xref: /freebsd/share/man/man5/pf.conf.5 (revision 7899f917b1c0ea178f1d2be0cfb452086d079d23)
1.\"	$OpenBSD: pf.conf.5,v 1.406 2009/01/31 19:37:12 sobrado Exp $
2.\"
3.\" Copyright (c) 2002, Daniel Hartmeier
4.\" All rights reserved.
5.\"
6.\" Redistribution and use in source and binary forms, with or without
7.\" modification, are permitted provided that the following conditions
8.\" are met:
9.\"
10.\"    - Redistributions of source code must retain the above copyright
11.\"      notice, this list of conditions and the following disclaimer.
12.\"    - Redistributions in binary form must reproduce the above
13.\"      copyright notice, this list of conditions and the following
14.\"      disclaimer in the documentation and/or other materials provided
15.\"      with the distribution.
16.\"
17.\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
18.\" "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
19.\" LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
20.\" FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
21.\" COPYRIGHT HOLDERS OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
22.\" INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
23.\" BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
24.\" LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
25.\" CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
26.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
27.\" ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
28.\" POSSIBILITY OF SUCH DAMAGE.
29.\"
30.Dd September 4, 2024
31.Dt PF.CONF 5
32.Os
33.Sh NAME
34.Nm pf.conf
35.Nd packet filter configuration file
36.Sh DESCRIPTION
37The
38.Xr pf 4
39packet filter modifies, drops or passes packets according to rules or
40definitions specified in
41.Nm pf.conf .
42.Sh STATEMENT ORDER
43There are eight types of statements in
44.Nm pf.conf :
45.Bl -tag -width xxxx
46.It Cm Macros
47User-defined variables may be defined and used later, simplifying
48the configuration file.
49Macros must be defined before they are referenced in
50.Nm pf.conf .
51.It Cm Tables
52Tables provide a mechanism for increasing the performance and flexibility of
53rules with large numbers of source or destination addresses.
54.It Cm Options
55Options tune the behaviour of the packet filtering engine.
56.It Cm Ethernet Filtering
57Ethernet filtering provides rule-based blocking or passing of Ethernet packets.
58.It Cm Traffic Normalization Li (e.g. Em scrub )
59Traffic normalization protects internal machines against inconsistencies
60in Internet protocols and implementations.
61.It Cm Queueing
62Queueing provides rule-based bandwidth control.
63.It Cm Translation Li (Various forms of NAT)
64Translation rules specify how addresses are to be mapped or redirected to
65other addresses.
66.It Cm Packet Filtering
67Packet filtering provides rule-based blocking or passing of packets.
68.El
69.Pp
70With the exception of
71.Cm macros
72and
73.Cm tables ,
74the types of statements should be grouped and appear in
75.Nm pf.conf
76in the order shown above, as this matches the operation of the underlying
77packet filtering engine.
78By default
79.Xr pfctl 8
80enforces this order (see
81.Ar set require-order
82below).
83.Pp
84Comments can be put anywhere in the file using a hash mark
85.Pq Sq # ,
86and extend to the end of the current line.
87.Pp
88Additional configuration files can be included with the
89.Ic include
90keyword, for example:
91.Bd -literal -offset indent
92include "/etc/pf/sub.filter.conf"
93.Ed
94.Sh MACROS
95Macros can be defined that will later be expanded in context.
96Macro names must start with a letter, and may contain letters, digits
97and underscores.
98Macro names may not be reserved words (for example
99.Ar pass ,
100.Ar in ,
101.Ar out ) .
102Macros are not expanded inside quotes.
103.Pp
104For example,
105.Bd -literal -offset indent
106ext_if = \&"kue0\&"
107all_ifs = \&"{\&" $ext_if lo0 \&"}\&"
108pass out on $ext_if from any to any
109pass in  on $ext_if proto tcp from any to any port 25
110.Ed
111.Sh TABLES
112Tables are named structures which can hold a collection of addresses and
113networks.
114Lookups against tables in
115.Xr pf 4
116are relatively fast, making a single rule with tables much more efficient,
117in terms of
118processor usage and memory consumption, than a large number of rules which
119differ only in IP address (either created explicitly or automatically by rule
120expansion).
121.Pp
122Tables can be used as the source or destination of filter rules,
123.Ar scrub
124rules
125or
126translation rules such as
127.Ar nat
128or
129.Ar rdr
130(see below for details on the various rule types).
131Tables can also be used for the redirect address of
132.Ar nat
133and
134.Ar rdr
135rules and in the routing options of filter rules, but only for
136.Ar round-robin
137pools.
138.Pp
139Tables can be defined with any of the following
140.Xr pfctl 8
141mechanisms.
142As with macros, reserved words may not be used as table names.
143.Bl -tag -width "manually"
144.It Ar manually
145Persistent tables can be manually created with the
146.Ar add
147or
148.Ar replace
149option of
150.Xr pfctl 8 ,
151before or after the ruleset has been loaded.
152.It Pa pf.conf
153Table definitions can be placed directly in this file, and loaded at the
154same time as other rules are loaded, atomically.
155Table definitions inside
156.Nm pf.conf
157use the
158.Ar table
159statement, and are especially useful to define non-persistent tables.
160The contents of a pre-existing table defined without a list of addresses
161to initialize it is not altered when
162.Nm pf.conf
163is loaded.
164A table initialized with the empty list,
165.Li { } ,
166will be cleared on load.
167.El
168.Pp
169Tables may be defined with the following attributes:
170.Bl -tag -width persist
171.It Ar persist
172The
173.Ar persist
174flag forces the kernel to keep the table even when no rules refer to it.
175If the flag is not set, the kernel will automatically remove the table
176when the last rule referring to it is flushed.
177.It Ar const
178The
179.Ar const
180flag prevents the user from altering the contents of the table once it
181has been created.
182Without that flag,
183.Xr pfctl 8
184can be used to add or remove addresses from the table at any time, even
185when running with
186.Xr securelevel 7
187= 2.
188.It Ar counters
189The
190.Ar counters
191flag enables per-address packet and byte counters which can be displayed with
192.Xr pfctl 8 .
193Note that this feature carries significant memory overhead for large tables.
194.El
195.Pp
196For example,
197.Bd -literal -offset indent
198table \*(Ltprivate\*(Gt const { 10/8, 172.16/12, 192.168/16 }
199table \*(Ltbadhosts\*(Gt persist
200block on fxp0 from { \*(Ltprivate\*(Gt, \*(Ltbadhosts\*(Gt } to any
201.Ed
202.Pp
203creates a table called private, to hold RFC 1918 private network
204blocks, and a table called badhosts, which is initially empty.
205A filter rule is set up to block all traffic coming from addresses listed in
206either table.
207The private table cannot have its contents changed and the badhosts table
208will exist even when no active filter rules reference it.
209Addresses may later be added to the badhosts table, so that traffic from
210these hosts can be blocked by using
211.Bd -literal -offset indent
212# pfctl -t badhosts -Tadd 204.92.77.111
213.Ed
214.Pp
215A table can also be initialized with an address list specified in one or more
216external files, using the following syntax:
217.Bd -literal -offset indent
218table \*(Ltspam\*(Gt persist file \&"/etc/spammers\&" file \&"/etc/openrelays\&"
219block on fxp0 from \*(Ltspam\*(Gt to any
220.Ed
221.Pp
222The files
223.Pa /etc/spammers
224and
225.Pa /etc/openrelays
226list IP addresses, one per line.
227Any lines beginning with a # are treated as comments and ignored.
228In addition to being specified by IP address, hosts may also be
229specified by their hostname.
230When the resolver is called to add a hostname to a table,
231.Em all
232resulting IPv4 and IPv6 addresses are placed into the table.
233IP addresses can also be entered in a table by specifying a valid interface
234name, a valid interface group or the
235.Em self
236keyword, in which case all addresses assigned to the interface(s) will be
237added to the table.
238.Sh OPTIONS
239.Xr pf 4
240may be tuned for various situations using the
241.Ar set
242command.
243.Bl -tag -width xxxx
244.It Ar set timeout
245.Pp
246.Bl -tag -width "src.track" -compact
247.It Ar interval
248Interval between purging expired states and fragments.
249.It Ar frag
250Seconds before an unassembled fragment is expired.
251.It Ar src.track
252Length of time to retain a source tracking entry after the last state
253expires.
254.El
255.Pp
256When a packet matches a stateful connection, the seconds to live for the
257connection will be updated to that of the
258.Ar proto.modifier
259which corresponds to the connection state.
260Each packet which matches this state will reset the TTL.
261Tuning these values may improve the performance of the
262firewall at the risk of dropping valid idle connections.
263.Pp
264.Bl -tag -width xxxx -compact
265.It Ar tcp.first
266The state after the first packet.
267.It Ar tcp.opening
268The state before the destination host ever sends a packet.
269.It Ar tcp.established
270The fully established state.
271.It Ar tcp.closing
272The state after the first FIN has been sent.
273.It Ar tcp.finwait
274The state after both FINs have been exchanged and the connection is closed.
275Some hosts (notably web servers on Solaris) send TCP packets even after closing
276the connection.
277Increasing
278.Ar tcp.finwait
279(and possibly
280.Ar tcp.closing )
281can prevent blocking of such packets.
282.It Ar tcp.closed
283The state after one endpoint sends an RST.
284.El
285.Pp
286SCTP timeout are handled similar to TCP, but with its own set of states:
287.Pp
288.Bl -tag -width xxxx -compact
289.It Ar sctp.first
290The state after the first packet.
291.It Ar sctp.opening
292The state before the destination host ever sends a packet.
293.It Ar sctp.established
294The fully established state.
295.It Ar sctp.closing
296The state after the first SHUTDOWN chunk has been sent.
297.It Ar sctp.closed
298The state after SHUTDOWN_ACK has been exchanged and the connection is closed.
299.El
300.Pp
301ICMP and UDP are handled in a fashion similar to TCP, but with a much more
302limited set of states:
303.Pp
304.Bl -tag -width xxxx -compact
305.It Ar udp.first
306The state after the first packet.
307.It Ar udp.single
308The state if the source host sends more than one packet but the destination
309host has never sent one back.
310.It Ar udp.multiple
311The state if both hosts have sent packets.
312.It Ar icmp.first
313The state after the first packet.
314.It Ar icmp.error
315The state after an ICMP error came back in response to an ICMP packet.
316.El
317.Pp
318Other protocols are handled similarly to UDP:
319.Pp
320.Bl -tag -width xxxx -compact
321.It Ar other.first
322.It Ar other.single
323.It Ar other.multiple
324.El
325.Pp
326Timeout values can be reduced adaptively as the number of state table
327entries grows.
328.Pp
329.Bl -tag -width xxxx -compact
330.It Ar adaptive.start
331When the number of state entries exceeds this value, adaptive scaling
332begins.
333All timeout values are scaled linearly with factor
334(adaptive.end - number of states) / (adaptive.end - adaptive.start).
335.It Ar adaptive.end
336When reaching this number of state entries, all timeout values become
337zero, effectively purging all state entries immediately.
338This value is used to define the scale factor, it should not actually
339be reached (set a lower state limit, see below).
340.El
341.Pp
342Adaptive timeouts are enabled by default, with an adaptive.start value
343equal to 60% of the state limit, and an adaptive.end value equal to
344120% of the state limit.
345They can be disabled by setting both adaptive.start and adaptive.end to 0.
346.Pp
347The adaptive timeout values can be defined both globally and for each rule.
348When used on a per-rule basis, the values relate to the number of
349states created by the rule, otherwise to the total number of
350states.
351.Pp
352For example:
353.Bd -literal -offset indent
354set timeout tcp.first 120
355set timeout tcp.established 86400
356set timeout { adaptive.start 6000, adaptive.end 12000 }
357set limit states 10000
358.Ed
359.Pp
360With 9000 state table entries, the timeout values are scaled to 50%
361(tcp.first 60, tcp.established 43200).
362.It Ar set loginterface
363Enable collection of packet and byte count statistics for the given
364interface or interface group.
365These statistics can be viewed using
366.Bd -literal -offset indent
367# pfctl -s info
368.Ed
369.Pp
370In this example
371.Xr pf 4
372collects statistics on the interface named dc0:
373.Bd -literal -offset indent
374set loginterface dc0
375.Ed
376.Pp
377One can disable the loginterface using:
378.Bd -literal -offset indent
379set loginterface none
380.Ed
381.It Ar set limit
382Sets hard limits on the memory pools used by the packet filter.
383See
384.Xr zone 9
385for an explanation of memory pools.
386.Pp
387For example,
388.Bd -literal -offset indent
389set limit states 20000
390.Ed
391.Pp
392sets the maximum number of entries in the memory pool used by state table
393entries (generated by
394.Ar pass
395rules which do not specify
396.Ar no state )
397to 20000.
398Using
399.Bd -literal -offset indent
400set limit frags 20000
401.Ed
402.Pp
403sets the maximum number of entries in the memory pool used for fragment
404reassembly (generated by the
405.Ar set reassemble
406option or
407.Ar scrub
408rules) to 20000.
409Using
410.Bd -literal -offset indent
411set limit src-nodes 2000
412.Ed
413.Pp
414sets the maximum number of entries in the memory pool used for tracking
415source IP addresses (generated by the
416.Ar sticky-address
417and
418.Ar src.track
419options) to 2000.
420Using
421.Bd -literal -offset indent
422set limit table-entries 100000
423.Ed
424.Pp
425sets the limit on the overall number of addresses that can be stored
426in tables to 100000.
427.Pp
428Various limits can be combined on a single line:
429.Bd -literal -offset indent
430set limit { states 20000, frags 20000, src-nodes 2000 }
431.Ed
432.It Ar set ruleset-optimization
433.Bl -tag -width xxxxxxxx -compact
434.It Ar none
435Disable the ruleset optimizer.
436.It Ar basic
437Enable basic ruleset optimization.
438This is the default behaviour.
439Basic ruleset optimization does four things to improve the
440performance of ruleset evaluations:
441.Pp
442.Bl -enum -compact
443.It
444remove duplicate rules
445.It
446remove rules that are a subset of another rule
447.It
448combine multiple rules into a table when advantageous
449.It
450re-order the rules to improve evaluation performance
451.El
452.Pp
453.It Ar profile
454Uses the currently loaded ruleset as a feedback profile to tailor the
455ordering of quick rules to actual network traffic.
456.El
457.Pp
458It is important to note that the ruleset optimizer will modify the ruleset
459to improve performance.
460A side effect of the ruleset modification is that per-rule accounting
461statistics will have different meanings than before.
462If per-rule accounting is important for billing purposes or whatnot,
463either the ruleset optimizer should not be used or a label field should
464be added to all of the accounting rules to act as optimization barriers.
465.Pp
466Optimization can also be set as a command-line argument to
467.Xr pfctl 8 ,
468overriding the settings in
469.Nm .
470.It Ar set optimization
471Optimize state timeouts for one of the following network environments:
472.Pp
473.Bl -tag -width xxxx -compact
474.It Ar normal
475A normal network environment.
476Suitable for almost all networks.
477.It Ar high-latency
478A high-latency environment (such as a satellite connection).
479.It Ar satellite
480Alias for
481.Ar high-latency .
482.It Ar aggressive
483Aggressively expire connections.
484This can greatly reduce the memory usage of the firewall at the cost of
485dropping idle connections early.
486.It Ar conservative
487Extremely conservative settings.
488Avoid dropping legitimate connections at the
489expense of greater memory utilization (possibly much greater on a busy
490network) and slightly increased processor utilization.
491.El
492.Pp
493For example:
494.Bd -literal -offset indent
495set optimization aggressive
496.Ed
497.It Ar set reassemble yes | no Op Cm no-df
498The
499.Cm reassemble
500option is used to enable or disable the reassembly of fragmented packets,
501and can be set to
502.Cm yes
503or
504.Cm no .
505If
506.Cm no-df
507is also specified, fragments with the
508.Dq dont-fragment
509bit set are reassembled too,
510instead of being dropped;
511the reassembled packet will have the
512.Dq dont-fragment
513bit cleared.
514The default value is
515.Cm no .
516.Pp
517This option is ignored if there are pre-FreeBSD 14
518.Cm scrub
519rules present.
520.It Ar set block-policy
521The
522.Ar block-policy
523option sets the default behaviour for the packet
524.Ar block
525action:
526.Pp
527.Bl -tag -width xxxxxxxx -compact
528.It Ar drop
529Packet is silently dropped.
530.It Ar return
531A TCP RST is returned for blocked TCP packets,
532an SCTP ABORT chunk is returned for blocked SCTP packets,
533an ICMP UNREACHABLE is returned for blocked UDP packets,
534and all other packets are silently dropped.
535.El
536.Pp
537For example:
538.Bd -literal -offset indent
539set block-policy return
540.Ed
541.It Ar set fail-policy
542The
543.Ar fail-policy
544option sets the behaviour of rules which should pass a packet but were
545unable to do so.
546This might happen when a nat or route-to rule uses an empty table as list
547of targets or if a rule fails to create state or source node.
548The following
549.Ar block
550actions are possible:
551.Pp
552.Bl -tag -width xxxxxxxx -compact
553.It Ar drop
554Incoming packet is silently dropped.
555.It Ar return
556Incoming packet is dropped and TCP RST is returned for TCP packets,
557an SCTP ABORT chunk is returned for blocked SCTP packets,
558an ICMP UNREACHABLE is returned for UDP packets,
559and no response is sent for other packets.
560.El
561.Pp
562For example:
563.Bd -literal -offset indent
564set fail-policy return
565.Ed
566.It Ar set state-policy
567The
568.Ar state-policy
569option sets the default behaviour for states:
570.Pp
571.Bl -tag -width group-bound -compact
572.It Ar if-bound
573States are bound to interface.
574.It Ar floating
575States can match packets on any interfaces (the default).
576.El
577.Pp
578For example:
579.Bd -literal -offset indent
580set state-policy if-bound
581.Ed
582.It Ar set syncookies never | always | adaptive
583When
584.Cm syncookies
585are active, pf will answer each incoming TCP SYN with a syncookie SYNACK,
586without allocating any resources.
587Upon reception of the client's ACK in response to the syncookie
588SYNACK, pf will evaluate the ruleset and create state if the ruleset
589permits it, complete the three way handshake with the target host and
590continue the connection with synproxy in place.
591This allows pf to be resilient against large synflood attacks which would
592run the state table against its limits otherwise.
593Due to the blind answers to every incoming SYN syncookies share the caveats of
594synproxy, namely seemingly accepting connections that will be dropped later on.
595.Pp
596.Bl -tag -width adaptive -compact
597.It Cm never
598pf will never send syncookie SYNACKs (the default).
599.It Cm always
600pf will always send syncookie SYNACKs.
601.It Cm adaptive
602pf will enable syncookie mode when a given percentage of the state table
603is used up by half-open TCP connections, as in, those that saw the initial
604SYN but didn't finish the three way handshake.
605The thresholds for entering and leaving syncookie mode can be specified using
606.Bd -literal -offset indent
607set syncookies adaptive (start 25%, end 12%)
608.Ed
609.El
610.It Ar set state-defaults
611The
612.Ar state-defaults
613option sets the state options for states created from rules
614without an explicit
615.Ar keep state .
616For example:
617.Bd -literal -offset indent
618set state-defaults no-sync
619.Ed
620.It Ar set hostid
621The 32-bit
622.Ar hostid
623identifies this firewall's state table entries to other firewalls
624in a
625.Xr pfsync 4
626failover cluster.
627By default the hostid is set to a pseudo-random value, however it may be
628desirable to manually configure it, for example to more easily identify the
629source of state table entries.
630.Bd -literal -offset indent
631set hostid 1
632.Ed
633.Pp
634The hostid may be specified in either decimal or hexadecimal.
635.It Ar set require-order
636By default
637.Xr pfctl 8
638enforces an ordering of the statement types in the ruleset to:
639.Em options ,
640.Em normalization ,
641.Em queueing ,
642.Em translation ,
643.Em filtering .
644Setting this option to
645.Ar no
646disables this enforcement.
647There may be non-trivial and non-obvious implications to an out of
648order ruleset.
649Consider carefully before disabling the order enforcement.
650.It Ar set fingerprints
651Load fingerprints of known operating systems from the given filename.
652By default fingerprints of known operating systems are automatically
653loaded from
654.Xr pf.os 5
655in
656.Pa /etc
657but can be overridden via this option.
658Setting this option may leave a small period of time where the fingerprints
659referenced by the currently active ruleset are inconsistent until the new
660ruleset finishes loading.
661.Pp
662For example:
663.Pp
664.Dl set fingerprints \&"/etc/pf.os.devel\&"
665.It Ar set skip on Aq Ar ifspec
666List interfaces for which packets should not be filtered.
667Packets passing in or out on such interfaces are passed as if pf was
668disabled, i.e. pf does not process them in any way.
669This can be useful on loopback and other virtual interfaces, when
670packet filtering is not desired and can have unexpected effects.
671For example:
672.Pp
673.Dl set skip on lo0
674.It Ar set debug
675Set the debug
676.Ar level
677to one of the following:
678.Pp
679.Bl -tag -width xxxxxxxxxxxx -compact
680.It Ar none
681Don't generate debug messages.
682.It Ar urgent
683Generate debug messages only for serious errors.
684.It Ar misc
685Generate debug messages for various errors.
686.It Ar loud
687Generate debug messages for common conditions.
688.El
689.It Ar set keepcounters
690Preserve rule counters across rule updates.
691Usually rule counters are reset to zero on every update of the ruleset.
692With
693.Ar keepcounters
694set pf will attempt to find matching rules between old and new rulesets
695and preserve the rule counters.
696.El
697.Sh ETHERNET FILTERING
698.Xr pf 4
699has the ability to
700.Ar block
701and
702.Ar pass
703packets based on attributes of their Ethernet (layer 2) header.
704.Pp
705For each packet processed by the packet filter, the filter rules are
706evaluated in sequential order, from first to last.
707The last matching rule decides what action is taken.
708If no rule matches the packet, the default action is to pass
709the packet.
710.Pp
711The following actions can be used in the filter:
712.Bl -tag -width xxxx
713.It Ar block
714The packet is blocked.
715Unlike for layer 3 traffic the packet is always silently dropped.
716.It Ar pass
717The packet is passed;
718no state is created for layer 2 traffic.
719.El
720.Sh PARAMETERS
721The rule parameters specify the packets to which a rule applies.
722A packet always comes in on, or goes out through, one interface.
723Most parameters are optional.
724If a parameter is specified, the rule only applies to packets with
725matching attributes.
726Certain parameters can be expressed as lists, in which case
727.Xr pfctl 8
728generates all needed rule combinations.
729.Bl -tag -width xxxx
730.It Ar in No or Ar out
731This rule applies to incoming or outgoing packets.
732If neither
733.Ar in
734nor
735.Ar out
736are specified, the rule will match packets in both directions.
737.It Ar quick
738If a packet matches a rule which has the
739.Ar quick
740option set, this rule
741is considered the last matching rule, and evaluation of subsequent rules
742is skipped.
743.It Ar on Aq Ar ifspec
744This rule applies only to packets coming in on, or going out through, this
745particular interface or interface group.
746For more information on interface groups,
747see the
748.Ic group
749keyword in
750.Xr ifconfig 8 .
751.It Ar bridge-to Aq interface
752Packets matching this rule will be sent out of the specified interface without
753further processing.
754.It Ar proto Aq Ar protocol
755This rule applies only to packets of this protocol.
756Note that Ethernet protocol numbers are different from those used in
757.Xr ip 4
758and
759.Xr ip6 4 .
760.It Xo
761.Ar from Aq Ar source
762.Ar to Aq Ar dest
763.Xc
764This rule applies only to packets with the specified source and destination
765MAC addresses.
766.It Xo Ar queue Aq Ar queue
767.Xc
768Packets matching this rule will be assigned to the specified queue.
769See
770.Sx QUEUEING
771for setup details.
772.Pp
773.It Ar tag Aq Ar string
774Packets matching this rule will be tagged with the
775specified string.
776The tag acts as an internal marker that can be used to
777identify these packets later on.
778This can be used, for example, to provide trust between
779interfaces and to determine if packets have been
780processed by translation rules.
781Tags are
782.Qq sticky ,
783meaning that the packet will be tagged even if the rule
784is not the last matching rule.
785Further matching rules can replace the tag with a
786new one but will not remove a previously applied tag.
787A packet is only ever assigned one tag at a time.
788.It Ar tagged Aq Ar string
789Used to specify that packets must already be tagged with the given tag in order
790to match the rule.
791Inverse tag matching can also be done by specifying the !  operator before the
792tagged keyword.
793.El
794.Sh TRAFFIC NORMALIZATION
795Traffic normalization is a broad umbrella term
796for aspects of the packet filter which deal with
797verifying packets, packet fragments, spoofed traffic,
798and other irregularities.
799.Ss Scrub
800Scrub involves sanitising packet content in such a way
801that there are no ambiguities in packet interpretation on the receiving side.
802It is invoked with the
803.Cm scrub
804option, added to filter rules.
805.Pp
806Parameters are specified enclosed in parentheses.
807At least one of the following parameters must be specified:
808.Bl -tag -width xxxx
809.It Ar no-df
810Clears the
811.Ar dont-fragment
812bit from a matching IP packet.
813Some operating systems are known to generate fragmented packets with the
814.Ar dont-fragment
815bit set.
816This is particularly true with NFS.
817.Ar Scrub
818will drop such fragmented
819.Ar dont-fragment
820packets unless
821.Ar no-df
822is specified.
823.Pp
824Unfortunately some operating systems also generate their
825.Ar dont-fragment
826packets with a zero IP identification field.
827Clearing the
828.Ar dont-fragment
829bit on packets with a zero IP ID may cause deleterious results if an
830upstream router later fragments the packet.
831Using the
832.Ar random-id
833modifier (see below) is recommended in combination with the
834.Ar no-df
835modifier to ensure unique IP identifiers.
836.It Ar min-ttl Aq Ar number
837Enforces a minimum TTL for matching IP packets.
838.It Ar max-mss Aq Ar number
839Enforces a maximum MSS for matching TCP packets.
840.It Xo Ar set-tos Aq Ar string
841.No \*(Ba Aq Ar number
842.Xc
843Enforces a
844.Em TOS
845for matching IP packets.
846.Em TOS
847may be
848given as one of
849.Ar critical ,
850.Ar inetcontrol ,
851.Ar lowdelay ,
852.Ar netcontrol ,
853.Ar throughput ,
854.Ar reliability ,
855or one of the DiffServ Code Points:
856.Ar ef ,
857.Ar va ,
858.Ar af11 No ... Ar af43 ,
859.Ar cs0 No ... Ar cs7 ;
860or as either hex or decimal.
861.It Ar random-id
862Replaces the IP identification field with random values to compensate
863for predictable values generated by many hosts.
864This option only applies to packets that are not fragmented
865after the optional fragment reassembly.
866.It Ar reassemble tcp
867Statefully normalizes TCP connections.
868.Ar reassemble tcp
869performs the following normalizations:
870.Pp
871.Bl -tag -width timeout -compact
872.It ttl
873Neither side of the connection is allowed to reduce their IP TTL.
874An attacker may send a packet such that it reaches the firewall, affects
875the firewall state, and expires before reaching the destination host.
876.Ar reassemble tcp
877will raise the TTL of all packets back up to the highest value seen on
878the connection.
879.It timestamp modulation
880Modern TCP stacks will send a timestamp on every TCP packet and echo
881the other endpoint's timestamp back to them.
882Many operating systems will merely start the timestamp at zero when
883first booted, and increment it several times a second.
884The uptime of the host can be deduced by reading the timestamp and multiplying
885by a constant.
886Also observing several different timestamps can be used to count hosts
887behind a NAT device.
888And spoofing TCP packets into a connection requires knowing or guessing
889valid timestamps.
890Timestamps merely need to be monotonically increasing and not derived off a
891guessable base time.
892.Ar reassemble tcp
893will cause
894.Ar scrub
895to modulate the TCP timestamps with a random number.
896.It extended PAWS checks
897There is a problem with TCP on long fat pipes, in that a packet might get
898delayed for longer than it takes the connection to wrap its 32-bit sequence
899space.
900In such an occurrence, the old packet would be indistinguishable from a
901new packet and would be accepted as such.
902The solution to this is called PAWS: Protection Against Wrapped Sequence
903numbers.
904It protects against it by making sure the timestamp on each packet does
905not go backwards.
906.Ar reassemble tcp
907also makes sure the timestamp on the packet does not go forward more
908than the RFC allows.
909By doing this,
910.Xr pf 4
911artificially extends the security of TCP sequence numbers by 10 to 18
912bits when the host uses appropriately randomized timestamps, since a
913blind attacker would have to guess the timestamp as well.
914.El
915.El
916.Pp
917For example,
918.Bd -literal -offset indent
919match in all scrub (no-df random-id max-mss 1440)
920.Ed
921.Ss Scrub ruleset (pre-FreeBSD 14)
922In order to maintain compatibility with older releases of FreeBSD
923.Ar scrub
924rules can also be specified in their own ruleset.
925In such case they are invoked with the
926.Ar scrub
927directive.
928If there are such rules present they determine packet reassembly behaviour.
929When no such rules are present the option
930.Ar set reassembly
931takes precedence.
932The
933.Ar scrub
934rules can take all parameters specified above for a
935.Ar scrub
936option of filter rules and 2 more parameters controlling fragment reassembly:
937.Bl -tag -width xxxx
938.It Ar fragment reassemble
939Using
940.Ar scrub
941rules, fragments can be reassembled by normalization.
942In this case, fragments are buffered until they form a complete
943packet, and only the completed packet is passed on to the filter.
944The advantage is that filter rules have to deal only with complete
945packets, and can ignore fragments.
946The drawback of caching fragments is the additional memory cost.
947This is the default behaviour unless no fragment reassemble is specified.
948.It Ar no fragment reassemble
949Do not reassemble fragments.
950.El
951.Pp
952For example,
953.Bd -literal -offset indent
954scrub in on $ext_if all fragment reassemble
955.Ed
956.Pp
957The
958.Ar no
959option prefixed to a scrub rule causes matching packets to remain unscrubbed,
960much in the same way as
961.Ar drop quick
962works in the packet filter (see below).
963This mechanism should be used when it is necessary to exclude specific packets
964from broader scrub rules.
965.Pp
966.Ar scrub
967rules in the
968.Ar scrub
969ruleset are evaluated for every packet before stateful filtering.
970This means excessive usage of them will cause performance penalty.
971.Ar scrub reassemble tcp
972rules must not have the direction (in/out) specified.
973.Sh QUEUEING with ALTQ
974The ALTQ system is currently not available in the GENERIC kernel nor as
975loadable modules.
976In order to use the herein after called queueing options one has to use a
977custom built kernel.
978Please refer to
979.Xr altq 4
980to learn about the related kernel options.
981.Pp
982Packets can be assigned to queues for the purpose of bandwidth
983control.
984At least two declarations are required to configure queues, and later
985any packet filtering rule can reference the defined queues by name.
986During the filtering component of
987.Nm pf.conf ,
988the last referenced
989.Ar queue
990name is where any packets from
991.Ar pass
992rules will be queued, while for
993.Ar block
994rules it specifies where any resulting ICMP or TCP RST
995packets should be queued.
996The
997.Ar scheduler
998defines the algorithm used to decide which packets get delayed, dropped, or
999sent out immediately.
1000There are three
1001.Ar schedulers
1002currently supported.
1003.Bl -tag -width xxxx
1004.It Ar cbq
1005Class Based Queueing.
1006.Ar Queues
1007attached to an interface build a tree, thus each
1008.Ar queue
1009can have further child
1010.Ar queues .
1011Each queue can have a
1012.Ar priority
1013and a
1014.Ar bandwidth
1015assigned.
1016.Ar Priority
1017mainly controls the time packets take to get sent out, while
1018.Ar bandwidth
1019has primarily effects on throughput.
1020.Ar cbq
1021achieves both partitioning and sharing of link bandwidth
1022by hierarchically structured classes.
1023Each class has its own
1024.Ar queue
1025and is assigned its share of
1026.Ar bandwidth .
1027A child class can borrow bandwidth from its parent class
1028as long as excess bandwidth is available
1029(see the option
1030.Ar borrow ,
1031below).
1032.It Ar priq
1033Priority Queueing.
1034.Ar Queues
1035are flat attached to the interface, thus,
1036.Ar queues
1037cannot have further child
1038.Ar queues .
1039Each
1040.Ar queue
1041has a unique
1042.Ar priority
1043assigned, ranging from 0 to 15.
1044Packets in the
1045.Ar queue
1046with the highest
1047.Ar priority
1048are processed first.
1049.It Ar hfsc
1050Hierarchical Fair Service Curve.
1051.Ar Queues
1052attached to an interface build a tree, thus each
1053.Ar queue
1054can have further child
1055.Ar queues .
1056Each queue can have a
1057.Ar priority
1058and a
1059.Ar bandwidth
1060assigned.
1061.Ar Priority
1062mainly controls the time packets take to get sent out, while
1063.Ar bandwidth
1064primarily affects throughput.
1065.Ar hfsc
1066supports both link-sharing and guaranteed real-time services.
1067It employs a service curve based QoS model,
1068and its unique feature is an ability to decouple
1069.Ar delay
1070and
1071.Ar bandwidth
1072allocation.
1073.El
1074.Pp
1075The interfaces on which queueing should be activated are declared using
1076the
1077.Ar altq on
1078declaration.
1079.Ar altq on
1080has the following keywords:
1081.Bl -tag -width xxxx
1082.It Aq Ar interface
1083Queueing is enabled on the named interface.
1084.It Aq Ar scheduler
1085Specifies which queueing scheduler to use.
1086Currently supported values
1087are
1088.Ar cbq
1089for Class Based Queueing,
1090.Ar priq
1091for Priority Queueing and
1092.Ar hfsc
1093for the Hierarchical Fair Service Curve scheduler.
1094.It Ar bandwidth Aq Ar bw
1095The maximum bitrate for all queues on an
1096interface may be specified using the
1097.Ar bandwidth
1098keyword.
1099The value can be specified as an absolute value or as a
1100percentage of the interface bandwidth.
1101When using an absolute value, the suffixes
1102.Ar b ,
1103.Ar Kb ,
1104.Ar Mb ,
1105and
1106.Ar Gb
1107are used to represent bits, kilobits, megabits, and
1108gigabits per second, respectively.
1109The value must not exceed the interface bandwidth.
1110If
1111.Ar bandwidth
1112is not specified, the interface bandwidth is used
1113(but take note that some interfaces do not know their bandwidth,
1114or can adapt their bandwidth rates).
1115.It Ar qlimit Aq Ar limit
1116The maximum number of packets held in the queue.
1117The default is 50.
1118.It Ar tbrsize Aq Ar size
1119Adjusts the size, in bytes, of the token bucket regulator.
1120If not specified, heuristics based on the
1121interface bandwidth are used to determine the size.
1122.It Ar queue Aq Ar list
1123Defines a list of subqueues to create on an interface.
1124.El
1125.Pp
1126In the following example, the interface dc0
1127should queue up to 5Mbps in four second-level queues using
1128Class Based Queueing.
1129Those four queues will be shown in a later example.
1130.Bd -literal -offset indent
1131altq on dc0 cbq bandwidth 5Mb queue { std, http, mail, ssh }
1132.Ed
1133.Pp
1134Once interfaces are activated for queueing using the
1135.Ar altq
1136directive, a sequence of
1137.Ar queue
1138directives may be defined.
1139The name associated with a
1140.Ar queue
1141must match a queue defined in the
1142.Ar altq
1143directive (e.g. mail), or, except for the
1144.Ar priq
1145.Ar scheduler ,
1146in a parent
1147.Ar queue
1148declaration.
1149The following keywords can be used:
1150.Bl -tag -width xxxx
1151.It Ar on Aq Ar interface
1152Specifies the interface the queue operates on.
1153If not given, it operates on all matching interfaces.
1154.It Ar bandwidth Aq Ar bw
1155Specifies the maximum bitrate to be processed by the queue.
1156This value must not exceed the value of the parent
1157.Ar queue
1158and can be specified as an absolute value or a percentage of the parent
1159queue's bandwidth.
1160If not specified, defaults to 100% of the parent queue's bandwidth.
1161The
1162.Ar priq
1163scheduler does not support bandwidth specification.
1164.It Ar priority Aq Ar level
1165Between queues a priority level can be set.
1166For
1167.Ar cbq
1168and
1169.Ar hfsc ,
1170the range is 0 to 7 and for
1171.Ar priq ,
1172the range is 0 to 15.
1173The default for all is 1.
1174.Ar Priq
1175queues with a higher priority are always served first.
1176.Ar Cbq
1177and
1178.Ar Hfsc
1179queues with a higher priority are preferred in the case of overload.
1180.It Ar qlimit Aq Ar limit
1181The maximum number of packets held in the queue.
1182The default is 50.
1183.El
1184.Pp
1185The
1186.Ar scheduler
1187can get additional parameters with
1188.Xo Aq Ar scheduler
1189.Pf ( Aq Ar parameters ) .
1190.Xc
1191Parameters are as follows:
1192.Bl -tag -width Fl
1193.It Ar default
1194Packets not matched by another queue are assigned to this one.
1195Exactly one default queue is required.
1196.It Ar red
1197Enable RED (Random Early Detection) on this queue.
1198RED drops packets with a probability proportional to the average
1199queue length.
1200.It Ar rio
1201Enables RIO on this queue.
1202RIO is RED with IN/OUT, thus running
1203RED two times more than RIO would achieve the same effect.
1204RIO is currently not supported in the GENERIC kernel.
1205.It Ar ecn
1206Enables ECN (Explicit Congestion Notification) on this queue.
1207ECN implies RED.
1208.El
1209.Pp
1210The
1211.Ar cbq
1212.Ar scheduler
1213supports an additional option:
1214.Bl -tag -width Fl
1215.It Ar borrow
1216The queue can borrow bandwidth from the parent.
1217.El
1218.Pp
1219The
1220.Ar hfsc
1221.Ar scheduler
1222supports some additional options:
1223.Bl -tag -width Fl
1224.It Ar realtime Aq Ar sc
1225The minimum required bandwidth for the queue.
1226.It Ar upperlimit Aq Ar sc
1227The maximum allowed bandwidth for the queue.
1228.It Ar linkshare Aq Ar sc
1229The bandwidth share of a backlogged queue.
1230.El
1231.Pp
1232.Aq Ar sc
1233is an acronym for
1234.Ar service curve .
1235.Pp
1236The format for service curve specifications is
1237.Ar ( m1 , d , m2 ) .
1238.Ar m2
1239controls the bandwidth assigned to the queue.
1240.Ar m1
1241and
1242.Ar d
1243are optional and can be used to control the initial bandwidth assignment.
1244For the first
1245.Ar d
1246milliseconds the queue gets the bandwidth given as
1247.Ar m1 ,
1248afterwards the value given in
1249.Ar m2 .
1250.Pp
1251Furthermore, with
1252.Ar cbq
1253and
1254.Ar hfsc ,
1255child queues can be specified as in an
1256.Ar altq
1257declaration, thus building a tree of queues using a part of
1258their parent's bandwidth.
1259.Pp
1260Packets can be assigned to queues based on filter rules by using the
1261.Ar queue
1262keyword.
1263Normally only one
1264.Ar queue
1265is specified; when a second one is specified it will instead be used for
1266packets which have a
1267.Em TOS
1268of
1269.Em lowdelay
1270and for TCP ACKs with no data payload.
1271.Pp
1272To continue the previous example, the examples below would specify the
1273four referenced
1274queues, plus a few child queues.
1275Interactive
1276.Xr ssh 1
1277sessions get priority over bulk transfers like
1278.Xr scp 1
1279and
1280.Xr sftp 1 .
1281The queues may then be referenced by filtering rules (see
1282.Sx PACKET FILTERING
1283below).
1284.Bd -literal
1285queue std bandwidth 10% cbq(default)
1286queue http bandwidth 60% priority 2 cbq(borrow red) \e
1287      { employees, developers }
1288queue  developers bandwidth 75% cbq(borrow)
1289queue  employees bandwidth 15%
1290queue mail bandwidth 10% priority 0 cbq(borrow ecn)
1291queue ssh bandwidth 20% cbq(borrow) { ssh_interactive, ssh_bulk }
1292queue  ssh_interactive bandwidth 50% priority 7 cbq(borrow)
1293queue  ssh_bulk bandwidth 50% priority 0 cbq(borrow)
1294
1295block return out on dc0 inet all queue std
1296pass out on dc0 inet proto tcp from $developerhosts to any port 80 \e
1297      queue developers
1298pass out on dc0 inet proto tcp from $employeehosts to any port 80 \e
1299      queue employees
1300pass out on dc0 inet proto tcp from any to any port 22 \e
1301      queue(ssh_bulk, ssh_interactive)
1302pass out on dc0 inet proto tcp from any to any port 25 \e
1303      queue mail
1304.Ed
1305.Sh QUEUEING with dummynet
1306Queueing can also be done with
1307.Xr dummynet 4 .
1308Queues and pipes can be created with
1309.Xr dnctl 8 .
1310.Pp
1311Packets can be assigned to queues and pipes using
1312.Ar dnqueue
1313and
1314.Ar dnpipe
1315respectively.
1316.Pp
1317Both
1318.Ar dnqueue
1319and
1320.Ar dnpipe
1321take either a single pipe or queue number or two numbers as arguments.
1322The first pipe or queue number will be used to shape the traffic in the rule
1323direction, the second will be used to shape the traffic in the reverse
1324direction.
1325If the rule does not specify a direction the first packet to create state will
1326be shaped according to the first number, and the response traffic according to
1327the second.
1328.Pp
1329If the
1330.Xr dummynet 4
1331module is not loaded any traffic sent into a queue or pipe will be dropped.
1332.Sh TRANSLATION
1333Translation rules modify either the source or destination address of the
1334packets associated with a stateful connection.
1335A stateful connection is automatically created to track packets matching
1336such a rule as long as they are not blocked by the filtering section of
1337.Nm pf.conf .
1338The translation engine modifies the specified address and/or port in the
1339packet, recalculates IP, TCP and UDP checksums as necessary, and passes
1340it to the packet filter for evaluation.
1341.Pp
1342Since translation occurs before filtering the filter
1343engine will see packets as they look after any
1344addresses and ports have been translated.
1345Filter rules will therefore have to filter based on the translated
1346address and port number.
1347Packets that match a translation rule are only automatically passed if
1348the
1349.Ar pass
1350modifier is given, otherwise they are
1351still subject to
1352.Ar block
1353and
1354.Ar pass
1355rules.
1356.Pp
1357The state entry created permits
1358.Xr pf 4
1359to keep track of the original address for traffic associated with that state
1360and correctly direct return traffic for that connection.
1361.Pp
1362Various types of translation are possible with pf:
1363.Bl -tag -width xxxx
1364.It Ar binat
1365A
1366.Ar binat
1367rule specifies a bidirectional mapping between an external IP netblock
1368and an internal IP netblock.
1369.It Ar nat
1370A
1371.Ar nat
1372rule specifies that IP addresses are to be changed as the packet
1373traverses the given interface.
1374This technique allows one or more IP addresses
1375on the translating host to support network traffic for a larger range of
1376machines on an "inside" network.
1377Although in theory any IP address can be used on the inside, it is strongly
1378recommended that one of the address ranges defined by RFC 1918 be used.
1379These netblocks are:
1380.Bd -literal
138110.0.0.0 - 10.255.255.255 (all of net 10, i.e., 10/8)
1382172.16.0.0 - 172.31.255.255 (i.e., 172.16/12)
1383192.168.0.0 - 192.168.255.255 (i.e., 192.168/16)
1384.Ed
1385.It Pa rdr
1386The packet is redirected to another destination and possibly a
1387different port.
1388.Ar rdr
1389rules can optionally specify port ranges instead of single ports.
1390rdr ... port 2000:2999 -\*(Gt ... port 4000
1391redirects ports 2000 to 2999 (inclusive) to port 4000.
1392rdr ... port 2000:2999 -\*(Gt ... port 4000:*
1393redirects port 2000 to 4000, 2001 to 4001, ..., 2999 to 4999.
1394.El
1395.Pp
1396In addition to modifying the address, some translation rules may modify
1397source or destination ports for
1398.Xr tcp 4
1399or
1400.Xr udp 4
1401connections; implicitly in the case of
1402.Ar nat
1403rules and both implicitly and explicitly in the case of
1404.Ar rdr
1405rules.
1406A
1407.Ar rdr
1408rule may cause the source port to be modified if doing so avoids a conflict
1409with an existing connection.
1410A random source port in the range 50001-65535 is chosen in this case; to
1411avoid excessive CPU consumption, the number of searches for a free port is
1412limited by the
1413.Va net.pf.rdr_srcport_rewrite_tries
1414sysctl.
1415Port numbers are never translated with a
1416.Ar binat
1417rule.
1418.Pp
1419Evaluation order of the translation rules is dependent on the type
1420of the translation rules and of the direction of a packet.
1421.Ar binat
1422rules are always evaluated first.
1423Then either the
1424.Ar rdr
1425rules are evaluated on an inbound packet or the
1426.Ar nat
1427rules on an outbound packet.
1428Rules of the same type are evaluated in the same order in which they
1429appear in the ruleset.
1430The first matching rule decides what action is taken.
1431.Pp
1432The
1433.Ar no
1434option prefixed to a translation rule causes packets to remain untranslated,
1435much in the same way as
1436.Ar drop quick
1437works in the packet filter (see below).
1438If no rule matches the packet it is passed to the filter engine unmodified.
1439.Pp
1440Translation rules apply only to packets that pass through
1441the specified interface, and if no interface is specified,
1442translation is applied to packets on all interfaces.
1443For instance, redirecting port 80 on an external interface to an internal
1444web server will only work for connections originating from the outside.
1445Connections to the address of the external interface from local hosts will
1446not be redirected, since such packets do not actually pass through the
1447external interface.
1448Redirections cannot reflect packets back through the interface they arrive
1449on, they can only be redirected to hosts connected to different interfaces
1450or to the firewall itself.
1451.Pp
1452Note that redirecting external incoming connections to the loopback
1453address, as in
1454.Bd -literal -offset indent
1455rdr on ne3 inet proto tcp to port smtp -\*(Gt 127.0.0.1 port spamd
1456.Ed
1457.Pp
1458will effectively allow an external host to connect to daemons
1459bound solely to the loopback address, circumventing the traditional
1460blocking of such connections on a real interface.
1461Unless this effect is desired, any of the local non-loopback addresses
1462should be used as redirection target instead, which allows external
1463connections only to daemons bound to this address or not bound to
1464any address.
1465.Pp
1466See
1467.Sx TRANSLATION EXAMPLES
1468below.
1469.Sh PACKET FILTERING
1470.Xr pf 4
1471has the ability to
1472.Ar block
1473,
1474.Ar pass
1475and
1476.Ar match
1477packets based on attributes of their layer 3 (see
1478.Xr ip 4
1479and
1480.Xr ip6 4 )
1481and layer 4 (see
1482.Xr icmp 4 ,
1483.Xr icmp6 4 ,
1484.Xr tcp 4 ,
1485.Xr sctp 4 ,
1486.Xr udp 4 )
1487headers.
1488In addition, packets may also be
1489assigned to queues for the purpose of bandwidth control.
1490.Pp
1491For each packet processed by the packet filter, the filter rules are
1492evaluated in sequential order, from first to last.
1493For
1494.Ar block
1495and
1496.Ar pass
1497, the last matching rule decides what action is taken.
1498For
1499.Ar match
1500, rules are evaluated every time they match; the pass/block state of a packet
1501remains unchanged.
1502If no rule matches the packet, the default action is to pass
1503the packet.
1504.Pp
1505The following actions can be used in the filter:
1506.Bl -tag -width xxxx
1507.It Ar block
1508The packet is blocked.
1509There are a number of ways in which a
1510.Ar block
1511rule can behave when blocking a packet.
1512The default behaviour is to
1513.Ar drop
1514packets silently, however this can be overridden or made
1515explicit either globally, by setting the
1516.Ar block-policy
1517option, or on a per-rule basis with one of the following options:
1518.Pp
1519.Bl -tag -width xxxx -compact
1520.It Ar drop
1521The packet is silently dropped.
1522.It Ar return-rst
1523This applies only to
1524.Xr tcp 4
1525packets, and issues a TCP RST which closes the
1526connection.
1527.It Ar return-icmp
1528.It Ar return-icmp6
1529This causes ICMP messages to be returned for packets which match the rule.
1530By default this is an ICMP UNREACHABLE message, however this
1531can be overridden by specifying a message as a code or number.
1532.It Ar return
1533This causes a TCP RST to be returned for
1534.Xr tcp 4
1535packets, an SCTP ABORT for SCTP
1536and an ICMP UNREACHABLE for UDP and other packets.
1537.El
1538.Pp
1539Options returning ICMP packets currently have no effect if
1540.Xr pf 4
1541operates on a
1542.Xr if_bridge 4 ,
1543as the code to support this feature has not yet been implemented.
1544.Pp
1545The simplest mechanism to block everything by default and only pass
1546packets that match explicit rules is specify a first filter rule of:
1547.Bd -literal -offset indent
1548block all
1549.Ed
1550.It Ar match
1551The packet is matched.
1552This mechanism is used to provide fine grained filtering without altering the
1553block/pass state of a packet.
1554.Ar match
1555rules differ from
1556.Ar block
1557and
1558.Ar pass
1559rules in that parameters are set for every rule a packet matches, not only
1560on the last matching rule.
1561For the following parameters, this means that the parameter effectively becomes
1562"sticky" until explicitly overridden:
1563.Ar queue ,
1564.Ar dnpipe ,
1565.Ar dnqueue ,
1566.Ar rtable ,
1567.Ar scrub
1568.
1569.It Ar pass
1570The packet is passed;
1571state is created unless the
1572.Ar no state
1573option is specified.
1574.El
1575.Pp
1576By default
1577.Xr pf 4
1578filters packets statefully; the first time a packet matches a
1579.Ar pass
1580rule, a state entry is created; for subsequent packets the filter checks
1581whether the packet matches any state.
1582If it does, the packet is passed without evaluation of any rules.
1583After the connection is closed or times out, the state entry is automatically
1584removed.
1585.Pp
1586This has several advantages.
1587For TCP connections, comparing a packet to a state involves checking
1588its sequence numbers, as well as TCP timestamps if a
1589.Ar scrub reassemble tcp
1590rule applies to the connection.
1591If these values are outside the narrow windows of expected
1592values, the packet is dropped.
1593This prevents spoofing attacks, such as when an attacker sends packets with
1594a fake source address/port but does not know the connection's sequence
1595numbers.
1596Similarly,
1597.Xr pf 4
1598knows how to match ICMP replies to states.
1599For example,
1600.Bd -literal -offset indent
1601pass out inet proto icmp all icmp-type echoreq
1602.Ed
1603.Pp
1604allows echo requests (such as those created by
1605.Xr ping 8 )
1606out statefully, and matches incoming echo replies correctly to states.
1607.Pp
1608Also, looking up states is usually faster than evaluating rules.
1609If there are 50 rules, all of them are evaluated sequentially in O(n).
1610Even with 50000 states, only 16 comparisons are needed to match a
1611state, since states are stored in a binary search tree that allows
1612searches in O(log2 n).
1613.Pp
1614Furthermore, correct handling of ICMP error messages is critical to
1615many protocols, particularly TCP.
1616.Xr pf 4
1617matches ICMP error messages to the correct connection, checks them against
1618connection parameters, and passes them if appropriate.
1619For example if an ICMP source quench message referring to a stateful TCP
1620connection arrives, it will be matched to the state and get passed.
1621.Pp
1622Finally, state tracking is required for
1623.Ar nat , binat No and Ar rdr
1624rules, in order to track address and port translations and reverse the
1625translation on returning packets.
1626.Pp
1627.Xr pf 4
1628will also create state for other protocols which are effectively stateless by
1629nature.
1630UDP packets are matched to states using only host addresses and ports,
1631and other protocols are matched to states using only the host addresses.
1632.Pp
1633If stateless filtering of individual packets is desired,
1634the
1635.Ar no state
1636keyword can be used to specify that state will not be created
1637if this is the last matching rule.
1638A number of parameters can also be set to affect how
1639.Xr pf 4
1640handles state tracking.
1641See
1642.Sx STATEFUL TRACKING OPTIONS
1643below for further details.
1644.Sh PARAMETERS
1645The rule parameters specify the packets to which a rule applies.
1646A packet always comes in on, or goes out through, one interface.
1647Most parameters are optional.
1648If a parameter is specified, the rule only applies to packets with
1649matching attributes.
1650Certain parameters can be expressed as lists, in which case
1651.Xr pfctl 8
1652generates all needed rule combinations.
1653.Bl -tag -width xxxx
1654.It Ar in No or Ar out
1655This rule applies to incoming or outgoing packets.
1656If neither
1657.Ar in
1658nor
1659.Ar out
1660are specified, the rule will match packets in both directions.
1661.It Ar log
1662In addition to the action specified, a log message is generated.
1663Only the packet that establishes the state is logged,
1664unless the
1665.Ar no state
1666option is specified.
1667The logged packets are sent to a
1668.Xr pflog 4
1669interface, by default
1670.Ar pflog0 .
1671This interface is monitored by the
1672.Xr pflogd 8
1673logging daemon, which dumps the logged packets to the file
1674.Pa /var/log/pflog
1675in
1676.Xr pcap 3
1677binary format.
1678.It Ar log (all)
1679Used to force logging of all packets for a connection.
1680This is not necessary when
1681.Ar no state
1682is explicitly specified.
1683As with
1684.Ar log ,
1685packets are logged to
1686.Xr pflog 4 .
1687.It Ar log (user)
1688Logs the
1689.Ux
1690user ID of the user that owns the socket and the PID of the process that
1691has the socket open where the packet is sourced from or destined to
1692(depending on which socket is local).
1693This is in addition to the normal information logged.
1694.Pp
1695Only the first packet
1696logged via
1697.Ar log (all, user)
1698will have the user credentials logged when using stateful matching.
1699.It Ar log (to Aq Ar interface )
1700Send logs to the specified
1701.Xr pflog 4
1702interface instead of
1703.Ar pflog0 .
1704.It Ar quick
1705If a packet matches a rule which has the
1706.Ar quick
1707option set, this rule
1708is considered the last matching rule, and evaluation of subsequent rules
1709is skipped.
1710.It Ar on Aq Ar interface
1711This rule applies only to packets coming in on, or going out through, this
1712particular interface or interface group.
1713For more information on interface groups,
1714see the
1715.Ic group
1716keyword in
1717.Xr ifconfig 8 .
1718.It Aq Ar af
1719This rule applies only to packets of this address family.
1720Supported values are
1721.Ar inet
1722and
1723.Ar inet6 .
1724.It Ar proto Aq Ar protocol
1725This rule applies only to packets of this protocol.
1726Common protocols are
1727.Xr icmp 4 ,
1728.Xr icmp6 4 ,
1729.Xr tcp 4 ,
1730.Xr sctp 4 ,
1731and
1732.Xr udp 4 .
1733For a list of all the protocol name to number mappings used by
1734.Xr pfctl 8 ,
1735see the file
1736.Pa /etc/protocols .
1737.It Xo
1738.Ar from Aq Ar source
1739.Ar port Aq Ar source
1740.Ar os Aq Ar source
1741.Ar to Aq Ar dest
1742.Ar port Aq Ar dest
1743.Xc
1744This rule applies only to packets with the specified source and destination
1745addresses and ports.
1746.Pp
1747Addresses can be specified in CIDR notation (matching netblocks), as
1748symbolic host names, interface names or interface group names, or as any
1749of the following keywords:
1750.Pp
1751.Bl -tag -width xxxxxxxxxxxxxx -compact
1752.It Ar any
1753Any address.
1754.It Ar no-route
1755Any address which is not currently routable.
1756.It Ar urpf-failed
1757Any source address that fails a unicast reverse path forwarding (URPF)
1758check, i.e. packets coming in on an interface other than that which holds
1759the route back to the packet's source address.
1760.It Aq Ar table
1761Any address that matches the given table.
1762.El
1763.Pp
1764Ranges of addresses are specified by using the
1765.Sq -
1766operator.
1767For instance:
1768.Dq 10.1.1.10 - 10.1.1.12
1769means all addresses from 10.1.1.10 to 10.1.1.12,
1770hence addresses 10.1.1.10, 10.1.1.11, and 10.1.1.12.
1771.Pp
1772Interface names and interface group names can have modifiers appended:
1773.Pp
1774.Bl -tag -width xxxxxxxxxxxx -compact
1775.It Ar :network
1776Translates to the network(s) attached to the interface.
1777.It Ar :broadcast
1778Translates to the interface's broadcast address(es).
1779.It Ar :peer
1780Translates to the point-to-point interface's peer address(es).
1781.It Ar :0
1782Do not include interface aliases.
1783.El
1784.Pp
1785Host names may also have the
1786.Ar :0
1787option appended to restrict the name resolution to the first of each
1788v4 and non-link-local v6 address found.
1789.Pp
1790Host name resolution and interface to address translation are done at
1791ruleset load-time.
1792When the address of an interface (or host name) changes (under DHCP or PPP,
1793for instance), the ruleset must be reloaded for the change to be reflected
1794in the kernel.
1795Surrounding the interface name (and optional modifiers) in parentheses
1796changes this behaviour.
1797When the interface name is surrounded by parentheses, the rule is
1798automatically updated whenever the interface changes its address.
1799The ruleset does not need to be reloaded.
1800This is especially useful with
1801.Ar nat .
1802.Pp
1803Ports can be specified either by number or by name.
1804For example, port 80 can be specified as
1805.Em www .
1806For a list of all port name to number mappings used by
1807.Xr pfctl 8 ,
1808see the file
1809.Pa /etc/services .
1810.Pp
1811Ports and ranges of ports are specified by using these operators:
1812.Bd -literal -offset indent
1813=	(equal)
1814!=	(unequal)
1815\*(Lt	(less than)
1816\*(Le	(less than or equal)
1817\*(Gt	(greater than)
1818\*(Ge	(greater than or equal)
1819:	(range including boundaries)
1820\*(Gt\*(Lt	(range excluding boundaries)
1821\*(Lt\*(Gt	(except range)
1822.Ed
1823.Pp
1824.Sq \*(Gt\*(Lt ,
1825.Sq \*(Lt\*(Gt
1826and
1827.Sq \&:
1828are binary operators (they take two arguments).
1829For instance:
1830.Bl -tag -width Fl
1831.It Ar port 2000:2004
1832means
1833.Sq all ports \*(Ge 2000 and \*(Le 2004 ,
1834hence ports 2000, 2001, 2002, 2003 and 2004.
1835.It Ar port 2000 \*(Gt\*(Lt 2004
1836means
1837.Sq all ports \*(Gt 2000 and \*(Lt 2004 ,
1838hence ports 2001, 2002 and 2003.
1839.It Ar port 2000 \*(Lt\*(Gt 2004
1840means
1841.Sq all ports \*(Lt 2000 or \*(Gt 2004 ,
1842hence ports 1-1999 and 2005-65535.
1843.El
1844.Pp
1845The operating system of the source host can be specified in the case of TCP
1846rules with the
1847.Ar OS
1848modifier.
1849See the
1850.Sx OPERATING SYSTEM FINGERPRINTING
1851section for more information.
1852.Pp
1853The host, port and OS specifications are optional, as in the following examples:
1854.Bd -literal -offset indent
1855pass in all
1856pass in from any to any
1857pass in proto tcp from any port \*(Le 1024 to any
1858pass in proto tcp from any to any port 25
1859pass in proto tcp from 10.0.0.0/8 port \*(Gt 1024 \e
1860      to ! 10.1.2.3 port != ssh
1861pass in proto tcp from any os "OpenBSD"
1862.Ed
1863.It Ar all
1864This is equivalent to "from any to any".
1865.It Ar group Aq Ar group
1866Similar to
1867.Ar user ,
1868this rule only applies to packets of sockets owned by the specified group.
1869.It Ar user Aq Ar user
1870This rule only applies to packets of sockets owned by the specified user.
1871For outgoing connections initiated from the firewall, this is the user
1872that opened the connection.
1873For incoming connections to the firewall itself, this is the user that
1874listens on the destination port.
1875For forwarded connections, where the firewall is not a connection endpoint,
1876the user and group are
1877.Em unknown .
1878.Pp
1879All packets, both outgoing and incoming, of one connection are associated
1880with the same user and group.
1881Only TCP and UDP packets can be associated with users; for other protocols
1882these parameters are ignored.
1883.Pp
1884User and group refer to the effective (as opposed to the real) IDs, in
1885case the socket is created by a setuid/setgid process.
1886User and group IDs are stored when a socket is created;
1887when a process creates a listening socket as root (for instance, by
1888binding to a privileged port) and subsequently changes to another
1889user ID (to drop privileges), the credentials will remain root.
1890.Pp
1891User and group IDs can be specified as either numbers or names.
1892The syntax is similar to the one for ports.
1893The value
1894.Em unknown
1895matches packets of forwarded connections.
1896.Em unknown
1897can only be used with the operators
1898.Cm =
1899and
1900.Cm != .
1901Other constructs like
1902.Cm user \*(Ge unknown
1903are invalid.
1904Forwarded packets with unknown user and group ID match only rules
1905that explicitly compare against
1906.Em unknown
1907with the operators
1908.Cm =
1909or
1910.Cm != .
1911For instance
1912.Cm user \*(Ge 0
1913does not match forwarded packets.
1914The following example allows only selected users to open outgoing
1915connections:
1916.Bd -literal -offset indent
1917block out proto { tcp, udp } all
1918pass  out proto { tcp, udp } all user { \*(Lt 1000, dhartmei }
1919.Ed
1920.It Xo Ar flags Aq Ar a
1921.Pf / Ns Aq Ar b
1922.No \*(Ba / Ns Aq Ar b
1923.No \*(Ba any
1924.Xc
1925This rule only applies to TCP packets that have the flags
1926.Aq Ar a
1927set out of set
1928.Aq Ar b .
1929Flags not specified in
1930.Aq Ar b
1931are ignored.
1932For stateful connections, the default is
1933.Ar flags S/SA .
1934To indicate that flags should not be checked at all, specify
1935.Ar flags any .
1936The flags are: (F)IN, (S)YN, (R)ST, (P)USH, (A)CK, (U)RG, (E)CE, and C(W)R.
1937.Bl -tag -width Fl
1938.It Ar flags S/S
1939Flag SYN is set.
1940The other flags are ignored.
1941.It Ar flags S/SA
1942This is the default setting for stateful connections.
1943Out of SYN and ACK, exactly SYN may be set.
1944SYN, SYN+PSH and SYN+RST match, but SYN+ACK, ACK and ACK+RST do not.
1945This is more restrictive than the previous example.
1946.It Ar flags /SFRA
1947If the first set is not specified, it defaults to none.
1948All of SYN, FIN, RST and ACK must be unset.
1949.El
1950.Pp
1951Because
1952.Ar flags S/SA
1953is applied by default (unless
1954.Ar no state
1955is specified), only the initial SYN packet of a TCP handshake will create
1956a state for a TCP connection.
1957It is possible to be less restrictive, and allow state creation from
1958intermediate
1959.Pq non-SYN
1960packets, by specifying
1961.Ar flags any .
1962This will cause
1963.Xr pf 4
1964to synchronize to existing connections, for instance
1965if one flushes the state table.
1966However, states created from such intermediate packets may be missing
1967connection details such as the TCP window scaling factor.
1968States which modify the packet flow, such as those affected by
1969.Ar nat , binat No or Ar rdr
1970rules,
1971.Ar modulate No or Ar synproxy state
1972options, or scrubbed with
1973.Ar reassemble tcp
1974will also not be recoverable from intermediate packets.
1975Such connections will stall and time out.
1976.It Xo Ar icmp-type Aq Ar type
1977.Ar code Aq Ar code
1978.Xc
1979.It Xo Ar icmp6-type Aq Ar type
1980.Ar code Aq Ar code
1981.Xc
1982This rule only applies to ICMP or ICMPv6 packets with the specified type
1983and code.
1984Text names for ICMP types and codes are listed in
1985.Xr icmp 4
1986and
1987.Xr icmp6 4 .
1988This parameter is only valid for rules that cover protocols ICMP or
1989ICMP6.
1990The protocol and the ICMP type indicator
1991.Po
1992.Ar icmp-type
1993or
1994.Ar icmp6-type
1995.Pc
1996must match.
1997.It Xo Ar tos Aq Ar string
1998.No \*(Ba Aq Ar number
1999.Xc
2000This rule applies to packets with the specified
2001.Em TOS
2002bits set.
2003.Em TOS
2004may be
2005given as one of
2006.Ar critical ,
2007.Ar inetcontrol ,
2008.Ar lowdelay ,
2009.Ar netcontrol ,
2010.Ar throughput ,
2011.Ar reliability ,
2012or one of the DiffServ Code Points:
2013.Ar ef ,
2014.Ar va ,
2015.Ar af11 No ... Ar af43 ,
2016.Ar cs0 No ... Ar cs7 ;
2017or as either hex or decimal.
2018.Pp
2019For example, the following rules are identical:
2020.Bd -literal -offset indent
2021pass all tos lowdelay
2022pass all tos 0x10
2023pass all tos 16
2024.Ed
2025.It Ar allow-opts
2026By default, IPv4 packets with IP options or IPv6 packets with routing
2027extension headers are blocked.
2028When
2029.Ar allow-opts
2030is specified for a
2031.Ar pass
2032rule, packets that pass the filter based on that rule (last matching)
2033do so even if they contain IP options or routing extension headers.
2034For packets that match state, the rule that initially created the
2035state is used.
2036The implicit
2037.Ar pass
2038rule that is used when a packet does not match any rules does not
2039allow IP options.
2040.It Ar label Aq Ar string
2041Adds a label (name) to the rule, which can be used to identify the rule.
2042For instance,
2043pfctl -s labels
2044shows per-rule statistics for rules that have labels.
2045.Pp
2046The following macros can be used in labels:
2047.Pp
2048.Bl -tag -width $srcaddr -compact -offset indent
2049.It Ar $if
2050The interface.
2051.It Ar $srcaddr
2052The source IP address.
2053.It Ar $dstaddr
2054The destination IP address.
2055.It Ar $srcport
2056The source port specification.
2057.It Ar $dstport
2058The destination port specification.
2059.It Ar $proto
2060The protocol name.
2061.It Ar $nr
2062The rule number.
2063.El
2064.Pp
2065For example:
2066.Bd -literal -offset indent
2067ips = \&"{ 1.2.3.4, 1.2.3.5 }\&"
2068pass in proto tcp from any to $ips \e
2069      port \*(Gt 1023 label \&"$dstaddr:$dstport\&"
2070.Ed
2071.Pp
2072expands to
2073.Bd -literal -offset indent
2074pass in inet proto tcp from any to 1.2.3.4 \e
2075      port \*(Gt 1023 label \&"1.2.3.4:\*(Gt1023\&"
2076pass in inet proto tcp from any to 1.2.3.5 \e
2077      port \*(Gt 1023 label \&"1.2.3.5:\*(Gt1023\&"
2078.Ed
2079.Pp
2080The macro expansion for the
2081.Ar label
2082directive occurs only at configuration file parse time, not during runtime.
2083.It Ar ridentifier Aq Ar number
2084Add an identifier (number) to the rule, which can be used to correlate the rule
2085to pflog entries, even after ruleset updates.
2086.It Xo Ar queue Aq Ar queue
2087.No \*(Ba ( Aq Ar queue ,
2088.Aq Ar queue )
2089.Xc
2090Packets matching this rule will be assigned to the specified queue.
2091If two queues are given, packets which have a
2092.Em TOS
2093of
2094.Em lowdelay
2095and TCP ACKs with no data payload will be assigned to the second one.
2096See
2097.Sx QUEUEING
2098for setup details.
2099.Pp
2100For example:
2101.Bd -literal -offset indent
2102pass in proto tcp to port 25 queue mail
2103pass in proto tcp to port 22 queue(ssh_bulk, ssh_prio)
2104.Ed
2105.It Cm set prio Ar priority | Pq Ar priority , priority
2106Packets matching this rule will be assigned a specific queueing priority.
2107Priorities are assigned as integers 0 through 7.
2108If the packet is transmitted on a
2109.Xr vlan 4
2110interface, the queueing priority will be written as the priority
2111code point in the 802.1Q VLAN header.
2112If two priorities are given, packets which have a TOS of
2113.Cm lowdelay
2114and TCP ACKs with no data payload will be assigned to the second one.
2115.Pp
2116For example:
2117.Bd -literal -offset indent
2118pass in proto tcp to port 25 set prio 2
2119pass in proto tcp to port 22 set prio (2, 5)
2120.Ed
2121.It Ar tag Aq Ar string
2122Packets matching this rule will be tagged with the
2123specified string.
2124The tag acts as an internal marker that can be used to
2125identify these packets later on.
2126This can be used, for example, to provide trust between
2127interfaces and to determine if packets have been
2128processed by translation rules.
2129Tags are
2130.Qq sticky ,
2131meaning that the packet will be tagged even if the rule
2132is not the last matching rule.
2133Further matching rules can replace the tag with a
2134new one but will not remove a previously applied tag.
2135A packet is only ever assigned one tag at a time.
2136Packet tagging can be done during
2137.Ar nat ,
2138.Ar rdr ,
2139.Ar binat
2140or
2141.Ar ether
2142rules in addition to filter rules.
2143Tags take the same macros as labels (see above).
2144.It Ar tagged Aq Ar string
2145Used with filter, translation or scrub rules
2146to specify that packets must already
2147be tagged with the given tag in order to match the rule.
2148Inverse tag matching can also be done
2149by specifying the
2150.Cm !\&
2151operator before the
2152.Ar tagged
2153keyword.
2154.It Ar rtable Aq Ar number
2155Used to select an alternate routing table for the routing lookup.
2156Only effective before the route lookup happened, i.e. when filtering inbound.
2157.It Xo Ar divert-to Aq Ar host
2158.Ar port Aq Ar port
2159.Xc
2160Used to
2161.Xr divert 4
2162packets to the given divert
2163.Ar port .
2164Historically
2165.Ox pf has another meaning for this, and
2166.Fx pf uses
2167this syntax to support
2168.Xr divert 4 instead. Hence,
2169.Ar host
2170has no meaning and can be set to anything like 127.0.0.1.
2171If a packet is re-injected and does not change direction then it will not be
2172re-diverted.
2173.It Ar divert-reply
2174It has no meaning in
2175.Fx pf .
2176.It Ar probability Aq Ar number
2177A probability attribute can be attached to a rule, with a value set between
21780 and 1, bounds not included.
2179In that case, the rule will be honoured using the given probability value
2180only.
2181For example, the following rule will drop 20% of incoming ICMP packets:
2182.Bd -literal -offset indent
2183block in proto icmp probability 20%
2184.Ed
2185.It Ar prio Aq Ar number
2186Only match packets which have the given queueing priority assigned.
2187.El
2188.Sh ROUTING
2189If a packet matches a rule with a route option set, the packet filter will
2190route the packet according to the type of route option.
2191When such a rule creates state, the route option is also applied to all
2192packets matching the same connection.
2193.Bl -tag -width xxxx
2194.It Ar route-to
2195The
2196.Ar route-to
2197option routes the packet to the specified interface with an optional address
2198for the next hop.
2199When a
2200.Ar route-to
2201rule creates state, only packets that pass in the same direction as the
2202filter rule specifies will be routed in this way.
2203Packets passing in the opposite direction (replies) are not affected
2204and are routed normally.
2205.It Ar reply-to
2206The
2207.Ar reply-to
2208option is similar to
2209.Ar route-to ,
2210but routes packets that pass in the opposite direction (replies) to the
2211specified interface.
2212Opposite direction is only defined in the context of a state entry, and
2213.Ar reply-to
2214is useful only in rules that create state.
2215It can be used on systems with multiple external connections to
2216route all outgoing packets of a connection through the interface
2217the incoming connection arrived through (symmetric routing enforcement).
2218.It Ar dup-to
2219The
2220.Ar dup-to
2221option creates a duplicate of the packet and routes it like
2222.Ar route-to .
2223The original packet gets routed as it normally would.
2224.El
2225.Sh POOL OPTIONS
2226For
2227.Ar nat
2228and
2229.Ar rdr
2230rules, (as well as for the
2231.Ar route-to ,
2232.Ar reply-to
2233and
2234.Ar dup-to
2235rule options) for which there is a single redirection address which has a
2236subnet mask smaller than 32 for IPv4 or 128 for IPv6 (more than one IP
2237address), a variety of different methods for assigning this address can be
2238used:
2239.Bl -tag -width xxxx
2240.It Ar bitmask
2241The
2242.Ar bitmask
2243option applies the network portion of the redirection address to the address
2244to be modified (source with
2245.Ar nat ,
2246destination with
2247.Ar rdr ) .
2248.It Ar random
2249The
2250.Ar random
2251option selects an address at random within the defined block of addresses.
2252.It Ar source-hash
2253The
2254.Ar source-hash
2255option uses a hash of the source address to determine the redirection address,
2256ensuring that the redirection address is always the same for a given source.
2257An optional key can be specified after this keyword either in hex or as a
2258string; by default
2259.Xr pfctl 8
2260randomly generates a key for source-hash every time the
2261ruleset is reloaded.
2262.It Ar round-robin
2263The
2264.Ar round-robin
2265option loops through the redirection address(es).
2266.Pp
2267When more than one redirection address is specified,
2268.Ar round-robin
2269is the only permitted pool type.
2270.It Ar static-port
2271With
2272.Ar nat
2273rules, the
2274.Ar static-port
2275option prevents
2276.Xr pf 4
2277from modifying the source port on TCP and UDP packets.
2278.It Xo Ar map-e-portset Aq Ar psid-offset
2279.No / Aq Ar psid-len
2280.No / Aq Ar psid
2281.Xc
2282With
2283.Ar nat
2284rules, the
2285.It Ar endpoint-independent
2286With
2287.Ar nat
2288rules, the
2289.Ar endpoint-independent
2290option caues
2291.Xr pf 4
2292to always map connections from a UDP source address and port to the same
2293NAT address and port.
2294This feature implements "full-cone" NAT behavior.
2295.Ar map-e-portset
2296option enables the source port translation of MAP-E (RFC 7597) Customer Edge.
2297In order to make the host act as a MAP-E Customer Edge, setting up a tunneling
2298interface and pass rules for encapsulated packets are required in addition
2299to the map-e-portset nat rule.
2300.Pp
2301For example:
2302.Bd -literal -offset indent
2303nat on $gif_mape_if from $int_if:network to any \e
2304      -> $ipv4_mape_src map-e-portset 6/8/0x34
2305.Ed
2306.Pp
2307sets PSID offset 6, PSID length 8, PSID 0x34.
2308.El
2309.Pp
2310Additionally, the
2311.Ar sticky-address
2312option can be specified to help ensure that multiple connections from the
2313same source are mapped to the same redirection address.
2314This option can be used with the
2315.Ar random
2316and
2317.Ar round-robin
2318pool options.
2319Note that by default these associations are destroyed as soon as there are
2320no longer states which refer to them; in order to make the mappings last
2321beyond the lifetime of the states, increase the global options with
2322.Ar set timeout src.track .
2323See
2324.Sx STATEFUL TRACKING OPTIONS
2325for more ways to control the source tracking.
2326.Sh STATE MODULATION
2327Much of the security derived from TCP is attributable to how well the
2328initial sequence numbers (ISNs) are chosen.
2329Some popular stack implementations choose
2330.Em very
2331poor ISNs and thus are normally susceptible to ISN prediction exploits.
2332By applying a
2333.Ar modulate state
2334rule to a TCP connection,
2335.Xr pf 4
2336will create a high quality random sequence number for each connection
2337endpoint.
2338.Pp
2339The
2340.Ar modulate state
2341directive implicitly keeps state on the rule and is
2342only applicable to TCP connections.
2343.Pp
2344For instance:
2345.Bd -literal -offset indent
2346block all
2347pass out proto tcp from any to any modulate state
2348pass in  proto tcp from any to any port 25 flags S/SFRA modulate state
2349.Ed
2350.Pp
2351Note that modulated connections will not recover when the state table
2352is lost (firewall reboot, flushing the state table, etc...).
2353.Xr pf 4
2354will not be able to infer a connection again after the state table flushes
2355the connection's modulator.
2356When the state is lost, the connection may be left dangling until the
2357respective endpoints time out the connection.
2358It is possible on a fast local network for the endpoints to start an ACK
2359storm while trying to resynchronize after the loss of the modulator.
2360The default
2361.Ar flags
2362settings (or a more strict equivalent) should be used on
2363.Ar modulate state
2364rules to prevent ACK storms.
2365.Pp
2366Note that alternative methods are available
2367to prevent loss of the state table
2368and allow for firewall failover.
2369See
2370.Xr carp 4
2371and
2372.Xr pfsync 4
2373for further information.
2374.Sh SYN PROXY
2375By default,
2376.Xr pf 4
2377passes packets that are part of a
2378.Xr tcp 4
2379handshake between the endpoints.
2380The
2381.Ar synproxy state
2382option can be used to cause
2383.Xr pf 4
2384itself to complete the handshake with the active endpoint, perform a handshake
2385with the passive endpoint, and then forward packets between the endpoints.
2386.Pp
2387No packets are sent to the passive endpoint before the active endpoint has
2388completed the handshake, hence so-called SYN floods with spoofed source
2389addresses will not reach the passive endpoint, as the sender can't complete the
2390handshake.
2391.Pp
2392The proxy is transparent to both endpoints, they each see a single
2393connection from/to the other endpoint.
2394.Xr pf 4
2395chooses random initial sequence numbers for both handshakes.
2396Once the handshakes are completed, the sequence number modulators
2397(see previous section) are used to translate further packets of the
2398connection.
2399.Ar synproxy state
2400includes
2401.Ar modulate state .
2402.Pp
2403Rules with
2404.Ar synproxy
2405will not work if
2406.Xr pf 4
2407operates on a
2408.Xr bridge 4 .
2409.Pp
2410Example:
2411.Bd -literal -offset indent
2412pass in proto tcp from any to any port www synproxy state
2413.Ed
2414.Sh STATEFUL TRACKING OPTIONS
2415A number of options related to stateful tracking can be applied on a
2416per-rule basis.
2417.Ar keep state ,
2418.Ar modulate state
2419and
2420.Ar synproxy state
2421support these options, and
2422.Ar keep state
2423must be specified explicitly to apply options to a rule.
2424.Pp
2425.Bl -tag -width xxxx -compact
2426.It Ar max Aq Ar number
2427Limits the number of concurrent states the rule may create.
2428When this limit is reached, further packets that would create
2429state will not match this rule until existing states time out.
2430.It Ar no-sync
2431Prevent state changes for states created by this rule from appearing on the
2432.Xr pfsync 4
2433interface.
2434.It Xo Aq Ar timeout
2435.Aq Ar seconds
2436.Xc
2437Changes the timeout values used for states created by this rule.
2438For a list of all valid timeout names, see
2439.Sx OPTIONS
2440above.
2441.It Ar sloppy
2442Uses a sloppy TCP connection tracker that does not check sequence
2443numbers at all, which makes insertion and ICMP teardown attacks way
2444easier.
2445This is intended to be used in situations where one does not see all
2446packets of a connection, e.g. in asymmetric routing situations.
2447Cannot be used with modulate or synproxy state.
2448.It Ar pflow
2449States created by this rule are exported on the
2450.Xr pflow 4
2451interface.
2452.El
2453.Pp
2454Multiple options can be specified, separated by commas:
2455.Bd -literal -offset indent
2456pass in proto tcp from any to any \e
2457      port www keep state \e
2458      (max 100, source-track rule, max-src-nodes 75, \e
2459      max-src-states 3, tcp.established 60, tcp.closing 5)
2460.Ed
2461.Pp
2462When the
2463.Ar source-track
2464keyword is specified, the number of states per source IP is tracked.
2465.Pp
2466.Bl -tag -width xxxx -compact
2467.It Ar source-track rule
2468The maximum number of states created by this rule is limited by the rule's
2469.Ar max-src-nodes
2470and
2471.Ar max-src-states
2472options.
2473Only state entries created by this particular rule count toward the rule's
2474limits.
2475.It Ar source-track global
2476The number of states created by all rules that use this option is limited.
2477Each rule can specify different
2478.Ar max-src-nodes
2479and
2480.Ar max-src-states
2481options, however state entries created by any participating rule count towards
2482each individual rule's limits.
2483.El
2484.Pp
2485The following limits can be set:
2486.Pp
2487.Bl -tag -width xxxx -compact
2488.It Ar max-src-nodes Aq Ar number
2489Limits the maximum number of source addresses which can simultaneously
2490have state table entries.
2491.It Ar max-src-states Aq Ar number
2492Limits the maximum number of simultaneous state entries that a single
2493source address can create with this rule.
2494.El
2495.Pp
2496For stateful TCP connections, limits on established connections (connections
2497which have completed the TCP 3-way handshake) can also be enforced
2498per source IP.
2499.Pp
2500.Bl -tag -width xxxx -compact
2501.It Ar max-src-conn Aq Ar number
2502Limits the maximum number of simultaneous TCP connections which have
2503completed the 3-way handshake that a single host can make.
2504.It Xo Ar max-src-conn-rate Aq Ar number
2505.No / Aq Ar seconds
2506.Xc
2507Limit the rate of new connections over a time interval.
2508The connection rate is an approximation calculated as a moving average.
2509.El
2510.Pp
2511Because the 3-way handshake ensures that the source address is not being
2512spoofed, more aggressive action can be taken based on these limits.
2513With the
2514.Ar overload Aq Ar table
2515state option, source IP addresses which hit either of the limits on
2516established connections will be added to the named table.
2517This table can be used in the ruleset to block further activity from
2518the offending host, redirect it to a tarpit process, or restrict its
2519bandwidth.
2520.Pp
2521The optional
2522.Ar flush
2523keyword kills all states created by the matching rule which originate
2524from the host which exceeds these limits.
2525The
2526.Ar global
2527modifier to the flush command kills all states originating from the
2528offending host, regardless of which rule created the state.
2529.Pp
2530For example, the following rules will protect the webserver against
2531hosts making more than 100 connections in 10 seconds.
2532Any host which connects faster than this rate will have its address added
2533to the
2534.Aq bad_hosts
2535table and have all states originating from it flushed.
2536Any new packets arriving from this host will be dropped unconditionally
2537by the block rule.
2538.Bd -literal -offset indent
2539block quick from \*(Ltbad_hosts\*(Gt
2540pass in on $ext_if proto tcp to $webserver port www keep state \e
2541	(max-src-conn-rate 100/10, overload \*(Ltbad_hosts\*(Gt flush global)
2542.Ed
2543.Sh OPERATING SYSTEM FINGERPRINTING
2544Passive OS Fingerprinting is a mechanism to inspect nuances of a TCP
2545connection's initial SYN packet and guess at the host's operating system.
2546Unfortunately these nuances are easily spoofed by an attacker so the
2547fingerprint is not useful in making security decisions.
2548But the fingerprint is typically accurate enough to make policy decisions
2549upon.
2550.Pp
2551The fingerprints may be specified by operating system class, by
2552version, or by subtype/patchlevel.
2553The class of an operating system is typically the vendor or genre
2554and would be
2555.Ox
2556for the
2557.Xr pf 4
2558firewall itself.
2559The version of the oldest available
2560.Ox
2561release on the main FTP site
2562would be 2.6 and the fingerprint would be written
2563.Pp
2564.Dl \&"OpenBSD 2.6\&"
2565.Pp
2566The subtype of an operating system is typically used to describe the
2567patchlevel if that patch led to changes in the TCP stack behavior.
2568In the case of
2569.Ox ,
2570the only subtype is for a fingerprint that was
2571normalized by the
2572.Ar no-df
2573scrub option and would be specified as
2574.Pp
2575.Dl \&"OpenBSD 3.3 no-df\&"
2576.Pp
2577Fingerprints for most popular operating systems are provided by
2578.Xr pf.os 5 .
2579Once
2580.Xr pf 4
2581is running, a complete list of known operating system fingerprints may
2582be listed by running:
2583.Pp
2584.Dl # pfctl -so
2585.Pp
2586Filter rules can enforce policy at any level of operating system specification
2587assuming a fingerprint is present.
2588Policy could limit traffic to approved operating systems or even ban traffic
2589from hosts that aren't at the latest service pack.
2590.Pp
2591The
2592.Ar unknown
2593class can also be used as the fingerprint which will match packets for
2594which no operating system fingerprint is known.
2595.Pp
2596Examples:
2597.Bd -literal -offset indent
2598pass  out proto tcp from any os OpenBSD
2599block out proto tcp from any os Doors
2600block out proto tcp from any os "Doors PT"
2601block out proto tcp from any os "Doors PT SP3"
2602block out from any os "unknown"
2603pass on lo0 proto tcp from any os "OpenBSD 3.3 lo0"
2604.Ed
2605.Pp
2606Operating system fingerprinting is limited only to the TCP SYN packet.
2607This means that it will not work on other protocols and will not match
2608a currently established connection.
2609.Pp
2610Caveat: operating system fingerprints are occasionally wrong.
2611There are three problems: an attacker can trivially craft his packets to
2612appear as any operating system he chooses;
2613an operating system patch could change the stack behavior and no fingerprints
2614will match it until the database is updated;
2615and multiple operating systems may have the same fingerprint.
2616.Sh BLOCKING SPOOFED TRAFFIC
2617"Spoofing" is the faking of IP addresses, typically for malicious
2618purposes.
2619The
2620.Ar antispoof
2621directive expands to a set of filter rules which will block all
2622traffic with a source IP from the network(s) directly connected
2623to the specified interface(s) from entering the system through
2624any other interface.
2625.Pp
2626For example, the line
2627.Bd -literal -offset indent
2628antispoof for lo0
2629.Ed
2630.Pp
2631expands to
2632.Bd -literal -offset indent
2633block drop in on ! lo0 inet from 127.0.0.1/8 to any
2634block drop in on ! lo0 inet6 from ::1 to any
2635.Ed
2636.Pp
2637For non-loopback interfaces, there are additional rules to block incoming
2638packets with a source IP address identical to the interface's IP(s).
2639For example, assuming the interface wi0 had an IP address of 10.0.0.1 and a
2640netmask of 255.255.255.0,
2641the line
2642.Bd -literal -offset indent
2643antispoof for wi0 inet
2644.Ed
2645.Pp
2646expands to
2647.Bd -literal -offset indent
2648block drop in on ! wi0 inet from 10.0.0.0/24 to any
2649block drop in inet from 10.0.0.1 to any
2650.Ed
2651.Pp
2652Caveat: Rules created by the
2653.Ar antispoof
2654directive interfere with packets sent over loopback interfaces
2655to local addresses.
2656One should pass these explicitly.
2657.Sh FRAGMENT HANDLING
2658The size of IP datagrams (packets) can be significantly larger than the
2659maximum transmission unit (MTU) of the network.
2660In cases when it is necessary or more efficient to send such large packets,
2661the large packet will be fragmented into many smaller packets that will each
2662fit onto the wire.
2663Unfortunately for a firewalling device, only the first logical fragment will
2664contain the necessary header information for the subprotocol that allows
2665.Xr pf 4
2666to filter on things such as TCP ports or to perform NAT.
2667.Pp
2668Besides the use of
2669.Ar set reassemble
2670option or
2671.Ar scrub
2672rules as described in
2673.Sx TRAFFIC NORMALIZATION
2674above, there are three options for handling fragments in the packet filter.
2675.Pp
2676One alternative is to filter individual fragments with filter rules.
2677If no
2678.Ar scrub
2679rule applies to a fragment or
2680.Ar set reassemble
2681is set to
2682.Cm no
2683, it is passed to the filter.
2684Filter rules with matching IP header parameters decide whether the
2685fragment is passed or blocked, in the same way as complete packets
2686are filtered.
2687Without reassembly, fragments can only be filtered based on IP header
2688fields (source/destination address, protocol), since subprotocol header
2689fields are not available (TCP/UDP port numbers, ICMP code/type).
2690The
2691.Ar fragment
2692option can be used to restrict filter rules to apply only to
2693fragments, but not complete packets.
2694Filter rules without the
2695.Ar fragment
2696option still apply to fragments, if they only specify IP header fields.
2697For instance, the rule
2698.Bd -literal -offset indent
2699pass in proto tcp from any to any port 80
2700.Ed
2701.Pp
2702never applies to a fragment, even if the fragment is part of a TCP
2703packet with destination port 80, because without reassembly this information
2704is not available for each fragment.
2705This also means that fragments cannot create new or match existing
2706state table entries, which makes stateful filtering and address
2707translation (NAT, redirection) for fragments impossible.
2708.Pp
2709It's also possible to reassemble only certain fragments by specifying
2710source or destination addresses or protocols as parameters in
2711.Ar scrub
2712rules.
2713.Pp
2714In most cases, the benefits of reassembly outweigh the additional
2715memory cost, and it's recommended to use
2716.Ar set reassemble
2717option or
2718.Ar scrub
2719rules with the
2720.Ar fragment reassemble
2721modifier to reassemble
2722all fragments.
2723.Pp
2724The memory allocated for fragment caching can be limited using
2725.Xr pfctl 8 .
2726Once this limit is reached, fragments that would have to be cached
2727are dropped until other entries time out.
2728The timeout value can also be adjusted.
2729.Pp
2730When forwarding reassembled IPv6 packets, pf refragments them with
2731the original maximum fragment size.
2732This allows the sender to determine the optimal fragment size by
2733path MTU discovery.
2734.Sh ANCHORS
2735Besides the main ruleset,
2736.Xr pfctl 8
2737can load rulesets into
2738.Ar anchor
2739attachment points.
2740An
2741.Ar anchor
2742is a container that can hold rules, address tables, and other anchors.
2743.Pp
2744An
2745.Ar anchor
2746has a name which specifies the path where
2747.Xr pfctl 8
2748can be used to access the anchor to perform operations on it, such as
2749attaching child anchors to it or loading rules into it.
2750Anchors may be nested, with components separated by
2751.Sq /
2752characters, similar to how file system hierarchies are laid out.
2753The main ruleset is actually the default anchor, so filter and
2754translation rules, for example, may also be contained in any anchor.
2755.Pp
2756An anchor can reference another
2757.Ar anchor
2758attachment point
2759using the following kinds
2760of rules:
2761.Bl -tag -width xxxx
2762.It Ar nat-anchor Aq Ar name
2763Evaluates the
2764.Ar nat
2765rules in the specified
2766.Ar anchor .
2767.It Ar rdr-anchor Aq Ar name
2768Evaluates the
2769.Ar rdr
2770rules in the specified
2771.Ar anchor .
2772.It Ar binat-anchor Aq Ar name
2773Evaluates the
2774.Ar binat
2775rules in the specified
2776.Ar anchor .
2777.It Ar anchor Aq Ar name
2778Evaluates the filter rules in the specified
2779.Ar anchor .
2780.It Xo Ar load anchor
2781.Aq Ar name
2782.Ar from Aq Ar file
2783.Xc
2784Loads the rules from the specified file into the
2785anchor
2786.Ar name .
2787.El
2788.Pp
2789When evaluation of the main ruleset reaches an
2790.Ar anchor
2791rule,
2792.Xr pf 4
2793will proceed to evaluate all rules specified in that anchor.
2794.Pp
2795Matching filter and translation rules marked with the
2796.Ar quick
2797option are final and abort the evaluation of the rules in other
2798anchors and the main ruleset.
2799If the
2800.Ar anchor
2801itself is marked with the
2802.Ar quick
2803option,
2804ruleset evaluation will terminate when the anchor is exited if the packet is
2805matched by any rule within the anchor.
2806.Pp
2807.Ar anchor
2808rules are evaluated relative to the anchor in which they are contained.
2809For example, all
2810.Ar anchor
2811rules specified in the main ruleset will reference anchor
2812attachment points underneath the main ruleset, and
2813.Ar anchor
2814rules specified in a file loaded from a
2815.Ar load anchor
2816rule will be attached under that anchor point.
2817.Pp
2818Rules may be contained in
2819.Ar anchor
2820attachment points which do not contain any rules when the main ruleset
2821is loaded, and later such anchors can be manipulated through
2822.Xr pfctl 8
2823without reloading the main ruleset or other anchors.
2824For example,
2825.Bd -literal -offset indent
2826ext_if = \&"kue0\&"
2827block on $ext_if all
2828anchor spam
2829pass out on $ext_if all
2830pass in on $ext_if proto tcp from any \e
2831      to $ext_if port smtp
2832.Ed
2833.Pp
2834blocks all packets on the external interface by default, then evaluates
2835all rules in the
2836.Ar anchor
2837named "spam", and finally passes all outgoing connections and
2838incoming connections to port 25.
2839.Bd -literal -offset indent
2840# echo \&"block in quick from 1.2.3.4 to any\&" \&| \e
2841      pfctl -a spam -f -
2842.Ed
2843.Pp
2844This loads a single rule into the
2845.Ar anchor ,
2846which blocks all packets from a specific address.
2847.Pp
2848The anchor can also be populated by adding a
2849.Ar load anchor
2850rule after the
2851.Ar anchor
2852rule:
2853.Bd -literal -offset indent
2854anchor spam
2855load anchor spam from "/etc/pf-spam.conf"
2856.Ed
2857.Pp
2858When
2859.Xr pfctl 8
2860loads
2861.Nm pf.conf ,
2862it will also load all the rules from the file
2863.Pa /etc/pf-spam.conf
2864into the anchor.
2865.Pp
2866Optionally,
2867.Ar anchor
2868rules can specify packet filtering parameters using the same syntax as
2869filter rules.
2870When parameters are used, the
2871.Ar anchor
2872rule is only evaluated for matching packets.
2873This allows conditional evaluation of anchors, like:
2874.Bd -literal -offset indent
2875block on $ext_if all
2876anchor spam proto tcp from any to any port smtp
2877pass out on $ext_if all
2878pass in on $ext_if proto tcp from any to $ext_if port smtp
2879.Ed
2880.Pp
2881The rules inside
2882.Ar anchor
2883spam are only evaluated for
2884.Ar tcp
2885packets with destination port 25.
2886Hence,
2887.Bd -literal -offset indent
2888# echo \&"block in quick from 1.2.3.4 to any" \&| \e
2889      pfctl -a spam -f -
2890.Ed
2891.Pp
2892will only block connections from 1.2.3.4 to port 25.
2893.Pp
2894Anchors may end with the asterisk
2895.Pq Sq *
2896character, which signifies that all anchors attached at that point
2897should be evaluated in the alphabetical ordering of their anchor name.
2898For example,
2899.Bd -literal -offset indent
2900anchor "spam/*"
2901.Ed
2902.Pp
2903will evaluate each rule in each anchor attached to the
2904.Li spam
2905anchor.
2906Note that it will only evaluate anchors that are directly attached to the
2907.Li spam
2908anchor, and will not descend to evaluate anchors recursively.
2909.Pp
2910Since anchors are evaluated relative to the anchor in which they are
2911contained, there is a mechanism for accessing the parent and ancestor
2912anchors of a given anchor.
2913Similar to file system path name resolution, if the sequence
2914.Dq ..
2915appears as an anchor path component, the parent anchor of the current
2916anchor in the path evaluation at that point will become the new current
2917anchor.
2918As an example, consider the following:
2919.Bd -literal -offset indent
2920# echo ' anchor "spam/allowed" ' | pfctl -f -
2921# echo -e ' anchor "../banned" \en pass' | \e
2922      pfctl -a spam/allowed -f -
2923.Ed
2924.Pp
2925Evaluation of the main ruleset will lead into the
2926.Li spam/allowed
2927anchor, which will evaluate the rules in the
2928.Li spam/banned
2929anchor, if any, before finally evaluating the
2930.Ar pass
2931rule.
2932.Pp
2933Filter rule
2934.Ar anchors
2935can also be loaded inline in the ruleset within a brace ('{' '}') delimited
2936block.
2937Brace delimited blocks may contain rules or other brace-delimited blocks.
2938When anchors are loaded this way the anchor name becomes optional.
2939.Bd -literal -offset indent
2940anchor "external" on $ext_if {
2941	block
2942	anchor out {
2943		pass proto tcp from any to port { 25, 80, 443 }
2944	}
2945	pass in proto tcp to any port 22
2946}
2947.Ed
2948.Pp
2949Since the parser specification for anchor names is a string, any
2950reference to an anchor name containing
2951.Sq /
2952characters will require double quote
2953.Pq Sq \&"
2954characters around the anchor name.
2955.Sh SCTP CONSIDERATIONS
2956.Xr pf 4
2957supports
2958.Xr sctp 4
2959connections.
2960It can match ports, track state and NAT SCTP traffic.
2961However, it will not alter port numbers during nat or rdr translations.
2962Doing so would break SCTP multihoming.
2963.Sh TRANSLATION EXAMPLES
2964This example maps incoming requests on port 80 to port 8080, on
2965which a daemon is running (because, for example, it is not run as root,
2966and therefore lacks permission to bind to port 80).
2967.Bd -literal
2968# use a macro for the interface name, so it can be changed easily
2969ext_if = \&"ne3\&"
2970
2971# map daemon on 8080 to appear to be on 80
2972rdr on $ext_if proto tcp from any to any port 80 -\*(Gt 127.0.0.1 port 8080
2973.Ed
2974.Pp
2975If the
2976.Ar pass
2977modifier is given, packets matching the translation rule are passed without
2978inspecting the filter rules:
2979.Bd -literal
2980rdr pass on $ext_if proto tcp from any to any port 80 -\*(Gt 127.0.0.1 \e
2981      port 8080
2982.Ed
2983.Pp
2984In the example below, vlan12 is configured as 192.168.168.1;
2985the machine translates all packets coming from 192.168.168.0/24 to 204.92.77.111
2986when they are going out any interface except vlan12.
2987This has the net effect of making traffic from the 192.168.168.0/24
2988network appear as though it is the Internet routable address
2989204.92.77.111 to nodes behind any interface on the router except
2990for the nodes on vlan12.
2991(Thus, 192.168.168.1 can talk to the 192.168.168.0/24 nodes.)
2992.Bd -literal
2993nat on ! vlan12 from 192.168.168.0/24 to any -\*(Gt 204.92.77.111
2994.Ed
2995.Pp
2996In the example below, the machine sits between a fake internal 144.19.74.*
2997network, and a routable external IP of 204.92.77.100.
2998The
2999.Ar no nat
3000rule excludes protocol AH from being translated.
3001.Bd -literal
3002# NO NAT
3003no nat on $ext_if proto ah from 144.19.74.0/24 to any
3004nat on $ext_if from 144.19.74.0/24 to any -\*(Gt 204.92.77.100
3005.Ed
3006.Pp
3007In the example below, packets bound for one specific server, as well as those
3008generated by the sysadmins are not proxied; all other connections are.
3009.Bd -literal
3010# NO RDR
3011no rdr on $int_if proto { tcp, udp } from any to $server port 80
3012no rdr on $int_if proto { tcp, udp } from $sysadmins to any port 80
3013rdr on $int_if proto { tcp, udp } from any to any port 80 -\*(Gt 127.0.0.1 \e
3014      port 80
3015.Ed
3016.Pp
3017This longer example uses both a NAT and a redirection.
3018The external interface has the address 157.161.48.183.
3019On localhost, we are running
3020.Xr ftp-proxy 8 ,
3021waiting for FTP sessions to be redirected to it.
3022The three mandatory anchors for
3023.Xr ftp-proxy 8
3024are omitted from this example; see the
3025.Xr ftp-proxy 8
3026manpage.
3027.Bd -literal
3028# NAT
3029# Translate outgoing packets' source addresses (any protocol).
3030# In this case, any address but the gateway's external address is mapped.
3031nat on $ext_if inet from ! ($ext_if) to any -\*(Gt ($ext_if)
3032
3033# NAT PROXYING
3034# Map outgoing packets' source port to an assigned proxy port instead of
3035# an arbitrary port.
3036# In this case, proxy outgoing isakmp with port 500 on the gateway.
3037nat on $ext_if inet proto udp from any port = isakmp to any -\*(Gt ($ext_if) \e
3038      port 500
3039
3040# BINAT
3041# Translate outgoing packets' source address (any protocol).
3042# Translate incoming packets' destination address to an internal machine
3043# (bidirectional).
3044binat on $ext_if from 10.1.2.150 to any -\*(Gt $ext_if
3045
3046# Translate packets arriving on $peer_if addressed to 172.22.16.0/20
3047# to the corresponding address in 172.21.16.0/20 (bidirectional).
3048binat on $peer_if from 172.21.16.0/20 to any -> 172.22.16.0/20
3049
3050# RDR
3051# Translate incoming packets' destination addresses.
3052# As an example, redirect a TCP and UDP port to an internal machine.
3053rdr on $ext_if inet proto tcp from any to ($ext_if) port 8080 \e
3054      -\*(Gt 10.1.2.151 port 22
3055rdr on $ext_if inet proto udp from any to ($ext_if) port 8080 \e
3056      -\*(Gt 10.1.2.151 port 53
3057
3058# RDR
3059# Translate outgoing ftp control connections to send them to localhost
3060# for proxying with ftp-proxy(8) running on port 8021.
3061rdr on $int_if proto tcp from any to any port 21 -\*(Gt 127.0.0.1 port 8021
3062.Ed
3063.Pp
3064In this example, a NAT gateway is set up to translate internal addresses
3065using a pool of public addresses (192.0.2.16/28) and to redirect
3066incoming web server connections to a group of web servers on the internal
3067network.
3068.Bd -literal
3069# NAT LOAD BALANCE
3070# Translate outgoing packets' source addresses using an address pool.
3071# A given source address is always translated to the same pool address by
3072# using the source-hash keyword.
3073nat on $ext_if inet from any to any -\*(Gt 192.0.2.16/28 source-hash
3074
3075# RDR ROUND ROBIN
3076# Translate incoming web server connections to a group of web servers on
3077# the internal network.
3078rdr on $ext_if proto tcp from any to any port 80 \e
3079      -\*(Gt { 10.1.2.155, 10.1.2.160, 10.1.2.161 } round-robin
3080.Ed
3081.Sh FILTER EXAMPLES
3082.Bd -literal
3083# The external interface is kue0
3084# (157.161.48.183, the only routable address)
3085# and the private network is 10.0.0.0/8, for which we are doing NAT.
3086
3087# Reassemble incoming traffic
3088set reassemble yes
3089
3090# use a macro for the interface name, so it can be changed easily
3091ext_if = \&"kue0\&"
3092
3093# block and log everything by default
3094block return log on $ext_if all
3095
3096# block anything coming from source we have no back routes for
3097block in from no-route to any
3098
3099# block packets whose ingress interface does not match the one in
3100# the route back to their source address
3101block in from urpf-failed to any
3102
3103# block and log outgoing packets that do not have our address as source,
3104# they are either spoofed or something is misconfigured (NAT disabled,
3105# for instance), we want to be nice and do not send out garbage.
3106block out log quick on $ext_if from ! 157.161.48.183 to any
3107
3108# silently drop broadcasts (cable modem noise)
3109block in quick on $ext_if from any to 255.255.255.255
3110
3111# block and log incoming packets from reserved address space and invalid
3112# addresses, they are either spoofed or misconfigured, we cannot reply to
3113# them anyway (hence, no return-rst).
3114block in log quick on $ext_if from { 10.0.0.0/8, 172.16.0.0/12, \e
3115      192.168.0.0/16, 255.255.255.255/32 } to any
3116
3117# ICMP
3118
3119# pass out/in certain ICMP queries and keep state (ping)
3120# state matching is done on host addresses and ICMP id (not type/code),
3121# so replies (like 0/0 for 8/0) will match queries
3122# ICMP error messages (which always refer to a TCP/UDP packet) are
3123# handled by the TCP/UDP states
3124pass on $ext_if inet proto icmp all icmp-type 8 code 0
3125
3126# UDP
3127
3128# pass out all UDP connections and keep state
3129pass out on $ext_if proto udp all
3130
3131# pass in certain UDP connections and keep state (DNS)
3132pass in on $ext_if proto udp from any to any port domain
3133
3134# TCP
3135
3136# pass out all TCP connections and modulate state
3137pass out on $ext_if proto tcp all modulate state
3138
3139# pass in certain TCP connections and keep state (SSH, SMTP, DNS, IDENT)
3140pass in on $ext_if proto tcp from any to any port { ssh, smtp, domain, \e
3141      auth }
3142
3143# Do not allow Windows 9x SMTP connections since they are typically
3144# a viral worm. Alternately we could limit these OSes to 1 connection each.
3145block in on $ext_if proto tcp from any os {"Windows 95", "Windows 98"} \e
3146      to any port smtp
3147
3148# IPv6
3149# pass in/out all IPv6 traffic: note that we have to enable this in two
3150# different ways, on both our physical interface and our tunnel
3151pass quick on gif0 inet6
3152pass quick on $ext_if proto ipv6
3153
3154# Packet Tagging
3155
3156# three interfaces: $int_if, $ext_if, and $wifi_if (wireless). NAT is
3157# being done on $ext_if for all outgoing packets. tag packets in on
3158# $int_if and pass those tagged packets out on $ext_if.  all other
3159# outgoing packets (i.e., packets from the wireless network) are only
3160# permitted to access port 80.
3161
3162pass in on $int_if from any to any tag INTNET
3163pass in on $wifi_if from any to any
3164
3165block out on $ext_if from any to any
3166pass out quick on $ext_if tagged INTNET
3167pass out on $ext_if proto tcp from any to any port 80
3168
3169# tag incoming packets as they are redirected to spamd(8). use the tag
3170# to pass those packets through the packet filter.
3171
3172rdr on $ext_if inet proto tcp from \*(Ltspammers\*(Gt to port smtp \e
3173	tag SPAMD -\*(Gt 127.0.0.1 port spamd
3174
3175block in on $ext_if
3176pass in on $ext_if inet proto tcp tagged SPAMD
3177.Ed
3178.Sh GRAMMAR
3179Syntax for
3180.Nm
3181in BNF:
3182.Bd -literal
3183line           = ( option | ether-rule | pf-rule | nat-rule | binat-rule |
3184                 rdr-rule | antispoof-rule | altq-rule | queue-rule |
3185                 trans-anchors | anchor-rule | anchor-close | load-anchor |
3186                 table-rule | include )
3187
3188option         = "set" ( [ "timeout" ( timeout | "{" timeout-list "}" ) ] |
3189                 [ "ruleset-optimization" [ "none" | "basic" | "profile" ]] |
3190                 [ "optimization" [ "default" | "normal" |
3191                 "high-latency" | "satellite" |
3192                 "aggressive" | "conservative" ] ]
3193                 [ "limit" ( limit-item | "{" limit-list "}" ) ] |
3194                 [ "loginterface" ( interface-name | "none" ) ] |
3195                 [ "block-policy" ( "drop" | "return" ) ] |
3196                 [ "state-policy" ( "if-bound" | "floating" ) ]
3197                 [ "state-defaults" state-opts ]
3198                 [ "require-order" ( "yes" | "no" ) ]
3199                 [ "fingerprints" filename ] |
3200                 [ "skip on" ifspec ] |
3201                 [ "debug" ( "none" | "urgent" | "misc" | "loud" ) ]
3202                 [ "keepcounters" ] )
3203
3204ether-rule     = "ether" etheraction [ ( "in" | "out" ) ]
3205                 [ "quick" ] [ "on" ifspec ] [ "bridge-to" interface-name ]
3206                 [ etherprotospec ] etherhosts [ "l3" hosts ]
3207                 [ etherfilteropt-list ]
3208
3209pf-rule        = action [ ( "in" | "out" ) ]
3210                 [ "log" [ "(" logopts ")"] ] [ "quick" ]
3211                 [ "on" ifspec ] [ route ] [ af ] [ protospec ]
3212                 hosts [ filteropt-list ]
3213
3214logopts        = logopt [ "," logopts ]
3215logopt         = "all" | "user" | "to" interface-name
3216
3217etherfilteropt-list = etherfilteropt-list etherfilteropt | etherfilteropt
3218etherfilteropt = "tag" string | "tagged" string | "queue" ( string ) |
3219                 "ridentifier" number | "label" string
3220
3221filteropt-list = filteropt-list filteropt | filteropt
3222filteropt      = user | group | flags | icmp-type | icmp6-type | "tos" tos |
3223                 ( "no" | "keep" | "modulate" | "synproxy" ) "state"
3224                 [ "(" state-opts ")" ] |
3225                 "fragment" | "no-df" | "min-ttl" number | "set-tos" tos |
3226                 "max-mss" number | "random-id" | "reassemble tcp" |
3227                 fragmentation | "allow-opts" |
3228                 "label" string | "tag" string | [ ! ] "tagged" string |
3229                 "set prio" ( number | "(" number [ [ "," ] number ] ")" ) |
3230                 "queue" ( string | "(" string [ [ "," ] string ] ")" ) |
3231                 "rtable" number | "probability" number"%" | "prio" number |
3232                 "dnpipe" ( number | "(" number "," number ")" ) |
3233                 "dnqueue" ( number | "(" number "," number ")" ) |
3234                 "ridentifier" number
3235
3236nat-rule       = [ "no" ] "nat" [ "pass" [ "log" [ "(" logopts ")" ] ] ]
3237                 [ "on" ifspec ] [ af ]
3238                 [ protospec ] hosts [ "tag" string ] [ "tagged" string ]
3239                 [ "-\*(Gt" ( redirhost | "{" redirhost-list "}" )
3240                 [ portspec ] [ pooltype ] [ "static-port" ]
3241                 [ "map-e-portset" number "/" number "/" number ] ]
3242
3243binat-rule     = [ "no" ] "binat" [ "pass" [ "log" [ "(" logopts ")" ] ] ]
3244                 [ "on" interface-name ] [ af ]
3245                 [ "proto" ( proto-name | proto-number ) ]
3246                 "from" address [ "/" mask-bits ] "to" ipspec
3247                 [ "tag" string ] [ "tagged" string ]
3248                 [ "-\*(Gt" address [ "/" mask-bits ] ]
3249
3250rdr-rule       = [ "no" ] "rdr" [ "pass" [ "log" [ "(" logopts ")" ] ] ]
3251                 [ "on" ifspec ] [ af ]
3252                 [ protospec ] hosts [ "tag" string ] [ "tagged" string ]
3253                 [ "-\*(Gt" ( redirhost | "{" redirhost-list "}" )
3254                 [ portspec ] [ pooltype ] ]
3255
3256antispoof-rule = "antispoof" [ "log" ] [ "quick" ]
3257                 "for" ifspec [ af ] [ "label" string ]
3258                 [ "ridentifier" number ]
3259
3260table-rule     = "table" "\*(Lt" string "\*(Gt" [ tableopts-list ]
3261tableopts-list = tableopts-list tableopts | tableopts
3262tableopts      = "persist" | "const" | "counters" | "file" string |
3263                 "{" [ tableaddr-list ] "}"
3264tableaddr-list = tableaddr-list [ "," ] tableaddr-spec | tableaddr-spec
3265tableaddr-spec = [ "!" ] tableaddr [ "/" mask-bits ]
3266tableaddr      = hostname | ifspec | "self" |
3267                 ipv4-dotted-quad | ipv6-coloned-hex
3268
3269altq-rule      = "altq on" interface-name queueopts-list
3270                 "queue" subqueue
3271queue-rule     = "queue" string [ "on" interface-name ] queueopts-list
3272                 subqueue
3273
3274anchor-rule    = "anchor" [ string ] [ ( "in" | "out" ) ] [ "on" ifspec ]
3275                 [ af ] [ protospec ] [ hosts ] [ filteropt-list ] [ "{" ]
3276
3277anchor-close   = "}"
3278
3279trans-anchors  = ( "nat-anchor" | "rdr-anchor" | "binat-anchor" ) string
3280                 [ "on" ifspec ] [ af ] [ "proto" ] [ protospec ] [ hosts ]
3281
3282load-anchor    = "load anchor" string "from" filename
3283
3284queueopts-list = queueopts-list queueopts | queueopts
3285queueopts      = [ "bandwidth" bandwidth-spec ] |
3286                 [ "qlimit" number ] | [ "tbrsize" number ] |
3287                 [ "priority" number ] | [ schedulers ]
3288schedulers     = ( cbq-def | priq-def | hfsc-def )
3289bandwidth-spec = "number" ( "b" | "Kb" | "Mb" | "Gb" | "%" )
3290
3291etheraction    = "pass" | "block"
3292action         = "pass" | "match" | "block" [ return ] | [ "no" ] "scrub"
3293return         = "drop" | "return" | "return-rst" [ "( ttl" number ")" ] |
3294                 "return-icmp" [ "(" icmpcode [ [ "," ] icmp6code ] ")" ] |
3295                 "return-icmp6" [ "(" icmp6code ")" ]
3296icmpcode       = ( icmp-code-name | icmp-code-number )
3297icmp6code      = ( icmp6-code-name | icmp6-code-number )
3298
3299ifspec         = ( [ "!" ] ( interface-name | interface-group ) ) |
3300                 "{" interface-list "}"
3301interface-list = [ "!" ] ( interface-name | interface-group )
3302                 [ [ "," ] interface-list ]
3303route          = ( "route-to" | "reply-to" | "dup-to" )
3304                 ( routehost | "{" routehost-list "}" )
3305                 [ pooltype ]
3306af             = "inet" | "inet6"
3307
3308etherprotospec = "proto" ( proto-number | "{" etherproto-list "}" )
3309etherproto-list	= proto-number [ [ "," ] etherproto-list ]
3310protospec      = "proto" ( proto-name | proto-number |
3311                 "{" proto-list "}" )
3312proto-list     = ( proto-name | proto-number ) [ [ "," ] proto-list ]
3313
3314etherhosts     = "from" macaddress "to" macaddress
3315macaddress     = mac | mac "/" masklen | mac "&" mask
3316
3317hosts          = "all" |
3318                 "from" ( "any" | "no-route" | "urpf-failed" | "self" | host |
3319                 "{" host-list "}" ) [ port ] [ os ]
3320                 "to"   ( "any" | "no-route" | "self" | host |
3321                 "{" host-list "}" ) [ port ]
3322
3323ipspec         = "any" | host | "{" host-list "}"
3324host           = [ "!" ] ( address [ "/" mask-bits ] | "\*(Lt" string "\*(Gt" )
3325redirhost      = address [ "/" mask-bits ]
3326routehost      = "(" interface-name [ address [ "/" mask-bits ] ] ")"
3327address        = ( interface-name | interface-group |
3328                 "(" ( interface-name | interface-group ) ")" |
3329                 hostname | ipv4-dotted-quad | ipv6-coloned-hex )
3330host-list      = host [ [ "," ] host-list ]
3331redirhost-list = redirhost [ [ "," ] redirhost-list ]
3332routehost-list = routehost [ [ "," ] routehost-list ]
3333
3334port           = "port" ( unary-op | binary-op | "{" op-list "}" )
3335portspec       = "port" ( number | name ) [ ":" ( "*" | number | name ) ]
3336os             = "os"  ( os-name | "{" os-list "}" )
3337user           = "user" ( unary-op | binary-op | "{" op-list "}" )
3338group          = "group" ( unary-op | binary-op | "{" op-list "}" )
3339
3340unary-op       = [ "=" | "!=" | "\*(Lt" | "\*(Le" | "\*(Gt" | "\*(Ge" ]
3341                 ( name | number )
3342binary-op      = number ( "\*(Lt\*(Gt" | "\*(Gt\*(Lt" | ":" ) number
3343op-list        = ( unary-op | binary-op ) [ [ "," ] op-list ]
3344
3345os-name        = operating-system-name
3346os-list        = os-name [ [ "," ] os-list ]
3347
3348flags          = "flags" ( [ flag-set ] "/"  flag-set | "any" )
3349flag-set       = [ "F" ] [ "S" ] [ "R" ] [ "P" ] [ "A" ] [ "U" ] [ "E" ]
3350                 [ "W" ]
3351
3352icmp-type      = "icmp-type" ( icmp-type-code | "{" icmp-list "}" )
3353icmp6-type     = "icmp6-type" ( icmp-type-code | "{" icmp-list "}" )
3354icmp-type-code = ( icmp-type-name | icmp-type-number )
3355                 [ "code" ( icmp-code-name | icmp-code-number ) ]
3356icmp-list      = icmp-type-code [ [ "," ] icmp-list ]
3357
3358tos            = ( "lowdelay" | "throughput" | "reliability" |
3359                 [ "0x" ] number )
3360
3361state-opts     = state-opt [ [ "," ] state-opts ]
3362state-opt      = ( "max" number | "no-sync" | timeout | "sloppy" |
3363                 "source-track" [ ( "rule" | "global" ) ] |
3364                 "max-src-nodes" number | "max-src-states" number |
3365                 "max-src-conn" number |
3366                 "max-src-conn-rate" number "/" number |
3367                 "overload" "\*(Lt" string "\*(Gt" [ "flush" ] |
3368                 "if-bound" | "floating" | "pflow" )
3369
3370fragmentation  = [ "fragment reassemble" ]
3371
3372timeout-list   = timeout [ [ "," ] timeout-list ]
3373timeout        = ( "tcp.first" | "tcp.opening" | "tcp.established" |
3374                 "tcp.closing" | "tcp.finwait" | "tcp.closed" |
3375                 "sctp.first" | "sctp.opening" | "sctp.established" |
3376                 "sctp.closing" | "sctp.closed" |
3377                 "udp.first" | "udp.single" | "udp.multiple" |
3378                 "icmp.first" | "icmp.error" |
3379                 "other.first" | "other.single" | "other.multiple" |
3380                 "frag" | "interval" | "src.track" |
3381                 "adaptive.start" | "adaptive.end" ) number
3382
3383limit-list     = limit-item [ [ "," ] limit-list ]
3384limit-item     = ( "states" | "frags" | "src-nodes" ) number
3385
3386pooltype       = ( "bitmask" | "random" |
3387                 "source-hash" [ ( hex-key | string-key ) ] |
3388                 "round-robin" ) [ sticky-address ]
3389
3390subqueue       = string | "{" queue-list "}"
3391queue-list     = string [ [ "," ] string ]
3392cbq-def        = "cbq" [ "(" cbq-opt [ [ "," ] cbq-opt ] ")" ]
3393priq-def       = "priq" [ "(" priq-opt [ [ "," ] priq-opt ] ")" ]
3394hfsc-def       = "hfsc" [ "(" hfsc-opt [ [ "," ] hfsc-opt ] ")" ]
3395cbq-opt        = ( "default" | "borrow" | "red" | "ecn" | "rio" )
3396priq-opt       = ( "default" | "red" | "ecn" | "rio" )
3397hfsc-opt       = ( "default" | "red" | "ecn" | "rio" |
3398                 linkshare-sc | realtime-sc | upperlimit-sc )
3399linkshare-sc   = "linkshare" sc-spec
3400realtime-sc    = "realtime" sc-spec
3401upperlimit-sc  = "upperlimit" sc-spec
3402sc-spec        = ( bandwidth-spec |
3403                 "(" bandwidth-spec number bandwidth-spec ")" )
3404include        = "include" filename
3405.Ed
3406.Sh FILES
3407.Bl -tag -width "/etc/protocols" -compact
3408.It Pa /etc/hosts
3409Host name database.
3410.It Pa /etc/pf.conf
3411Default location of the ruleset file.
3412The file has to be created manually as it is not installed with a
3413standard installation.
3414.It Pa /etc/pf.os
3415Default location of OS fingerprints.
3416.It Pa /etc/protocols
3417Protocol name database.
3418.It Pa /etc/services
3419Service name database.
3420.El
3421.Sh SEE ALSO
3422.Xr altq 4 ,
3423.Xr carp 4 ,
3424.Xr icmp 4 ,
3425.Xr icmp6 4 ,
3426.Xr ip 4 ,
3427.Xr ip6 4 ,
3428.Xr pf 4 ,
3429.Xr pflow 4 ,
3430.Xr pfsync 4 ,
3431.Xr tcp 4 ,
3432.Xr sctp 4 ,
3433.Xr udp 4 ,
3434.Xr hosts 5 ,
3435.Xr pf.os 5 ,
3436.Xr protocols 5 ,
3437.Xr services 5 ,
3438.Xr ftp-proxy 8 ,
3439.Xr pfctl 8 ,
3440.Xr pflogd 8
3441.Sh HISTORY
3442The
3443.Nm
3444file format first appeared in
3445.Ox 3.0 .
3446