3752.Cm fq_pie
3753scheduler.
3754The value must be in the range 1..20480.
3755.It Va net.inet.ip.dummynet.fqpie.max_burst : No 150000
3756The default maximum period of microseconds that
3757.Cm fq_pie
3758scheduler/AQM does not drop/mark packets.
3759The value must be in the range 1..10000000.
3760.It Va net.inet.ip.dummynet.fqpie.max_ecnth : No 99
3761The default maximum ECN probability threshold (scaled by 1000) for
3762.Cm fq_pie
3763scheduler/AQM.
3764The value must be in the range 1..7000.
3765.It Va net.inet.ip.dummynet.fqpie.quantum : No 1514
3766The default quantum (credit) of the
3767.Cm fq_pie
3768in unit of byte.
3769The value must be in the range 1..9000.
3770.It Va net.inet.ip.dummynet.fqpie.target : No 15000
3771The default
3772.Cm target
3773delay of the
3774.Cm fq_pie
3775in unit of microsecond.
3776The value must be in the range 1..5000000.
3777.It Va net.inet.ip.dummynet.fqpie.tupdate : No 15000
3778The default
3779.Cm tupdate
3780of the
3781.Cm fq_pie
3782in unit of microsecond.
3783The value must be in the range 1..5000000.
3784.It Va net.inet.ip.dummynet.hash_size : No 64
3785Default size of the hash table used for dynamic pipes/queues.
3786This value is used when no
3787.Cm buckets
3788option is specified when configuring a pipe/queue.
3789.It Va net.inet.ip.dummynet.io_fast : No 0
3790If set to a non-zero value,
3791the
3792.Dq fast
3793mode of
3794.Nm dummynet
3795operation (see above) is enabled.
3796.It Va net.inet.ip.dummynet.io_pkt
3797Number of packets passed to
3798.Nm dummynet .
3799.It Va net.inet.ip.dummynet.io_pkt_drop
3800Number of packets dropped by
3801.Nm dummynet .
3802.It Va net.inet.ip.dummynet.io_pkt_fast
3803Number of packets bypassed by the
3804.Nm dummynet
3805scheduler.
3806.It Va net.inet.ip.dummynet.max_chain_len : No 16
3807Target value for the maximum number of pipes/queues in a hash bucket.
3808The product
3809.Cm max_chain_len*hash_size
3810is used to determine the threshold over which empty pipes/queues
3811will be expired even when
3812.Cm net.inet.ip.dummynet.expire=0 .
3813.It Va net.inet.ip.dummynet.red_lookup_depth : No 256
3814.It Va net.inet.ip.dummynet.red_avg_pkt_size : No 512
3815.It Va net.inet.ip.dummynet.red_max_pkt_size : No 1500
3816Parameters used in the computations of the drop probability
3817for the RED algorithm.
3818.It Va net.inet.ip.dummynet.pie.alpha : No 125
3819The default
3820.Ar alpha
3821parameter (scaled by 1000) for
3822.Cm pie
3823AQM.
3824The value must be in the range 1..7000.
3825.It Va net.inet.ip.dummynet.pie.beta : No 1250
3826The default
3827.Ar beta
3828parameter (scaled by 1000) for
3829.Cm pie
3830AQM.
3831The value must be in the range 1..7000.
3832.It Va net.inet.ip.dummynet.pie.max_burst : No 150000
3833The default maximum period of microseconds that
3834.Cm pie
3835AQM does not drop/mark packets.
3836The value must be in the range 1..10000000.
3837.It Va net.inet.ip.dummynet.pie.max_ecnth : No 99
3838The default maximum ECN probability threshold (scaled by 1000) for
3839.Cm pie
3840AQM.
3841The value must be in the range 1..7000.
3842.It Va net.inet.ip.dummynet.pie.target : No 15000
3843The default
3844.Cm target
3845delay of
3846.Cm pie
3847AQM in unit of microsecond.
3848The value must be in the range 1..5000000.
3849.It Va net.inet.ip.dummynet.pie.tupdate : No 15000
3850The default
3851.Cm tupdate
3852of
3853.Cm pie
3854AQM in unit of microsecond.
3855The value must be in the range 1..5000000.
3856.It Va net.inet.ip.dummynet.pipe_byte_limit : No 1048576
3857.It Va net.inet.ip.dummynet.pipe_slot_limit : No 100
3858The maximum queue size that can be specified in bytes or packets.
3859These limits prevent accidental exhaustion of resources such as mbufs.
3860If you raise these limits,
3861you should make sure the system is configured so that sufficient resources
3862are available.
3863.It Va net.inet.ip.fw.autoinc_step : No 100
3864Delta between rule numbers when auto-generating them.
3865The value must be in the range 1..1000.
3866.It Va net.inet.ip.fw.curr_dyn_buckets : Va net.inet.ip.fw.dyn_buckets
3867The current number of buckets in the hash table for dynamic rules
3868(readonly).
3869.It Va net.inet.ip.fw.debug : No 1
3870Controls debugging messages produced by
3871.Nm .
3872.It Va net.inet.ip.fw.default_rule : No 65535
3873The default rule number (read-only).
3874By the design of
3875.Nm , the default rule is the last one, so its number
3876can also serve as the highest number allowed for a rule.
3877.It Va net.inet.ip.fw.dyn_buckets : No 256
3878The number of buckets in the hash table for dynamic rules.
3879Must be a power of 2, up to 65536.
3880It only takes effect when all dynamic rules have expired, so you
3881are advised to use a
3882.Cm flush
3883command to make sure that the hash table is resized.
3884.It Va net.inet.ip.fw.dyn_count : No 3
3885Current number of dynamic rules
3886(read-only).
3887.It Va net.inet.ip.fw.dyn_keepalive : No 1
3888Enables generation of keepalive packets for
3889.Cm keep-state
3890rules on TCP sessions.
3891A keepalive is generated to both
3892sides of the connection every 5 seconds for the last 20
3893seconds of the lifetime of the rule.
3894.It Va net.inet.ip.fw.dyn_max : No 8192
3895Maximum number of dynamic rules.
3896When you hit this limit, no more dynamic rules can be
3897installed until old ones expire.
3898.It Va net.inet.ip.fw.dyn_ack_lifetime : No 300
3899.It Va net.inet.ip.fw.dyn_syn_lifetime : No 20
3900.It Va net.inet.ip.fw.dyn_fin_lifetime : No 1
3901.It Va net.inet.ip.fw.dyn_rst_lifetime : No 1
3902.It Va net.inet.ip.fw.dyn_udp_lifetime : No 5
3903.It Va net.inet.ip.fw.dyn_short_lifetime : No 30
3904These variables control the lifetime, in seconds, of dynamic
3905rules.
3906Upon the initial SYN exchange the lifetime is kept short,
3907then increased after both SYN have been seen, then decreased
3908again during the final FIN exchange or when a RST is received.
3909Both
3910.Em dyn_fin_lifetime
3911and
3912.Em dyn_rst_lifetime
3913must be strictly lower than 5 seconds, the period of
3914repetition of keepalives.
3915The firewall enforces that.
3916.It Va net.inet.ip.fw.dyn_keep_states: No 0
3917Keep dynamic states on rule/set deletion.
3918States are relinked to default rule (65535).
3919This can be handly for ruleset reload.
3920Turned off by default.
3921.It Va net.inet.ip.fw.enable : No 1
3922Enables the firewall.
3923Setting this variable to 0 lets you run your machine without
3924firewall even if compiled in.
3925.It Va net.inet6.ip6.fw.enable : No 1
3926provides the same functionality as above for the IPv6 case.
3927.It Va net.inet.ip.fw.one_pass : No 1
3928When set, the packet exiting from the
3929.Nm dummynet
3930pipe or from
3931.Xr ng_ipfw 4
3932node is not passed though the firewall again.
3933Otherwise, after an action, the packet is
3934reinjected into the firewall at the next rule.
3935.It Va net.inet.ip.fw.tables_max : No 128
3936Maximum number of tables.
3937.It Va net.inet.ip.fw.verbose : No 1
3938Enables verbose messages.
3939.It Va net.inet.ip.fw.verbose_limit : No 0
3940Limits the number of messages produced by a verbose firewall.
3941.It Va net.inet6.ip6.fw.deny_unknown_exthdrs : No 1
3942If enabled packets with unknown IPv6 Extension Headers will be denied.
3943.It Va net.link.ether.ipfw : No 0
3944Controls whether layer-2 packets are passed to
3945.Nm .
3946Default is no.
3947.It Va net.link.bridge.ipfw : No 0
3948Controls whether bridged packets are passed to
3949.Nm .
3950Default is no.
3951.It Va net.inet.ip.fw.nat64_allow_private : No 0
3952Defines how
3953.Nm nat64
3954handles private IPv4 addresses:
3955.Bl -tag -width indent
3956.It Cm 0
3957Packets with private IPv4 will not be handled by translator
3958.It Cm 1
3959Translator will accept and process packets with private IPv4 addresses.
3960.El
3961.It Va net.inet.ip.fw.nat64_debug : No 0
3962Controls debugging messages produced by
3963.Nm ipfw_nat64
3964module.
3965.It Va net.inet.ip.fw.nat64_direct_output : No 0
3966Controls the output method used by
3967.Nm ipfw_nat64
3968module:
3969.Bl -tag -width indent
3970.It Cm 0
3971A packet is handled by
3972.Nm ipfw
3973twice.
3974First time an original packet is handled by
3975.Nm ipfw
3976and consumed by
3977.Nm ipfw_nat64
3978translator.
3979Then translated packet is queued via netisr to input processing again.
3980.It Cm 1
3981A packet is handled by
3982.Nm ipfw
3983only once, and after translation it will be pushed directly to outgoing
3984interface.
3985.El
3986.El
3987.Sh INTERNAL DIAGNOSTICS
3988There are some commands that may be useful to understand current state
3989of certain subsystems inside kernel module.
3990These commands provide debugging output which may change without notice.
3991.Pp
3992Currently the following commands are available as
3993.Cm internal
3994sub-options:
3995.Bl -tag -width indent
3996.It Cm iflist
3997Lists all interface which are currently tracked by
3998.Nm
3999with their in-kernel status.
4000.It Cm talist
4001List all table lookup algorithms currently available.
4002.El
4003.Sh EXAMPLES
4004There are far too many possible uses of
4005.Nm
4006so this Section will only give a small set of examples.
4007.Pp
4008.Ss BASIC PACKET FILTERING
4009This command adds an entry which denies all tcp packets from
4010.Em cracker.evil.org
4011to the telnet port of
4012.Em wolf.tambov.su
4013from being forwarded by the host:
4014.Pp
4015.Dl "ipfw add deny tcp from cracker.evil.org to wolf.tambov.su telnet"
4016.Pp
4017This one disallows any connection from the entire cracker's
4018network to my host:
4019.Pp
4020.Dl "ipfw add deny ip from 123.45.67.0/24 to my.host.org"
4021.Pp
4022A first and efficient way to limit access (not using dynamic rules)
4023is the use of the following rules:
4024.Pp
4025.Dl "ipfw add allow tcp from any to any established"
4026.Dl "ipfw add allow tcp from net1 portlist1 to net2 portlist2 setup"
4027.Dl "ipfw add allow tcp from net3 portlist3 to net3 portlist3 setup"
4028.Dl "..."
4029.Dl "ipfw add deny tcp from any to any"
4030.Pp
4031The first rule will be a quick match for normal TCP packets,
4032but it will not match the initial SYN packet, which will be
4033matched by the
4034.Cm setup
4035rules only for selected source/destination pairs.
4036All other SYN packets will be rejected by the final
4037.Cm deny
4038rule.
4039.Pp
4040If you administer one or more subnets, you can take advantage
4041of the address sets and or-blocks and write extremely
4042compact rulesets which selectively enable services to blocks
4043of clients, as below:
4044.Pp
4045.Dl "goodguys=\*q{ 10.1.2.0/24{20,35,66,18} or 10.2.3.0/28{6,3,11} }\*q"
4046.Dl "badguys=\*q10.1.2.0/24{8,38,60}\*q"
4047.Dl ""
4048.Dl "ipfw add allow ip from ${goodguys} to any"
4049.Dl "ipfw add deny ip from ${badguys} to any"
4050.Dl "... normal policies ..."
4051.Pp
4052The
4053.Cm verrevpath
4054option could be used to do automated anti-spoofing by adding the
4055following to the top of a ruleset:
4056.Pp
4057.Dl "ipfw add deny ip from any to any not verrevpath in"
4058.Pp
4059This rule drops all incoming packets that appear to be coming to the
4060system on the wrong interface.
4061For example, a packet with a source
4062address belonging to a host on a protected internal network would be
4063dropped if it tried to enter the system from an external interface.
4064.Pp
4065The
4066.Cm antispoof
4067option could be used to do similar but more restricted anti-spoofing
4068by adding the following to the top of a ruleset:
4069.Pp
4070.Dl "ipfw add deny ip from any to any not antispoof in"
4071.Pp
4072This rule drops all incoming packets that appear to be coming from another
4073directly connected system but on the wrong interface.
4074For example, a packet with a source address of
4075.Li 192.168.0.0/24 ,
4076configured on
4077.Li fxp0 ,
4078but coming in on
4079.Li fxp1
4080would be dropped.
4081.Pp
4082The
4083.Cm setdscp
4084option could be used to (re)mark user traffic,
4085by adding the following to the appropriate place in ruleset:
4086.Pp
4087.Dl "ipfw add setdscp be ip from any to any dscp af11,af21"
4088.Ss SELECTIVE MIRRORING
4089If your network has network traffic analyzer
4090connected to your host directly via dedicated interface
4091or remotely via RSPAN vlan, you can selectively mirror
4092some Ethernet layer2 frames to the analyzer.
4093.Pp
4094First, make sure your firewall is already configured and runs.
4095Then, enable layer2 processing if not already enabled:
4096.Pp
4097.Dl "sysctl net.link.ether.ipfw=1"
4098.Pp
4099Next, load needed additional kernel modules:
4100.Pp
4101.Dl "kldload ng_ether ng_ipfw"
4102.Pp
4103Optionally, make system load these modules automatically
4104at startup:
4105.Pp
4106.Dl sysrc kld_list+="ng_ether ng_ipfw"
4107.Pp
4108Next, configure
4109.Xr ng_ipfw 4
4110kernel module to transmit mirrored copies of layer2 frames
4111out via vlan900 interface:
4112.Pp
4113.Dl "ngctl connect ipfw: vlan900: 1 lower"
4114.Pp
4115Think of "1" here as of "mirroring instance index" and vlan900 is its
4116destination.
4117You can have arbitrary number of instances.
4118Refer to
4119.Xr ng_ipfw 4
4120for details.
4121.Pp
4122At last, actually start mirroring of selected frames using "instance 1".
4123For frames incoming from em0 interface:
4124.Pp
4125.Dl "ipfw add ngtee 1 ip from any to 192.168.0.1 layer2 in recv em0"
4126.Pp
4127For frames outgoing to em0 interface:
4128.Pp
4129.Dl "ipfw add ngtee 1 ip from any to 192.168.0.1 layer2 out xmit em0"
4130.Pp
4131For both incoming and outgoing frames while flowing through em0:
4132.Pp
4133.Dl "ipfw add ngtee 1 ip from any to 192.168.0.1 layer2 via em0"
4134.Pp
4135Make sure you do not perform mirroring for already duplicated frames
4136or kernel may hang as there is no safety net.
4137.Ss DYNAMIC RULES
4138In order to protect a site from flood attacks involving fake
4139TCP packets, it is safer to use dynamic rules:
4140.Pp
4141.Dl "ipfw add check-state"
4142.Dl "ipfw add deny tcp from any to any established"
4143.Dl "ipfw add allow tcp from my-net to any setup keep-state"
4144.Pp
4145This will let the firewall install dynamic rules only for
4146those connection which start with a regular SYN packet coming
4147from the inside of our network.
4148Dynamic rules are checked when encountering the first
4149occurrence of a
4150.Cm check-state ,
4151.Cm keep-state
4152or
4153.Cm limit
4154rule.
4155A
4156.Cm check-state
4157rule should usually be placed near the beginning of the
4158ruleset to minimize the amount of work scanning the ruleset.
4159Your mileage may vary.
4160.Pp
4161For more complex scenarios with dynamic rules
4162.Cm record-state
4163and
4164.Cm defer-action
4165can be used to precisely control creation and checking of dynamic rules.
4166Example of usage of these options are provided in
4167.Sx NETWORK ADDRESS TRANSLATION (NAT)
4168Section.
4169.Pp
4170To limit the number of connections a user can open
4171you can use the following type of rules:
4172.Pp
4173.Dl "ipfw add allow tcp from my-net/24 to any setup limit src-addr 10"
4174.Dl "ipfw add allow tcp from any to me setup limit src-addr 4"
4175.Pp
4176The former (assuming it runs on a gateway) will allow each host
4177on a /24 network to open at most 10 TCP connections.
4178The latter can be placed on a server to make sure that a single
4179client does not use more than 4 simultaneous connections.
4180.Pp
4181.Em BEWARE :
4182stateful rules can be subject to denial-of-service attacks
4183by a SYN-flood which opens a huge number of dynamic rules.
4184The effects of such attacks can be partially limited by
4185acting on a set of
4186.Xr sysctl 8
4187variables which control the operation of the firewall.
4188.Pp
4189Here is a good usage of the
4190.Cm list
4191command to see accounting records and timestamp information:
4192.Pp
4193.Dl ipfw -at list
4194.Pp
4195or in short form without timestamps:
4196.Pp
4197.Dl ipfw -a list
4198.Pp
4199which is equivalent to:
4200.Pp
4201.Dl ipfw show
4202.Pp
4203Next rule diverts all incoming packets from 192.168.2.0/24
4204to divert port 5000:
4205.Pp
4206.Dl ipfw divert 5000 ip from 192.168.2.0/24 to any in
4207.Ss TRAFFIC SHAPING
4208The following rules show some of the applications of
4209.Nm
4210and
4211.Nm dummynet
4212for simulations and the like.
4213.Pp
4214This rule drops random incoming packets with a probability
4215of 5%:
4216.Pp
4217.Dl "ipfw add prob 0.05 deny ip from any to any in"
4218.Pp
4219A similar effect can be achieved making use of
4220.Nm dummynet
4221pipes:
4222.Pp
4223.Dl "ipfw add pipe 10 ip from any to any"
4224.Dl "ipfw pipe 10 config plr 0.05"
4225.Pp
4226We can use pipes to artificially limit bandwidth, e.g.\& on a
4227machine acting as a router, if we want to limit traffic from
4228local clients on 192.168.2.0/24 we do:
4229.Pp
4230.Dl "ipfw add pipe 1 ip from 192.168.2.0/24 to any out"
4231.Dl "ipfw pipe 1 config bw 300Kbit/s queue 50KBytes"
4232.Pp
4233note that we use the
4234.Cm out
4235modifier so that the rule is not used twice.
4236Remember in fact that
4237.Nm
4238rules are checked both on incoming and outgoing packets.
4239.Pp
4240Should we want to simulate a bidirectional link with bandwidth
4241limitations, the correct way is the following:
4242.Pp
4243.Dl "ipfw add pipe 1 ip from any to any out"
4244.Dl "ipfw add pipe 2 ip from any to any in"
4245.Dl "ipfw pipe 1 config bw 64Kbit/s queue 10Kbytes"
4246.Dl "ipfw pipe 2 config bw 64Kbit/s queue 10Kbytes"
4247.Pp
4248The above can be very useful, e.g.\& if you want to see how
4249your fancy Web page will look for a residential user who
4250is connected only through a slow link.
4251You should not use only one pipe for both directions, unless
4252you want to simulate a half-duplex medium (e.g.\& AppleTalk,
4253Ethernet, IRDA).
4254It is not necessary that both pipes have the same configuration,
4255so we can also simulate asymmetric links.
4256.Pp
4257Should we want to verify network performance with the RED queue
4258management algorithm:
4259.Pp
4260.Dl "ipfw add pipe 1 ip from any to any"
4261.Dl "ipfw pipe 1 config bw 500Kbit/s queue 100 red 0.002/30/80/0.1"
4262.Pp
4263Another typical application of the traffic shaper is to
4264introduce some delay in the communication.
4265This can significantly affect applications which do a lot of Remote
4266Procedure Calls, and where the round-trip-time of the
4267connection often becomes a limiting factor much more than
4268bandwidth:
4269.Pp
4270.Dl "ipfw add pipe 1 ip from any to any out"
4271.Dl "ipfw add pipe 2 ip from any to any in"
4272.Dl "ipfw pipe 1 config delay 250ms bw 1Mbit/s"
4273.Dl "ipfw pipe 2 config delay 250ms bw 1Mbit/s"
4274.Pp
4275Per-flow queueing can be useful for a variety of purposes.
4276A very simple one is counting traffic:
4277.Pp
4278.Dl "ipfw add pipe 1 tcp from any to any"
4279.Dl "ipfw add pipe 1 udp from any to any"
4280.Dl "ipfw add pipe 1 ip from any to any"
4281.Dl "ipfw pipe 1 config mask all"
4282.Pp
4283The above set of rules will create queues (and collect
4284statistics) for all traffic.
4285Because the pipes have no limitations, the only effect is
4286collecting statistics.
4287Note that we need 3 rules, not just the last one, because
4288when
4289.Nm
4290tries to match IP packets it will not consider ports, so we
4291would not see connections on separate ports as different
4292ones.
4293.Pp
4294A more sophisticated example is limiting the outbound traffic
4295on a net with per-host limits, rather than per-network limits:
4296.Pp
4297.Dl "ipfw add pipe 1 ip from 192.168.2.0/24 to any out"
4298.Dl "ipfw add pipe 2 ip from any to 192.168.2.0/24 in"
4299.Dl "ipfw pipe 1 config mask src-ip 0x000000ff bw 200Kbit/s queue 20Kbytes"
4300.Dl "ipfw pipe 2 config mask dst-ip 0x000000ff bw 200Kbit/s queue 20Kbytes"
4301.Ss LOOKUP TABLES
4302In the following example, we need to create several traffic bandwidth
4303classes and we need different hosts/networks to fall into different classes.
4304We create one pipe for each class and configure them accordingly.
4305Then we create a single table and fill it with IP subnets and addresses.
4306For each subnet/host we set the argument equal to the number of the pipe
4307that it should use.
4308Then we classify traffic using a single rule:
4309.Pp
4310.Dl "ipfw pipe 1 config bw 1000Kbyte/s"
4311.Dl "ipfw pipe 4 config bw 4000Kbyte/s"
4312.Dl "..."
4313.Dl "ipfw table T1 create type addr"
4314.Dl "ipfw table T1 add 192.168.2.0/24 1"
4315.Dl "ipfw table T1 add 192.168.0.0/27 4"
4316.Dl "ipfw table T1 add 192.168.0.2 1"
4317.Dl "..."
4318.Dl "ipfw add pipe tablearg ip from 'table(T1)' to any"
4319.Pp
4320Using the
4321.Cm fwd
4322action, the table entries may include hostnames and IP addresses.
4323.Pp
4324.Dl "ipfw table T2 create type addr ftype ip"
4325.Dl "ipfw table T2 add 192.168.2.0/24 10.23.2.1"
4326.Dl "ipfw table T21 add 192.168.0.0/27 router1.dmz"
4327.Dl "..."
4328.Dl "ipfw add 100 fwd tablearg ip from any to table(1)"
4329.Pp
4330In the following example per-interface firewall is created:
4331.Pp
4332.Dl "ipfw table IN create type iface valtype skipto,fib"
4333.Dl "ipfw table IN add vlan20 12000,12"
4334.Dl "ipfw table IN add vlan30 13000,13"
4335.Dl "ipfw table OUT create type iface valtype skipto"
4336.Dl "ipfw table OUT add vlan20 22000"
4337.Dl "ipfw table OUT add vlan30 23000"
4338.Dl ".."
4339.Dl "ipfw add 100 setfib tablearg ip from any to any recv 'table(IN)' in"
4340.Dl "ipfw add 200 skipto tablearg ip from any to any recv 'table(IN)' in"
4341.Dl "ipfw add 300 skipto tablearg ip from any to any xmit 'table(OUT)' out"
4342.Pp
4343The following example illustrate usage of flow tables:
4344.Pp
4345.Dl "ipfw table fl create type flow:src-ip,proto,dst-ip,dst-port"
4346.Dl "ipfw table fl add 2a02:6b8:77::88,tcp,2a02:6b8:77::99,80 11"
4347.Dl "ipfw table fl add 10.0.0.1,udp,10.0.0.2,53 12"
4348.Dl ".."
4349.Dl "ipfw add 100 allow ip from any to any flow 'table(fl,11)' recv ix0"
4350.Ss SETS OF RULES
4351To add a set of rules atomically, e.g.\& set 18:
4352.Pp
4353.Dl "ipfw set disable 18"
4354.Dl "ipfw add NN set 18 ...         # repeat as needed"
4355.Dl "ipfw set enable 18"
4356.Pp
4357To delete a set of rules atomically the command is simply:
4358.Pp
4359.Dl "ipfw delete set 18"
4360.Pp
4361To test a ruleset and disable it and regain control if something goes wrong:
4362.Pp
4363.Dl "ipfw set disable 18"
4364.Dl "ipfw add NN set 18 ...         # repeat as needed"
4365.Dl "ipfw set enable 18; echo done; sleep 30 && ipfw set disable 18"
4366.Pp
4367Here if everything goes well, you press control-C before the "sleep"
4368terminates, and your ruleset will be left active.
4369Otherwise, e.g.\& if
4370you cannot access your box, the ruleset will be disabled after
4371the sleep terminates thus restoring the previous situation.
4372.Pp
4373To show rules of the specific set:
4374.Pp
4375.Dl "ipfw set 18 show"
4376.Pp
4377To show rules of the disabled set:
4378.Pp
4379.Dl "ipfw -S set 18 show"
4380.Pp
4381To clear a specific rule counters of the specific set:
4382.Pp
4383.Dl "ipfw set 18 zero NN"
4384.Pp
4385To delete a specific rule of the specific set:
4386.Pp
4387.Dl "ipfw set 18 delete NN"
4388.Ss NAT, REDIRECT AND LSNAT
4389First redirect all the traffic to nat instance 123:
4390.Pp
4391.Dl "ipfw add nat 123 all from any to any"
4392.Pp
4393Then to configure nat instance 123 to alias all the outgoing traffic with ip
4394192.168.0.123, blocking all incoming connections, trying to keep
4395same ports on both sides, clearing aliasing table on address change
4396and keeping a log of traffic/link statistics:
4397.Pp
4398.Dl "ipfw nat 123 config ip 192.168.0.123 log deny_in reset same_ports"
4399.Pp
4400Or to change address of instance 123, aliasing table will be cleared (see
4401reset option):
4402.Pp
4403.Dl "ipfw nat 123 config ip 10.0.0.1"
4404.Pp
4405To see configuration of nat instance 123:
4406.Pp
4407.Dl "ipfw nat 123 show config"
4408.Pp
4409To show logs of all the instances in range 111-999:
4410.Pp
4411.Dl "ipfw nat 111-999 show"
4412.Pp
4413To see configurations of all instances:
4414.Pp
4415.Dl "ipfw nat show config"
4416.Pp
4417Or a redirect rule with mixed modes could looks like:
4418.Pp
4419.Dl "ipfw nat 123 config redirect_addr 10.0.0.1 10.0.0.66"
4420.Dl "			 redirect_port tcp 192.168.0.1:80 500"
4421.Dl "			 redirect_proto udp 192.168.1.43 192.168.1.1"
4422.Dl "			 redirect_addr 192.168.0.10,192.168.0.11"
4423.Dl "			 	    10.0.0.100	# LSNAT"
4424.Dl "			 redirect_port tcp 192.168.0.1:80,192.168.0.10:22"
4425.Dl "			 	    500		# LSNAT"
4426.Pp
4427or it could be split in:
4428.Pp
4429.Dl "ipfw nat 1 config redirect_addr 10.0.0.1 10.0.0.66"
4430.Dl "ipfw nat 2 config redirect_port tcp 192.168.0.1:80 500"
4431.Dl "ipfw nat 3 config redirect_proto udp 192.168.1.43 192.168.1.1"
4432.Dl "ipfw nat 4 config redirect_addr 192.168.0.10,192.168.0.11,192.168.0.12"
4433.Dl "				         10.0.0.100"
4434.Dl "ipfw nat 5 config redirect_port tcp"
4435.Dl "			192.168.0.1:80,192.168.0.10:22,192.168.0.20:25 500"
4436.Pp
4437Sometimes you may want to mix NAT and dynamic rules. It could be achieved with
4438.Cm record-state
4439and
4440.Cm defer-action
4441options. Problem is, you need to create dynamic rule before NAT and check it
4442after NAT actions (or vice versa) to have consistent addresses and ports.
4443Rule with
4444.Cm keep-state
4445option will trigger activation of existing dynamic state, and action of such
4446rule will be performed as soon as rule is matched. In case of NAT and
4447.Cm allow
4448rule packet need to be passed to NAT, not allowed as soon is possible.
4449.Pp
4450There is example of set of rules to achieve this. Bear in mind that this
4451is exmaple only and it is not very useful by itself.
4452.Pp
4453On way out, after all checks place this rules:
4454.Pp
4455.Dl "ipfw add allow record-state skip-action"
4456.Dl "ipfw add nat 1"
4457.Pp
4458And on way in there should be something like this:
4459.Pp
4460.Dl "ipfw add nat 1"
4461.Dl "ipfw add check-state"
4462.Pp
4463Please note, that first rule on way out doesn't allow packet and doesn't
4464execute existing dynamic rules. All it does, create new dynamic rule with
4465.Cm allow
4466action, if it is not created yet. Later, this dynamic rule is used on way
4467in by
4468.Cm check-state
4469rule.
4470.Ss CONFIGURING CODEL, PIE, FQ-CODEL and FQ-PIE AQM
4471.Cm codel
4472and
4473.Cm pie
4474AQM can be configured for
4475.Nm dummynet
4476.Cm pipe
4477or
4478.Cm queue.
4479.Pp
4480To configure a
4481.Cm pipe
4482with
4483.Cm codel
4484AQM using default configuration for traffic from 192.168.0.0/24 and 1Mbits/s
4485rate limit, we do:
4486.Pp
4487.Dl "ipfw pipe 1 config bw 1mbits/s codel"
4488.Dl "ipfw add 100 pipe 1 ip from 192.168.0.0/24 to any"
4489.Pp
4490To configure a
4491.Cm queue
4492with
4493.Cm codel
4494AQM using different configurations parameters for traffic from
4495192.168.0.0/24 and 1Mbits/s rate limit, we do:
4496.Pp
4497.Dl "ipfw pipe 1 config bw 1mbits/s"
4498.Dl "ipfw queue 1 config pipe 1 codel target 8ms interval 160ms ecn"
4499.Dl "ipfw add 100 queue 1 ip from 192.168.0.0/24 to any"
4500.Pp
4501To configure a
4502.Cm pipe
4503with
4504.Cm pie
4505AQM using default configuration for traffic from 192.168.0.0/24 and 1Mbits/s
4506rate limit, we do:
4507.Pp
4508.Dl "ipfw pipe 1 config bw 1mbits/s pie"
4509.Dl "ipfw add 100 pipe 1 ip from 192.168.0.0/24 to any"
4510.Pp
4511To configure a
4512.Cm queue
4513with
4514.Cm pie
4515AQM using different configuration parameters for traffic from
4516192.168.0.0/24 and 1Mbits/s rate limit, we do:
4517.Pp
4518.Dl "ipfw pipe 1 config bw 1mbits/s"
4519.Dl "ipfw queue 1 config pipe 1 pie target 20ms tupdate 30ms ecn"
4520.Dl "ipfw add 100 queue 1 ip from 192.168.0.0/24 to any"
4521.Pp
4522.Cm fq_codel
4523and
4524.Cm fq_pie
4525AQM can be configured for
4526.Nm dummynet
4527schedulers.
4528.Pp
4529To configure
4530.Cm fq_codel
4531scheduler using different configurations parameters for traffic from
4532192.168.0.0/24 and 1Mbits/s rate limit, we do:
4533.Pp
4534.Dl "ipfw pipe 1 config bw 1mbits/s"
4535.Dl "ipfw sched 1 config pipe 1 type fq_codel"
4536.Dl "ipfw queue 1 config sched 1"
4537.Dl "ipfw add 100 queue 1 ip from 192.168.0.0/24 to any"
4538.Pp
4539To change
4540.Cm fq_codel
4541default configuration for a
4542.Cm sched
4543such as disable ECN and change the
4544.Ar target
4545to 10ms, we do:
4546.Pp
4547.Dl "ipfw sched 1 config pipe 1 type fq_codel target 10ms noecn"
4548.Pp
4549Similar to
4550.Cm fq_codel,
4551to configure
4552.Cm fq_pie
4553scheduler using different configurations parameters for traffic from
4554192.168.0.0/24 and 1Mbits/s rate limit, we do:
4555.Pp
4556.Dl "ipfw pipe 1 config bw 1mbits/s"
4557.Dl "ipfw sched 1 config pipe 1 type fq_pie"
4558.Dl "ipfw queue 1 config sched 1"
4559.Dl "ipfw add 100 queue 1 ip from 192.168.0.0/24 to any"
4560.Pp
4561The configurations of
4562.Cm fq_pie
4563.Cm sched
4564can be changed in a similar way as for
4565.Cm fq_codel
4566.Sh SEE ALSO
4567.Xr cpp 1 ,
4568.Xr m4 1 ,
4569.Xr altq 4 ,
4570.Xr divert 4 ,
4571.Xr dummynet 4 ,
4572.Xr if_bridge 4 ,
4573.Xr ip 4 ,
4574.Xr ipfirewall 4 ,
4575.Xr ng_ether 4 ,
4576.Xr ng_ipfw 4 ,
4577.Xr protocols 5 ,
4578.Xr services 5 ,
4579.Xr init 8 ,
4580.Xr kldload 8 ,
4581.Xr reboot 8 ,
4582.Xr sysctl 8 ,
4583.Xr sysrc 8 ,
4584.Xr syslogd 8
4585.Sh HISTORY
4586The
4587.Nm
4588utility first appeared in
4589.Fx 2.0 .
4590.Nm dummynet
4591was introduced in
4592.Fx 2.2.8 .
4593Stateful extensions were introduced in
4594.Fx 4.0 .
4595.Nm ipfw2
4596was introduced in Summer 2002.
4597.Sh AUTHORS
4598.An Ugen J. S. Antsilevich ,
4599.An Poul-Henning Kamp ,
4600.An Alex Nash ,
4601.An Archie Cobbs ,
4602.An Luigi Rizzo ,
4603.An Rasool Al-Saadi .
4604.Pp
4605.An -nosplit
4606API based upon code written by
4607.An Daniel Boulet
4608for BSDI.
4609.Pp
4610Dummynet has been introduced by Luigi Rizzo in 1997-1998.
4611.Pp
4612Some early work (1999-2000) on the
4613.Nm dummynet
4614traffic shaper supported by Akamba Corp.
4615.Pp
4616The ipfw core (ipfw2) has been completely redesigned and
4617reimplemented by Luigi Rizzo in summer 2002.
4618Further
4619actions and
4620options have been added by various developers over the years.
4621.Pp
4622.An -nosplit
4623In-kernel NAT support written by
4624.An Paolo Pisati Aq Mt piso@FreeBSD.org
4625as part of a Summer of Code 2005 project.
4626.Pp
4627SCTP
4628.Nm nat
4629support has been developed by
4630.An The Centre for Advanced Internet Architectures (CAIA) Aq http://www.caia.swin.edu.au .
4631The primary developers and maintainers are David Hayes and Jason But.
4632For further information visit:
4633.Aq http://www.caia.swin.edu.au/urp/SONATA
4634.Pp
4635Delay profiles have been developed by Alessandro Cerri and
4636Luigi Rizzo, supported by the
4637European Commission within Projects Onelab and Onelab2.
4638.Pp
4639CoDel, PIE, FQ-CoDel and FQ-PIE AQM for Dummynet have been implemented by
4640.An The Centre for Advanced Internet Architectures (CAIA)
4641in 2016, supported by The Comcast Innovation Fund.
4642The primary developer is
4643Rasool Al-Saadi.
4644.Sh BUGS
4645The syntax has grown over the years and sometimes it might be confusing.
4646Unfortunately, backward compatibility prevents cleaning up mistakes
4647made in the definition of the syntax.
4648.Pp
4649.Em !!! WARNING !!!
4650.Pp
4651Misconfiguring the firewall can put your computer in an unusable state,
4652possibly shutting down network services and requiring console access to
4653regain control of it.
4654.Pp
4655Incoming packet fragments diverted by
4656.Cm divert
4657are reassembled before delivery to the socket.
4658The action used on those packet is the one from the
4659rule which matches the first fragment of the packet.
4660.Pp
4661Packets diverted to userland, and then reinserted by a userland process
4662may lose various packet attributes.
4663The packet source interface name
4664will be preserved if it is shorter than 8 bytes and the userland process
4665saves and reuses the sockaddr_in
4666(as does
4667.Xr natd 8 ) ;
4668otherwise, it may be lost.
4669If a packet is reinserted in this manner, later rules may be incorrectly
4670applied, making the order of
4671.Cm divert
4672rules in the rule sequence very important.
4673.Pp
4674Dummynet drops all packets with IPv6 link-local addresses.
4675.Pp
4676Rules using
4677.Cm uid
4678or
4679.Cm gid
4680may not behave as expected.
4681In particular, incoming SYN packets may
4682have no uid or gid associated with them since they do not yet belong
4683to a TCP connection, and the uid/gid associated with a packet may not
4684be as expected if the associated process calls
4685.Xr setuid 2
4686or similar system calls.
4687.Pp
4688Rule syntax is subject to the command line environment and some patterns
4689may need to be escaped with the backslash character
4690or quoted appropriately.
4691.Pp
4692Due to the architecture of
4693.Xr libalias 3 ,
4694ipfw nat is not compatible with the TCP segmentation offloading (TSO).
4695Thus, to reliably nat your network traffic, please disable TSO
4696on your NICs using
4697.Xr ifconfig 8 .
4698.Pp
4699ICMP error messages are not implicitly matched by dynamic rules
4700for the respective conversations.
4701To avoid failures of network error detection and path MTU discovery,
4702ICMP error messages may need to be allowed explicitly through static
4703rules.
4704.Pp
4705Rules using
4706.Cm call
4707and
4708.Cm return
4709actions may lead to confusing behaviour if ruleset has mistakes,
4710and/or interaction with other subsystems (netgraph, dummynet, etc.) is used.
4711One possible case for this is packet leaving
4712.Nm
4713in subroutine on the input pass, while later on output encountering unpaired
4714.Cm return
4715first.
4716As the call stack is kept intact after input pass, packet will suddenly
4717return to the rule number used on input pass, not on output one.
4718Order of processing should be checked carefully to avoid such mistakes.
4719