xref: /freebsd/sbin/ipfw/ipfw.8 (revision 30d239bc4c510432e65a84fa1c14ed67a3ab1c92)
1.\"
2.\" $FreeBSD$
3.\"
4.Dd October 19, 2007
5.Dt IPFW 8
6.Os
7.Sh NAME
8.Nm ipfw
9.Nd IP firewall and traffic shaper control program
10.Sh SYNOPSIS
11.Nm
12.Op Fl cq
13.Cm add
14.Ar rule
15.Nm
16.Op Fl acdefnNStT
17.Op Cm set Ar N
18.Brq Cm list | show
19.Op Ar rule | first-last ...
20.Nm
21.Op Fl f | q
22.Op Cm set Ar N
23.Cm flush
24.Nm
25.Op Fl q
26.Op Cm set Ar N
27.Brq Cm delete | zero | resetlog
28.Op Ar number ...
29.Nm
30.Cm enable
31.Brq Cm firewall | altq | one_pass | debug | verbose | dyn_keepalive
32.Nm
33.Cm disable
34.Brq Cm firewall | altq | one_pass | debug | verbose | dyn_keepalive
35.Pp
36.Nm
37.Cm set Oo Cm disable Ar number ... Oc Op Cm enable Ar number ...
38.Nm
39.Cm set move
40.Op Cm rule
41.Ar number Cm to Ar number
42.Nm
43.Cm set swap Ar number number
44.Nm
45.Cm set show
46.Pp
47.Nm
48.Cm table Ar number Cm add Ar addr Ns Oo / Ns Ar masklen Oc Op Ar value
49.Nm
50.Cm table Ar number Cm delete Ar addr Ns Op / Ns Ar masklen
51.Nm
52.Cm table Ar number Cm flush
53.Nm
54.Cm table Ar number Cm list
55.Pp
56.Nm
57.Brq Cm pipe | queue
58.Ar number
59.Cm config
60.Ar config-options
61.Nm
62.Op Fl s Op Ar field
63.Brq Cm pipe | queue
64.Brq Cm delete | list | show
65.Op Ar number ...
66.Pp
67.Nm
68.Cm nat
69.Ar number
70.Cm config
71.Ar config-options
72.Pp
73.Nm
74.Op Fl cfnNqS
75.Oo
76.Fl p Ar preproc
77.Oo
78.Ar preproc-flags
79.Oc
80.Oc
81.Ar pathname
82.Sh DESCRIPTION
83The
84.Nm
85utility is the user interface for controlling the
86.Xr ipfw 4
87firewall and the
88.Xr dummynet 4
89traffic shaper in
90.Fx .
91.Pp
92An
93.Nm
94configuration, or
95.Em ruleset ,
96is made of a list of
97.Em rules
98numbered from 1 to 65535.
99Packets are passed to
100.Nm
101from a number of different places in the protocol stack
102(depending on the source and destination of the packet,
103it is possible that
104.Nm
105is invoked multiple times on the same packet).
106The packet passed to the firewall is compared
107against each of the rules in the firewall
108.Em ruleset .
109When a match is found, the action corresponding to the
110matching rule is performed.
111.Pp
112Depending on the action and certain system settings, packets
113can be reinjected into the firewall at some rule after the
114matching one for further processing.
115.Pp
116An
117.Nm
118ruleset always includes a
119.Em default
120rule (numbered 65535) which cannot be modified or deleted,
121and matches all packets.
122The action associated with the
123.Em default
124rule can be either
125.Cm deny
126or
127.Cm allow
128depending on how the kernel is configured.
129.Pp
130If the ruleset includes one or more rules with the
131.Cm keep-state
132or
133.Cm limit
134option, then
135.Nm
136assumes a
137.Em stateful
138behaviour, i.e., upon a match it will create dynamic rules matching
139the exact parameters (addresses and ports) of the matching packet.
140.Pp
141These dynamic rules, which have a limited lifetime, are checked
142at the first occurrence of a
143.Cm check-state ,
144.Cm keep-state
145or
146.Cm limit
147rule, and are typically used to open the firewall on-demand to
148legitimate traffic only.
149See the
150.Sx STATEFUL FIREWALL
151and
152.Sx EXAMPLES
153Sections below for more information on the stateful behaviour of
154.Nm .
155.Pp
156All rules (including dynamic ones) have a few associated counters:
157a packet count, a byte count, a log count and a timestamp
158indicating the time of the last match.
159Counters can be displayed or reset with
160.Nm
161commands.
162.Pp
163Rules can be added with the
164.Cm add
165command; deleted individually or in groups with the
166.Cm delete
167command, and globally (except those in set 31) with the
168.Cm flush
169command; displayed, optionally with the content of the
170counters, using the
171.Cm show
172and
173.Cm list
174commands.
175Finally, counters can be reset with the
176.Cm zero
177and
178.Cm resetlog
179commands.
180.Pp
181Also, each rule belongs to one of 32 different
182.Em sets
183, and there are
184.Nm
185commands to atomically manipulate sets, such as enable,
186disable, swap sets, move all rules in a set to another
187one, delete all rules in a set.
188These can be useful to
189install temporary configurations, or to test them.
190See Section
191.Sx SETS OF RULES
192for more information on
193.Em sets .
194.Pp
195The following options are available:
196.Bl -tag -width indent
197.It Fl a
198While listing, show counter values.
199The
200.Cm show
201command just implies this option.
202.It Fl b
203Only show the action and the comment, not the body of a rule.
204Implies
205.Fl c .
206.It Fl c
207When entering or showing rules, print them in compact form,
208i.e., without the optional "ip from any to any" string
209when this does not carry any additional information.
210.It Fl d
211While listing, show dynamic rules in addition to static ones.
212.It Fl e
213While listing, if the
214.Fl d
215option was specified, also show expired dynamic rules.
216.It Fl f
217Do not ask for confirmation for commands that can cause problems
218if misused,
219.No i.e. Cm flush .
220If there is no tty associated with the process, this is implied.
221.It Fl n
222Only check syntax of the command strings, without actually passing
223them to the kernel.
224.It Fl N
225Try to resolve addresses and service names in output.
226.It Fl q
227While
228.Cm add Ns ing ,
229.Cm zero Ns ing ,
230.Cm resetlog Ns ging
231or
232.Cm flush Ns ing ,
233be quiet about actions
234(implies
235.Fl f ) .
236This is useful for adjusting rules by executing multiple
237.Nm
238commands in a script
239(e.g.,
240.Ql sh\ /etc/rc.firewall ) ,
241or by processing a file of many
242.Nm
243rules across a remote login session.
244It also stops a table add or delete
245from failing if the entry already exists or is not present.
246If a
247.Cm flush
248is performed in normal (verbose) mode (with the default kernel
249configuration), it prints a message.
250Because all rules are flushed, the message might not be delivered
251to the login session, causing the remote login session to be closed
252and the remainder of the ruleset to not be processed.
253Access to the console would then be required to recover.
254.It Fl S
255While listing rules, show the
256.Em set
257each rule belongs to.
258If this flag is not specified, disabled rules will not be
259listed.
260.It Fl s Op Ar field
261While listing pipes, sort according to one of the four
262counters (total or current packets or bytes).
263.It Fl t
264While listing, show last match timestamp (converted with ctime()).
265.It Fl T
266While listing, show last match timestamp (as seconds from the epoch).
267This form can be more convenient for postprocessing by scripts.
268.El
269.Pp
270To ease configuration, rules can be put into a file which is
271processed using
272.Nm
273as shown in the last synopsis line.
274An absolute
275.Ar pathname
276must be used.
277The file will be read line by line and applied as arguments to the
278.Nm
279utility.
280.Pp
281Optionally, a preprocessor can be specified using
282.Fl p Ar preproc
283where
284.Ar pathname
285is to be piped through.
286Useful preprocessors include
287.Xr cpp 1
288and
289.Xr m4 1 .
290If
291.Ar preproc
292does not start with a slash
293.Pq Ql /
294as its first character, the usual
295.Ev PATH
296name search is performed.
297Care should be taken with this in environments where not all
298file systems are mounted (yet) by the time
299.Nm
300is being run (e.g.\& when they are mounted over NFS).
301Once
302.Fl p
303has been specified, any additional arguments as passed on to the preprocessor
304for interpretation.
305This allows for flexible configuration files (like conditionalizing
306them on the local hostname) and the use of macros to centralize
307frequently required arguments like IP addresses.
308.Pp
309The
310.Nm
311.Cm pipe
312and
313.Cm queue
314commands are used to configure the traffic shaper, as shown in the
315.Sx TRAFFIC SHAPER (DUMMYNET) CONFIGURATION
316Section below.
317.Pp
318If the world and the kernel get out of sync the
319.Nm
320ABI may break, preventing you from being able to add any rules.
321This can
322adversely effect the booting process.
323You can use
324.Nm
325.Cm disable
326.Cm firewall
327to temporarily disable the firewall to regain access to the network,
328allowing you to fix the problem.
329.Sh PACKET FLOW
330A packet is checked against the active ruleset in multiple places
331in the protocol stack, under control of several sysctl variables.
332These places and variables are shown below, and it is important to
333have this picture in mind in order to design a correct ruleset.
334.Bd -literal -offset indent
335       ^    to upper layers    V
336       |                       |
337       +----------->-----------+
338       ^                       V
339 [ip(6)_input]           [ip(6)_output]     net.inet(6).ip(6).fw.enable=1
340       |                       |
341       ^                       V
342 [ether_demux]        [ether_output_frame]  net.link.ether.ipfw=1
343       |                       |
344       +-->--[bdg_forward]-->--+            net.link.bridge.ipfw=1
345       ^                       V
346       |      to devices       |
347.Ed
348.Pp
349As can be noted from the above picture, the number of
350times the same packet goes through the firewall can
351vary between 0 and 4 depending on packet source and
352destination, and system configuration.
353.Pp
354Note that as packets flow through the stack, headers can be
355stripped or added to it, and so they may or may not be available
356for inspection.
357E.g., incoming packets will include the MAC header when
358.Nm
359is invoked from
360.Cm ether_demux() ,
361but the same packets will have the MAC header stripped off when
362.Nm
363is invoked from
364.Cm ip_input()
365or
366.Cm ip6_input() .
367.Pp
368Also note that each packet is always checked against the complete ruleset,
369irrespective of the place where the check occurs, or the source of the packet.
370If a rule contains some match patterns or actions which are not valid
371for the place of invocation (e.g.\& trying to match a MAC header within
372.Cm ip_input
373or
374.Cm ip6_input ),
375the match pattern will not match, but a
376.Cm not
377operator in front of such patterns
378.Em will
379cause the pattern to
380.Em always
381match on those packets.
382It is thus the responsibility of
383the programmer, if necessary, to write a suitable ruleset to
384differentiate among the possible places.
385.Cm skipto
386rules can be useful here, as an example:
387.Bd -literal -offset indent
388# packets from ether_demux or bdg_forward
389ipfw add 10 skipto 1000 all from any to any layer2 in
390# packets from ip_input
391ipfw add 10 skipto 2000 all from any to any not layer2 in
392# packets from ip_output
393ipfw add 10 skipto 3000 all from any to any not layer2 out
394# packets from ether_output_frame
395ipfw add 10 skipto 4000 all from any to any layer2 out
396.Ed
397.Pp
398(yes, at the moment there is no way to differentiate between
399ether_demux and bdg_forward).
400.Sh SYNTAX
401In general, each keyword or argument must be provided as
402a separate command line argument, with no leading or trailing
403spaces.
404Keywords are case-sensitive, whereas arguments may
405or may not be case-sensitive depending on their nature
406(e.g.\& uid's are, hostnames are not).
407.Pp
408In
409.Nm ipfw2
410you can introduce spaces after commas ',' to make
411the line more readable.
412You can also put the entire
413command (including flags) into a single argument.
414E.g., the following forms are equivalent:
415.Bd -literal -offset indent
416ipfw -q add deny src-ip 10.0.0.0/24,127.0.0.1/8
417ipfw -q add deny src-ip 10.0.0.0/24, 127.0.0.1/8
418ipfw "-q add deny src-ip 10.0.0.0/24, 127.0.0.1/8"
419.Ed
420.Sh RULE FORMAT
421The format of
422.Nm
423rules is the following:
424.Bd -ragged -offset indent
425.Bk -words
426.Op Ar rule_number
427.Op Cm set Ar set_number
428.Op Cm prob Ar match_probability
429.Ar action
430.Op Cm log Op Cm logamount Ar number
431.Op Cm altq Ar queue
432.Oo
433.Bro Cm tag | untag
434.Brc Ar number
435.Oc
436.Ar body
437.Ek
438.Ed
439.Pp
440where the body of the rule specifies which information is used
441for filtering packets, among the following:
442.Pp
443.Bl -tag -width "Source and dest. addresses and ports" -offset XXX -compact
444.It Layer-2 header fields
445When available
446.It IPv4 and IPv6 Protocol
447TCP, UDP, ICMP, etc.
448.It Source and dest. addresses and ports
449.It Direction
450See Section
451.Sx PACKET FLOW
452.It Transmit and receive interface
453By name or address
454.It Misc. IP header fields
455Version, type of service, datagram length, identification,
456fragment flag (non-zero IP offset),
457Time To Live
458.It IP options
459.It IPv6 Extension headers
460Fragmentation, Hop-by-Hop options,
461Routing Headers, Source routing rthdr0, Mobile IPv6 rthdr2, IPSec options.
462.It IPv6 Flow-ID
463.It Misc. TCP header fields
464TCP flags (SYN, FIN, ACK, RST, etc.),
465sequence number, acknowledgment number,
466window
467.It TCP options
468.It ICMP types
469for ICMP packets
470.It ICMP6 types
471for ICMP6 packets
472.It User/group ID
473When the packet can be associated with a local socket.
474.It Divert status
475Whether a packet came from a divert socket (e.g.,
476.Xr natd 8 ) .
477.El
478.Pp
479Note that some of the above information, e.g.\& source MAC or IP addresses and
480TCP/UDP ports, could easily be spoofed, so filtering on those fields
481alone might not guarantee the desired results.
482.Bl -tag -width indent
483.It Ar rule_number
484Each rule is associated with a
485.Ar rule_number
486in the range 1..65535, with the latter reserved for the
487.Em default
488rule.
489Rules are checked sequentially by rule number.
490Multiple rules can have the same number, in which case they are
491checked (and listed) according to the order in which they have
492been added.
493If a rule is entered without specifying a number, the kernel will
494assign one in such a way that the rule becomes the last one
495before the
496.Em default
497rule.
498Automatic rule numbers are assigned by incrementing the last
499non-default rule number by the value of the sysctl variable
500.Ar net.inet.ip.fw.autoinc_step
501which defaults to 100.
502If this is not possible (e.g.\& because we would go beyond the
503maximum allowed rule number), the number of the last
504non-default value is used instead.
505.It Cm set Ar set_number
506Each rule is associated with a
507.Ar set_number
508in the range 0..31.
509Sets can be individually disabled and enabled, so this parameter
510is of fundamental importance for atomic ruleset manipulation.
511It can be also used to simplify deletion of groups of rules.
512If a rule is entered without specifying a set number,
513set 0 will be used.
514.br
515Set 31 is special in that it cannot be disabled,
516and rules in set 31 are not deleted by the
517.Nm ipfw flush
518command (but you can delete them with the
519.Nm ipfw delete set 31
520command).
521Set 31 is also used for the
522.Em default
523rule.
524.It Cm prob Ar match_probability
525A match is only declared with the specified probability
526(floating point number between 0 and 1).
527This can be useful for a number of applications such as
528random packet drop or
529(in conjunction with
530.Xr dummynet 4 )
531to simulate the effect of multiple paths leading to out-of-order
532packet delivery.
533.Pp
534Note: this condition is checked before any other condition, including
535ones such as keep-state or check-state which might have side effects.
536.It Cm log Op Cm logamount Ar number
537When a packet matches a rule with the
538.Cm log
539keyword, a message will be
540logged to
541.Xr syslogd 8
542with a
543.Dv LOG_SECURITY
544facility.
545The logging only occurs if the sysctl variable
546.Em net.inet.ip.fw.verbose
547is set to 1
548(which is the default when the kernel is compiled with
549.Dv IPFIREWALL_VERBOSE )
550and the number of packets logged so far for that
551particular rule does not exceed the
552.Cm logamount
553parameter.
554If no
555.Cm logamount
556is specified, the limit is taken from the sysctl variable
557.Em net.inet.ip.fw.verbose_limit .
558In both cases, a value of 0 removes the logging limit.
559.Pp
560Once the limit is reached, logging can be re-enabled by
561clearing the logging counter or the packet counter for that entry, see the
562.Cm resetlog
563command.
564.Pp
565Note: logging is done after all other packet matching conditions
566have been successfully verified, and before performing the final
567action (accept, deny, etc.) on the packet.
568.It Cm tag Ar number
569When a packet matches a rule with the
570.Cm tag
571keyword, the numeric tag for the given
572.Ar number
573in the range 1..65534 will be attached to the packet.
574The tag acts as an internal marker (it is not sent out over
575the wire) that can be used to identify these packets later on.
576This can be used, for example, to provide trust between interfaces
577and to start doing policy-based filtering.
578A packet can have mutiple tags at the same time.
579Tags are "sticky", meaning once a tag is applied to a packet by a
580matching rule it exists until explicit removal.
581Tags are kept with the packet everywhere within the kernel, but are
582lost when packet leaves the kernel, for example, on transmitting
583packet out to the network or sending packet to a
584.Xr divert 4
585socket.
586.Pp
587To check for previously applied tags, use the
588.Cm tagged
589rule option.
590To delete previously applied tag, use the
591.Cm untag
592keyword.
593.Pp
594Note: since tags are kept with the packet everywhere in kernelspace,
595they can be set and unset anywhere in kernel network subsystem
596(using
597.Xr mbuf_tags 9
598facility), not only by means of
599.Xr ipfw 4
600.Cm tag
601and
602.Cm untag
603keywords.
604For example, there can be a specialized
605.Xr netgraph 4
606node doing traffic analyzing and tagging for later inspecting
607in firewall.
608.It Cm untag Ar number
609When a packet matches a rule with the
610.Cm untag
611keyword, the tag with the number
612.Ar number
613is searched among the tags attached to this packet and,
614if found, removed from it.
615Other tags bound to packet, if present, are left untouched.
616.It Cm altq Ar queue
617When a packet matches a rule with the
618.Cm altq
619keyword, the ALTQ identifier for the given
620.Ar queue
621(see
622.Xr altq 4 )
623will be attached.
624Note that this ALTQ tag is only meaningful for packets going "out" of IPFW,
625and not being rejected or going to divert sockets.
626Note that if there is insufficient memory at the time the packet is
627processed, it will not be tagged, so it is wise to make your ALTQ
628"default" queue policy account for this.
629If multiple
630.Cm altq
631rules match a single packet, only the first one adds the ALTQ classification
632tag.
633In doing so, traffic may be shaped by using
634.Cm count Cm altq Ar queue
635rules for classification early in the ruleset, then later applying
636the filtering decision.
637For example,
638.Cm check-state
639and
640.Cm keep-state
641rules may come later and provide the actual filtering decisions in
642addition to the fallback ALTQ tag.
643.Pp
644You must run
645.Xr pfctl 8
646to set up the queues before IPFW will be able to look them up by name,
647and if the ALTQ disciplines are rearranged, the rules in containing the
648queue identifiers in the kernel will likely have gone stale and need
649to be reloaded.
650Stale queue identifiers will probably result in misclassification.
651.Pp
652All system ALTQ processing can be turned on or off via
653.Nm
654.Cm enable Ar altq
655and
656.Nm
657.Cm disable Ar altq .
658The usage of
659.Em net.inet.ip.fw.one_pass
660is irrelevant to ALTQ traffic shaping, as the actual rule action is followed
661always after adding an ALTQ tag.
662.El
663.Ss RULE ACTIONS
664A rule can be associated with one of the following actions, which
665will be executed when the packet matches the body of the rule.
666.Bl -tag -width indent
667.It Cm allow | accept | pass | permit
668Allow packets that match rule.
669The search terminates.
670.It Cm check-state
671Checks the packet against the dynamic ruleset.
672If a match is found, execute the action associated with
673the rule which generated this dynamic rule, otherwise
674move to the next rule.
675.br
676.Cm Check-state
677rules do not have a body.
678If no
679.Cm check-state
680rule is found, the dynamic ruleset is checked at the first
681.Cm keep-state
682or
683.Cm limit
684rule.
685.It Cm count
686Update counters for all packets that match rule.
687The search continues with the next rule.
688.It Cm deny | drop
689Discard packets that match this rule.
690The search terminates.
691.It Cm divert Ar port
692Divert packets that match this rule to the
693.Xr divert 4
694socket bound to port
695.Ar port .
696The search terminates.
697.It Cm fwd | forward Ar ipaddr | tablearg Ns Op , Ns Ar port
698Change the next-hop on matching packets to
699.Ar ipaddr ,
700which can be an IP address or a host name.
701The next hop can also be supplied by the last table
702looked up for the packet by using the
703.Cm tablearg
704keyword instead of an explicit address.
705The search terminates if this rule matches.
706.Pp
707If
708.Ar ipaddr
709is a local address, then matching packets will be forwarded to
710.Ar port
711(or the port number in the packet if one is not specified in the rule)
712on the local machine.
713.br
714If
715.Ar ipaddr
716is not a local address, then the port number
717(if specified) is ignored, and the packet will be
718forwarded to the remote address, using the route as found in
719the local routing table for that IP.
720.br
721A
722.Ar fwd
723rule will not match layer-2 packets (those received
724on ether_input, ether_output, or bridged).
725.br
726The
727.Cm fwd
728action does not change the contents of the packet at all.
729In particular, the destination address remains unmodified, so
730packets forwarded to another system will usually be rejected by that system
731unless there is a matching rule on that system to capture them.
732For packets forwarded locally,
733the local address of the socket will be
734set to the original destination address of the packet.
735This makes the
736.Xr netstat 1
737entry look rather weird but is intended for
738use with transparent proxy servers.
739.Pp
740To enable
741.Cm fwd
742a custom kernel needs to be compiled with the option
743.Cd "options IPFIREWALL_FORWARD" .
744.It Cm nat Ar nat_nr
745Pass packet to a
746nat instance
747(for network address translation, address redirect, etc.):
748see the
749.Sx NETWORK ADDRESS TRANSLATION (NAT)
750Section for further information.
751.It Cm pipe Ar pipe_nr
752Pass packet to a
753.Xr dummynet 4
754.Dq pipe
755(for bandwidth limitation, delay, etc.).
756See the
757.Sx TRAFFIC SHAPER (DUMMYNET) CONFIGURATION
758Section for further information.
759The search terminates; however, on exit from the pipe and if
760the
761.Xr sysctl 8
762variable
763.Em net.inet.ip.fw.one_pass
764is not set, the packet is passed again to the firewall code
765starting from the next rule.
766.It Cm queue Ar queue_nr
767Pass packet to a
768.Xr dummynet 4
769.Dq queue
770(for bandwidth limitation using WF2Q+).
771.It Cm reject
772(Deprecated).
773Synonym for
774.Cm unreach host .
775.It Cm reset
776Discard packets that match this rule, and if the
777packet is a TCP packet, try to send a TCP reset (RST) notice.
778The search terminates.
779.It Cm reset6
780Discard packets that match this rule, and if the
781packet is a TCP packet, try to send a TCP reset (RST) notice.
782The search terminates.
783.It Cm skipto Ar number
784Skip all subsequent rules numbered less than
785.Ar number .
786The search continues with the first rule numbered
787.Ar number
788or higher.
789.It Cm tee Ar port
790Send a copy of packets matching this rule to the
791.Xr divert 4
792socket bound to port
793.Ar port .
794The search continues with the next rule.
795.It Cm unreach Ar code
796Discard packets that match this rule, and try to send an ICMP
797unreachable notice with code
798.Ar code ,
799where
800.Ar code
801is a number from 0 to 255, or one of these aliases:
802.Cm net , host , protocol , port ,
803.Cm needfrag , srcfail , net-unknown , host-unknown ,
804.Cm isolated , net-prohib , host-prohib , tosnet ,
805.Cm toshost , filter-prohib , host-precedence
806or
807.Cm precedence-cutoff .
808The search terminates.
809.It Cm unreach6 Ar code
810Discard packets that match this rule, and try to send an ICMPv6
811unreachable notice with code
812.Ar code ,
813where
814.Ar code
815is a number from 0, 1, 3 or 4, or one of these aliases:
816.Cm no-route, admin-prohib, address
817or
818.Cm port .
819The search terminates.
820.It Cm netgraph Ar cookie
821Divert packet into netgraph with given
822.Ar cookie .
823The search terminates.
824If packet is later returned from netgraph it is either
825accepted or continues with the next rule, depending on
826.Em net.inet.ip.fw.one_pass
827sysctl variable.
828.It Cm ngtee Ar cookie
829A copy of packet is diverted into netgraph, original
830packet is either accepted or continues with the next rule, depending on
831.Em net.inet.ip.fw.one_pass
832sysctl variable.
833See
834.Xr ng_ipfw 4
835for more information on
836.Cm netgraph
837and
838.Cm ngtee
839actions.
840.El
841.Ss RULE BODY
842The body of a rule contains zero or more patterns (such as
843specific source and destination addresses or ports,
844protocol options, incoming or outgoing interfaces, etc.)
845that the packet must match in order to be recognised.
846In general, the patterns are connected by (implicit)
847.Cm and
848operators -- i.e., all must match in order for the
849rule to match.
850Individual patterns can be prefixed by the
851.Cm not
852operator to reverse the result of the match, as in
853.Pp
854.Dl "ipfw add 100 allow ip from not 1.2.3.4 to any"
855.Pp
856Additionally, sets of alternative match patterns
857.Pq Em or-blocks
858can be constructed by putting the patterns in
859lists enclosed between parentheses ( ) or braces { }, and
860using the
861.Cm or
862operator as follows:
863.Pp
864.Dl "ipfw add 100 allow ip from { x or not y or z } to any"
865.Pp
866Only one level of parentheses is allowed.
867Beware that most shells have special meanings for parentheses
868or braces, so it is advisable to put a backslash \\ in front of them
869to prevent such interpretations.
870.Pp
871The body of a rule must in general include a source and destination
872address specifier.
873The keyword
874.Ar any
875can be used in various places to specify that the content of
876a required field is irrelevant.
877.Pp
878The rule body has the following format:
879.Bd -ragged -offset indent
880.Op Ar proto Cm from Ar src Cm to Ar dst
881.Op Ar options
882.Ed
883.Pp
884The first part (proto from src to dst) is for backward
885compatibility with earlier versions of
886.Fx .
887In modern
888.Fx
889any match pattern (including MAC headers, IP protocols,
890addresses and ports) can be specified in the
891.Ar options
892section.
893.Pp
894Rule fields have the following meaning:
895.Bl -tag -width indent
896.It Ar proto : protocol | Cm { Ar protocol Cm or ... }
897.It Ar protocol : Oo Cm not Oc Ar protocol-name | protocol-number
898An IP protocol specified by number or name
899(for a complete list see
900.Pa /etc/protocols ) ,
901or one of the following keywords:
902.Bl -tag -width indent
903.It Cm ip4 | ipv4
904Matches IPv4 packets.
905.It Cm ip6 | ipv6
906Matches IPv6 packets.
907.It Cm ip | all
908Matches any packet.
909.El
910.Pp
911The
912.Cm ipv6
913in
914.Cm proto
915option will be treated as inner protocol.
916And, the
917.Cm ipv4
918is not available in
919.Cm proto
920option.
921.Pp
922The
923.Cm { Ar protocol Cm or ... }
924format (an
925.Em or-block )
926is provided for convenience only but its use is deprecated.
927.It Ar src No and Ar dst : Bro Cm addr | Cm { Ar addr Cm or ... } Brc Op Oo Cm not Oc Ar ports
928An address (or a list, see below)
929optionally followed by
930.Ar ports
931specifiers.
932.Pp
933The second format
934.Em ( or-block
935with multiple addresses) is provided for convenience only and
936its use is discouraged.
937.It Ar addr : Oo Cm not Oc Bro
938.Cm any | me | me6 |
939.Cm table Ns Pq Ar number Ns Op , Ns Ar value
940.Ar | addr-list | addr-set
941.Brc
942.It Cm any
943matches any IP address.
944.It Cm me
945matches any IP address configured on an interface in the system.
946.It Cm me6
947matches any IPv6 address configured on an interface in the system.
948The address list is evaluated at the time the packet is
949analysed.
950.It Cm table Ns Pq Ar number Ns Op , Ns Ar value
951Matches any IPv4 address for which an entry exists in the lookup table
952.Ar number .
953If an optional 32-bit unsigned
954.Ar value
955is also specified, an entry will match only if it has this value.
956See the
957.Sx LOOKUP TABLES
958section below for more information on lookup tables.
959.It Ar addr-list : ip-addr Ns Op Ns , Ns Ar addr-list
960.It Ar ip-addr :
961A host or subnet address specified in one of the following ways:
962.Bl -tag -width indent
963.It Ar numeric-ip | hostname
964Matches a single IPv4 address, specified as dotted-quad or a hostname.
965Hostnames are resolved at the time the rule is added to the firewall list.
966.It Ar addr Ns / Ns Ar masklen
967Matches all addresses with base
968.Ar addr
969(specified as an IP address, a network number, or a hostname)
970and mask width of
971.Cm masklen
972bits.
973As an example, 1.2.3.4/25 or 1.2.3.0/25 will match
974all IP numbers from 1.2.3.0 to 1.2.3.127 .
975.It Ar addr Ns : Ns Ar mask
976Matches all addresses with base
977.Ar addr
978(specified as an IP address, a network number, or a hostname)
979and the mask of
980.Ar mask ,
981specified as a dotted quad.
982As an example, 1.2.3.4:255.0.255.0 or 1.0.3.0:255.0.255.0 will match
9831.*.3.*.
984This form is advised only for non-contiguous
985masks.
986It is better to resort to the
987.Ar addr Ns / Ns Ar masklen
988format for contiguous masks, which is more compact and less
989error-prone.
990.El
991.It Ar addr-set : addr Ns Oo Ns / Ns Ar masklen Oc Ns Cm { Ns Ar list Ns Cm }
992.It Ar list : Bro Ar num | num-num Brc Ns Op Ns , Ns Ar list
993Matches all addresses with base address
994.Ar addr
995(specified as an IP address, a network number, or a hostname)
996and whose last byte is in the list between braces { } .
997Note that there must be no spaces between braces and
998numbers (spaces after commas are allowed).
999Elements of the list can be specified as single entries
1000or ranges.
1001The
1002.Ar masklen
1003field is used to limit the size of the set of addresses,
1004and can have any value between 24 and 32.
1005If not specified,
1006it will be assumed as 24.
1007.br
1008This format is particularly useful to handle sparse address sets
1009within a single rule.
1010Because the matching occurs using a
1011bitmask, it takes constant time and dramatically reduces
1012the complexity of rulesets.
1013.br
1014As an example, an address specified as 1.2.3.4/24{128,35-55,89}
1015or 1.2.3.0/24{128,35-55,89}
1016will match the following IP addresses:
1017.br
10181.2.3.128, 1.2.3.35 to 1.2.3.55, 1.2.3.89 .
1019.It Ar addr6-list : ip6-addr Ns Op Ns , Ns Ar addr6-list
1020.It Ar ip6-addr :
1021A host or subnet specified one of the following ways:
1022.Pp
1023.Bl -tag -width indent
1024.It Ar numeric-ip | hostname
1025Matches a single IPv6 address as allowed by
1026.Xr inet_pton 3
1027or a hostname.
1028Hostnames are resolved at the time the rule is added to the firewall
1029list.
1030.It Ar addr Ns / Ns Ar masklen
1031Matches all IPv6 addresses with base
1032.Ar addr
1033(specified as allowed by
1034.Xr inet_pton
1035or a hostname)
1036and mask width of
1037.Cm masklen
1038bits.
1039.El
1040.Pp
1041No support for sets of IPv6 addresses is provided because IPv6 addresses
1042are typically random past the initial prefix.
1043.It Ar ports : Bro Ar port | port Ns \&- Ns Ar port Ns Brc Ns Op , Ns Ar ports
1044For protocols which support port numbers (such as TCP and UDP), optional
1045.Cm ports
1046may be specified as one or more ports or port ranges, separated
1047by commas but no spaces, and an optional
1048.Cm not
1049operator.
1050The
1051.Ql \&-
1052notation specifies a range of ports (including boundaries).
1053.Pp
1054Service names (from
1055.Pa /etc/services )
1056may be used instead of numeric port values.
1057The length of the port list is limited to 30 ports or ranges,
1058though one can specify larger ranges by using an
1059.Em or-block
1060in the
1061.Cm options
1062section of the rule.
1063.Pp
1064A backslash
1065.Pq Ql \e
1066can be used to escape the dash
1067.Pq Ql -
1068character in a service name (from a shell, the backslash must be
1069typed twice to avoid the shell itself interpreting it as an escape
1070character).
1071.Pp
1072.Dl "ipfw add count tcp from any ftp\e\e-data-ftp to any"
1073.Pp
1074Fragmented packets which have a non-zero offset (i.e., not the first
1075fragment) will never match a rule which has one or more port
1076specifications.
1077See the
1078.Cm frag
1079option for details on matching fragmented packets.
1080.El
1081.Ss RULE OPTIONS (MATCH PATTERNS)
1082Additional match patterns can be used within
1083rules.
1084Zero or more of these so-called
1085.Em options
1086can be present in a rule, optionally prefixed by the
1087.Cm not
1088operand, and possibly grouped into
1089.Em or-blocks .
1090.Pp
1091The following match patterns can be used (listed in alphabetical order):
1092.Bl -tag -width indent
1093.It Cm // this is a comment.
1094Inserts the specified text as a comment in the rule.
1095Everything following // is considered as a comment and stored in the rule.
1096You can have comment-only rules, which are listed as having a
1097.Cm count
1098action followed by the comment.
1099.It Cm bridged
1100Alias for
1101.Cm layer2 .
1102.It Cm diverted
1103Matches only packets generated by a divert socket.
1104.It Cm diverted-loopback
1105Matches only packets coming from a divert socket back into the IP stack
1106input for delivery.
1107.It Cm diverted-output
1108Matches only packets going from a divert socket back outward to the IP
1109stack output for delivery.
1110.It Cm dst-ip Ar ip-address
1111Matches IPv4 packets whose destination IP is one of the address(es)
1112specified as argument.
1113.It Bro Cm dst-ip6 | dst-ipv6 Brc Ar ip6-address
1114Matches IPv6 packets whose destination IP is one of the address(es)
1115specified as argument.
1116.It Cm dst-port Ar ports
1117Matches IP packets whose destination port is one of the port(s)
1118specified as argument.
1119.It Cm established
1120Matches TCP packets that have the RST or ACK bits set.
1121.It Cm ext6hdr Ar header
1122Matches IPv6 packets containing the extended header given by
1123.Ar header .
1124Supported headers are:
1125.Pp
1126Fragment,
1127.Pq Cm frag ,
1128Hop-to-hop options
1129.Pq Cm hopopt ,
1130any type of Routing Header
1131.Pq Cm route ,
1132Source routing Routing Header Type 0
1133.Pq Cm rthdr0 ,
1134Mobile IPv6 Routing Header Type 2
1135.Pq Cm rthdr2 ,
1136Destination options
1137.Pq Cm dstopt ,
1138IPSec authentication headers
1139.Pq Cm ah ,
1140and IPSec encapsulated security payload headers
1141.Pq Cm esp .
1142.It Cm flow-id Ar labels
1143Matches IPv6 packets containing any of the flow labels given in
1144.Ar labels .
1145.Ar labels
1146is a comma seperate list of numeric flow labels.
1147.It Cm frag
1148Matches packets that are fragments and not the first
1149fragment of an IP datagram.
1150Note that these packets will not have
1151the next protocol header (e.g.\& TCP, UDP) so options that look into
1152these headers cannot match.
1153.It Cm gid Ar group
1154Matches all TCP or UDP packets sent by or received for a
1155.Ar group .
1156A
1157.Ar group
1158may be specified by name or number.
1159.It Cm jail Ar prisonID
1160Matches all TCP or UDP packets sent by or received for the
1161jail whos prison ID is
1162.Ar prisonID .
1163.It Cm icmptypes Ar types
1164Matches ICMP packets whose ICMP type is in the list
1165.Ar types .
1166The list may be specified as any combination of
1167individual types (numeric) separated by commas.
1168.Em Ranges are not allowed.
1169The supported ICMP types are:
1170.Pp
1171echo reply
1172.Pq Cm 0 ,
1173destination unreachable
1174.Pq Cm 3 ,
1175source quench
1176.Pq Cm 4 ,
1177redirect
1178.Pq Cm 5 ,
1179echo request
1180.Pq Cm 8 ,
1181router advertisement
1182.Pq Cm 9 ,
1183router solicitation
1184.Pq Cm 10 ,
1185time-to-live exceeded
1186.Pq Cm 11 ,
1187IP header bad
1188.Pq Cm 12 ,
1189timestamp request
1190.Pq Cm 13 ,
1191timestamp reply
1192.Pq Cm 14 ,
1193information request
1194.Pq Cm 15 ,
1195information reply
1196.Pq Cm 16 ,
1197address mask request
1198.Pq Cm 17
1199and address mask reply
1200.Pq Cm 18 .
1201.It Cm icmp6types Ar types
1202Matches ICMP6 packets whose ICMP6 type is in the list of
1203.Ar types .
1204The list may be specified as any combination of
1205individual types (numeric) separated by commas.
1206.Em Ranges are not allowed.
1207.It Cm in | out
1208Matches incoming or outgoing packets, respectively.
1209.Cm in
1210and
1211.Cm out
1212are mutually exclusive (in fact,
1213.Cm out
1214is implemented as
1215.Cm not in Ns No ).
1216.It Cm ipid Ar id-list
1217Matches IPv4 packets whose
1218.Cm ip_id
1219field has value included in
1220.Ar id-list ,
1221which is either a single value or a list of values or ranges
1222specified in the same way as
1223.Ar ports .
1224.It Cm iplen Ar len-list
1225Matches IP packets whose total length, including header and data, is
1226in the set
1227.Ar len-list ,
1228which is either a single value or a list of values or ranges
1229specified in the same way as
1230.Ar ports .
1231.It Cm ipoptions Ar spec
1232Matches packets whose IPv4 header contains the comma separated list of
1233options specified in
1234.Ar spec .
1235The supported IP options are:
1236.Pp
1237.Cm ssrr
1238(strict source route),
1239.Cm lsrr
1240(loose source route),
1241.Cm rr
1242(record packet route) and
1243.Cm ts
1244(timestamp).
1245The absence of a particular option may be denoted
1246with a
1247.Ql \&! .
1248.It Cm ipprecedence Ar precedence
1249Matches IPv4 packets whose precedence field is equal to
1250.Ar precedence .
1251.It Cm ipsec
1252Matches packets that have IPSEC history associated with them
1253(i.e., the packet comes encapsulated in IPSEC, the kernel
1254has IPSEC support and IPSEC_FILTERTUNNEL option, and can correctly
1255decapsulate it).
1256.Pp
1257Note that specifying
1258.Cm ipsec
1259is different from specifying
1260.Cm proto Ar ipsec
1261as the latter will only look at the specific IP protocol field,
1262irrespective of IPSEC kernel support and the validity of the IPSEC data.
1263.Pp
1264Further note that this flag is silently ignored in kernels without
1265IPSEC support.
1266It does not affect rule processing when given and the
1267rules are handled as if with no
1268.Cm ipsec
1269flag.
1270.It Cm iptos Ar spec
1271Matches IPv4 packets whose
1272.Cm tos
1273field contains the comma separated list of
1274service types specified in
1275.Ar spec .
1276The supported IP types of service are:
1277.Pp
1278.Cm lowdelay
1279.Pq Dv IPTOS_LOWDELAY ,
1280.Cm throughput
1281.Pq Dv IPTOS_THROUGHPUT ,
1282.Cm reliability
1283.Pq Dv IPTOS_RELIABILITY ,
1284.Cm mincost
1285.Pq Dv IPTOS_MINCOST ,
1286.Cm congestion
1287.Pq Dv IPTOS_ECN_CE .
1288The absence of a particular type may be denoted
1289with a
1290.Ql \&! .
1291.It Cm ipttl Ar ttl-list
1292Matches IPv4 packets whose time to live is included in
1293.Ar ttl-list ,
1294which is either a single value or a list of values or ranges
1295specified in the same way as
1296.Ar ports .
1297.It Cm ipversion Ar ver
1298Matches IP packets whose IP version field is
1299.Ar ver .
1300.It Cm keep-state
1301Upon a match, the firewall will create a dynamic rule, whose
1302default behaviour is to match bidirectional traffic between
1303source and destination IP/port using the same protocol.
1304The rule has a limited lifetime (controlled by a set of
1305.Xr sysctl 8
1306variables), and the lifetime is refreshed every time a matching
1307packet is found.
1308.It Cm layer2
1309Matches only layer2 packets, i.e., those passed to
1310.Nm
1311from ether_demux() and ether_output_frame().
1312.It Cm limit Bro Cm src-addr | src-port | dst-addr | dst-port Brc Ar N
1313The firewall will only allow
1314.Ar N
1315connections with the same
1316set of parameters as specified in the rule.
1317One or more
1318of source and destination addresses and ports can be
1319specified.
1320Currently,
1321only IPv4 flows are supported.
1322.It Cm { MAC | mac } Ar dst-mac src-mac
1323Match packets with a given
1324.Ar dst-mac
1325and
1326.Ar src-mac
1327addresses, specified as the
1328.Cm any
1329keyword (matching any MAC address), or six groups of hex digits
1330separated by colons,
1331and optionally followed by a mask indicating the significant bits.
1332The mask may be specified using either of the following methods:
1333.Bl -enum -width indent
1334.It
1335A slash
1336.Pq /
1337followed by the number of significant bits.
1338For example, an address with 33 significant bits could be specified as:
1339.Pp
1340.Dl "MAC 10:20:30:40:50:60/33 any"
1341.Pp
1342.It
1343An ampersand
1344.Pq &
1345followed by a bitmask specified as six groups of hex digits separated
1346by colons.
1347For example, an address in which the last 16 bits are significant could
1348be specified as:
1349.Pp
1350.Dl "MAC 10:20:30:40:50:60&00:00:00:00:ff:ff any"
1351.Pp
1352Note that the ampersand character has a special meaning in many shells
1353and should generally be escaped.
1354.Pp
1355.El
1356Note that the order of MAC addresses (destination first,
1357source second) is
1358the same as on the wire, but the opposite of the one used for
1359IP addresses.
1360.It Cm mac-type Ar mac-type
1361Matches packets whose Ethernet Type field
1362corresponds to one of those specified as argument.
1363.Ar mac-type
1364is specified in the same way as
1365.Cm port numbers
1366(i.e., one or more comma-separated single values or ranges).
1367You can use symbolic names for known values such as
1368.Em vlan , ipv4, ipv6 .
1369Values can be entered as decimal or hexadecimal (if prefixed by 0x),
1370and they are always printed as hexadecimal (unless the
1371.Cm -N
1372option is used, in which case symbolic resolution will be attempted).
1373.It Cm proto Ar protocol
1374Matches packets with the corresponding IP protocol.
1375.It Cm recv | xmit | via Brq Ar ifX | Ar if Ns Cm * | Ar ipno | Ar any
1376Matches packets received, transmitted or going through,
1377respectively, the interface specified by exact name
1378.Ns No ( Ar ifX Ns No ),
1379by device name
1380.Ns No ( Ar if Ns Ar * Ns No ),
1381by IP address, or through some interface.
1382.Pp
1383The
1384.Cm via
1385keyword causes the interface to always be checked.
1386If
1387.Cm recv
1388or
1389.Cm xmit
1390is used instead of
1391.Cm via ,
1392then only the receive or transmit interface (respectively)
1393is checked.
1394By specifying both, it is possible to match packets based on
1395both receive and transmit interface, e.g.:
1396.Pp
1397.Dl "ipfw add deny ip from any to any out recv ed0 xmit ed1"
1398.Pp
1399The
1400.Cm recv
1401interface can be tested on either incoming or outgoing packets,
1402while the
1403.Cm xmit
1404interface can only be tested on outgoing packets.
1405So
1406.Cm out
1407is required (and
1408.Cm in
1409is invalid) whenever
1410.Cm xmit
1411is used.
1412.Pp
1413A packet may not have a receive or transmit interface: packets
1414originating from the local host have no receive interface,
1415while packets destined for the local host have no transmit
1416interface.
1417.It Cm setup
1418Matches TCP packets that have the SYN bit set but no ACK bit.
1419This is the short form of
1420.Dq Li tcpflags\ syn,!ack .
1421.It Cm src-ip Ar ip-address
1422Matches IPv4 packets whose source IP is one of the address(es)
1423specified as an argument.
1424.It Cm src-ip6 Ar ip6-address
1425Matches IPv6 packets whose source IP is one of the address(es)
1426specified as an argument.
1427.It Cm src-port Ar ports
1428Matches IP packets whose source port is one of the port(s)
1429specified as argument.
1430.It Cm tagged Ar tag-list
1431Matches packets whose tags are included in
1432.Ar tag-list ,
1433which is either a single value or a list of values or ranges
1434specified in the same way as
1435.Ar ports .
1436Tags can be applied to the packet using
1437.Cm tag
1438rule action parameter (see it's description for details on tags).
1439.It Cm tcpack Ar ack
1440TCP packets only.
1441Match if the TCP header acknowledgment number field is set to
1442.Ar ack .
1443.It Cm tcpdatalen Ar tcpdatalen-list
1444Matches TCP packets whose length of TCP data is
1445.Ar tcpdatalen-list ,
1446which is either a single value or a list of values or ranges
1447specified in the same way as
1448.Ar ports .
1449.It Cm tcpflags Ar spec
1450TCP packets only.
1451Match if the TCP header contains the comma separated list of
1452flags specified in
1453.Ar spec .
1454The supported TCP flags are:
1455.Pp
1456.Cm fin ,
1457.Cm syn ,
1458.Cm rst ,
1459.Cm psh ,
1460.Cm ack
1461and
1462.Cm urg .
1463The absence of a particular flag may be denoted
1464with a
1465.Ql \&! .
1466A rule which contains a
1467.Cm tcpflags
1468specification can never match a fragmented packet which has
1469a non-zero offset.
1470See the
1471.Cm frag
1472option for details on matching fragmented packets.
1473.It Cm tcpseq Ar seq
1474TCP packets only.
1475Match if the TCP header sequence number field is set to
1476.Ar seq .
1477.It Cm tcpwin Ar win
1478TCP packets only.
1479Match if the TCP header window field is set to
1480.Ar win .
1481.It Cm tcpoptions Ar spec
1482TCP packets only.
1483Match if the TCP header contains the comma separated list of
1484options specified in
1485.Ar spec .
1486The supported TCP options are:
1487.Pp
1488.Cm mss
1489(maximum segment size),
1490.Cm window
1491(tcp window advertisement),
1492.Cm sack
1493(selective ack),
1494.Cm ts
1495(rfc1323 timestamp) and
1496.Cm cc
1497(rfc1644 t/tcp connection count).
1498The absence of a particular option may be denoted
1499with a
1500.Ql \&! .
1501.It Cm uid Ar user
1502Match all TCP or UDP packets sent by or received for a
1503.Ar user .
1504A
1505.Ar user
1506may be matched by name or identification number.
1507.It Cm verrevpath
1508For incoming packets,
1509a routing table lookup is done on the packet's source address.
1510If the interface on which the packet entered the system matches the
1511outgoing interface for the route,
1512the packet matches.
1513If the interfaces do not match up,
1514the packet does not match.
1515All outgoing packets or packets with no incoming interface match.
1516.Pp
1517The name and functionality of the option is intentionally similar to
1518the Cisco IOS command:
1519.Pp
1520.Dl ip verify unicast reverse-path
1521.Pp
1522This option can be used to make anti-spoofing rules to reject all
1523packets with source addresses not from this interface.
1524See also the option
1525.Cm antispoof .
1526.It Cm versrcreach
1527For incoming packets,
1528a routing table lookup is done on the packet's source address.
1529If a route to the source address exists, but not the default route
1530or a blackhole/reject route, the packet matches.
1531Otherwise, the packet does not match.
1532All outgoing packets match.
1533.Pp
1534The name and functionality of the option is intentionally similar to
1535the Cisco IOS command:
1536.Pp
1537.Dl ip verify unicast source reachable-via any
1538.Pp
1539This option can be used to make anti-spoofing rules to reject all
1540packets whose source address is unreachable.
1541.It Cm antispoof
1542For incoming packets, the packet's source address is checked if it
1543belongs to a directly connected network.
1544If the network is directly connected, then the interface the packet
1545came on in is compared to the interface the network is connected to.
1546When incoming interface and directly connected interface are not the
1547same, the packet does not match.
1548Otherwise, the packet does match.
1549All outgoing packets match.
1550.Pp
1551This option can be used to make anti-spoofing rules to reject all
1552packets that pretend to be from a directly connected network but do
1553not come in through that interface.
1554This option is similar to but more restricted than
1555.Cm verrevpath
1556because it engages only on packets with source addresses of directly
1557connected networks instead of all source addresses.
1558.El
1559.Sh LOOKUP TABLES
1560Lookup tables are useful to handle large sparse address sets,
1561typically from a hundred to several thousands of entries.
1562There may be up to 128 different lookup tables, numbered 0 to 127.
1563.Pp
1564Each entry is represented by an
1565.Ar addr Ns Op / Ns Ar masklen
1566and will match all addresses with base
1567.Ar addr
1568(specified as an IP address or a hostname)
1569and mask width of
1570.Ar masklen
1571bits.
1572If
1573.Ar masklen
1574is not specified, it defaults to 32.
1575When looking up an IP address in a table, the most specific
1576entry will match.
1577Associated with each entry is a 32-bit unsigned
1578.Ar value ,
1579which can optionally be checked by a rule matching code.
1580When adding an entry, if
1581.Ar value
1582is not specified, it defaults to 0.
1583.Pp
1584An entry can be added to a table
1585.Pq Cm add ,
1586removed from a table
1587.Pq Cm delete ,
1588a table can be examined
1589.Pq Cm list
1590or flushed
1591.Pq Cm flush .
1592.Pp
1593Internally, each table is stored in a Radix tree, the same way as
1594the routing table (see
1595.Xr route 4 ) .
1596.Pp
1597Lookup tables currently support IPv4 addresses only.
1598.Pp
1599The
1600.Cm tablearg
1601feature provides the ability to use a value, looked up in the table, as
1602the argument for a rule action, action parameter or rule option.
1603This can significantly reduce number of rules in some configurations.
1604The
1605.Cm tablearg
1606argument can be used with the following actions:
1607.Cm pipe , queue, divert, tee, netgraph, ngtee, fwd
1608action parameters:
1609.Cm tag, untag,
1610rule options:
1611.Cm limit, tagged.
1612.Pp
1613When used with
1614.Cm fwd
1615it is possible to supply table entries with values
1616that are in the form of IP addresses or hostnames.
1617See the
1618.Sx EXAMPLES
1619Section for example usage of tables and the tablearg keyword.
1620.Sh SETS OF RULES
1621Each rule belongs to one of 32 different
1622.Em sets
1623, numbered 0 to 31.
1624Set 31 is reserved for the default rule.
1625.Pp
1626By default, rules are put in set 0, unless you use the
1627.Cm set N
1628attribute when entering a new rule.
1629Sets can be individually and atomically enabled or disabled,
1630so this mechanism permits an easy way to store multiple configurations
1631of the firewall and quickly (and atomically) switch between them.
1632The command to enable/disable sets is
1633.Bd -ragged -offset indent
1634.Nm
1635.Cm set Oo Cm disable Ar number ... Oc Op Cm enable Ar number ...
1636.Ed
1637.Pp
1638where multiple
1639.Cm enable
1640or
1641.Cm disable
1642sections can be specified.
1643Command execution is atomic on all the sets specified in the command.
1644By default, all sets are enabled.
1645.Pp
1646When you disable a set, its rules behave as if they do not exist
1647in the firewall configuration, with only one exception:
1648.Bd -ragged -offset indent
1649dynamic rules created from a rule before it had been disabled
1650will still be active until they expire.
1651In order to delete
1652dynamic rules you have to explicitly delete the parent rule
1653which generated them.
1654.Ed
1655.Pp
1656The set number of rules can be changed with the command
1657.Bd -ragged -offset indent
1658.Nm
1659.Cm set move
1660.Brq Cm rule Ar rule-number | old-set
1661.Cm to Ar new-set
1662.Ed
1663.Pp
1664Also, you can atomically swap two rulesets with the command
1665.Bd -ragged -offset indent
1666.Nm
1667.Cm set swap Ar first-set second-set
1668.Ed
1669.Pp
1670See the
1671.Sx EXAMPLES
1672Section on some possible uses of sets of rules.
1673.Sh STATEFUL FIREWALL
1674Stateful operation is a way for the firewall to dynamically
1675create rules for specific flows when packets that
1676match a given pattern are detected.
1677Support for stateful
1678operation comes through the
1679.Cm check-state , keep-state
1680and
1681.Cm limit
1682options of
1683.Nm rules .
1684.Pp
1685Dynamic rules are created when a packet matches a
1686.Cm keep-state
1687or
1688.Cm limit
1689rule, causing the creation of a
1690.Em dynamic
1691rule which will match all and only packets with
1692a given
1693.Em protocol
1694between a
1695.Em src-ip/src-port dst-ip/dst-port
1696pair of addresses
1697.Em ( src
1698and
1699.Em dst
1700are used here only to denote the initial match addresses, but they
1701are completely equivalent afterwards).
1702Dynamic rules will be checked at the first
1703.Cm check-state, keep-state
1704or
1705.Cm limit
1706occurrence, and the action performed upon a match will be the same
1707as in the parent rule.
1708.Pp
1709Note that no additional attributes other than protocol and IP addresses
1710and ports are checked on dynamic rules.
1711.Pp
1712The typical use of dynamic rules is to keep a closed firewall configuration,
1713but let the first TCP SYN packet from the inside network install a
1714dynamic rule for the flow so that packets belonging to that session
1715will be allowed through the firewall:
1716.Pp
1717.Dl "ipfw add check-state"
1718.Dl "ipfw add allow tcp from my-subnet to any setup keep-state"
1719.Dl "ipfw add deny tcp from any to any"
1720.Pp
1721A similar approach can be used for UDP, where an UDP packet coming
1722from the inside will install a dynamic rule to let the response through
1723the firewall:
1724.Pp
1725.Dl "ipfw add check-state"
1726.Dl "ipfw add allow udp from my-subnet to any keep-state"
1727.Dl "ipfw add deny udp from any to any"
1728.Pp
1729Dynamic rules expire after some time, which depends on the status
1730of the flow and the setting of some
1731.Cm sysctl
1732variables.
1733See Section
1734.Sx SYSCTL VARIABLES
1735for more details.
1736For TCP sessions, dynamic rules can be instructed to periodically
1737send keepalive packets to refresh the state of the rule when it is
1738about to expire.
1739.Pp
1740See Section
1741.Sx EXAMPLES
1742for more examples on how to use dynamic rules.
1743.Sh TRAFFIC SHAPER (DUMMYNET) CONFIGURATION
1744.Nm
1745is also the user interface for the
1746.Xr dummynet 4
1747traffic shaper.
1748.Pp
1749.Nm dummynet
1750operates by first using the firewall to classify packets and divide them into
1751.Em flows ,
1752using any match pattern that can be used in
1753.Nm
1754rules.
1755Depending on local policies, a flow can contain packets for a single
1756TCP connection, or from/to a given host, or entire subnet, or a
1757protocol type, etc.
1758.Pp
1759Packets belonging to the same flow are then passed to either of two
1760different objects, which implement the traffic regulation:
1761.Bl -hang -offset XXXX
1762.It Em pipe
1763A pipe emulates a link with given bandwidth, propagation delay,
1764queue size and packet loss rate.
1765Packets are queued in front of the pipe as they come out from the classifier,
1766and then transferred to the pipe according to the pipe's parameters.
1767.Pp
1768.It Em queue
1769A queue
1770is an abstraction used to implement the WF2Q+
1771(Worst-case Fair Weighted Fair Queueing) policy, which is
1772an efficient variant of the WFQ policy.
1773.br
1774The queue associates a
1775.Em weight
1776and a reference pipe to each flow, and then all backlogged (i.e.,
1777with packets queued) flows linked to the same pipe share the pipe's
1778bandwidth proportionally to their weights.
1779Note that weights are not priorities; a flow with a lower weight
1780is still guaranteed to get its fraction of the bandwidth even if a
1781flow with a higher weight is permanently backlogged.
1782.Pp
1783.El
1784In practice,
1785.Em pipes
1786can be used to set hard limits to the bandwidth that a flow can use, whereas
1787.Em queues
1788can be used to determine how different flow share the available bandwidth.
1789.Pp
1790The
1791.Em pipe
1792and
1793.Em queue
1794configuration commands are the following:
1795.Bd -ragged -offset indent
1796.Cm pipe Ar number Cm config Ar pipe-configuration
1797.Pp
1798.Cm queue Ar number Cm config Ar queue-configuration
1799.Ed
1800.Pp
1801The following parameters can be configured for a pipe:
1802.Pp
1803.Bl -tag -width indent -compact
1804.It Cm bw Ar bandwidth | device
1805Bandwidth, measured in
1806.Sm off
1807.Op Cm K | M
1808.Brq Cm bit/s | Byte/s .
1809.Sm on
1810.Pp
1811A value of 0 (default) means unlimited bandwidth.
1812The unit must immediately follow the number, as in
1813.Pp
1814.Dl "ipfw pipe 1 config bw 300Kbit/s"
1815.Pp
1816If a device name is specified instead of a numeric value, as in
1817.Pp
1818.Dl "ipfw pipe 1 config bw tun0"
1819.Pp
1820then the transmit clock is supplied by the specified device.
1821At the moment only the
1822.Xr tun 4
1823device supports this
1824functionality, for use in conjunction with
1825.Xr ppp 8 .
1826.Pp
1827.It Cm delay Ar ms-delay
1828Propagation delay, measured in milliseconds.
1829The value is rounded to the next multiple of the clock tick
1830(typically 10ms, but it is a good practice to run kernels
1831with
1832.Dq "options HZ=1000"
1833to reduce
1834the granularity to 1ms or less).
1835Default value is 0, meaning no delay.
1836.El
1837.Pp
1838The following parameters can be configured for a queue:
1839.Pp
1840.Bl -tag -width indent -compact
1841.It Cm pipe Ar pipe_nr
1842Connects a queue to the specified pipe.
1843Multiple queues (with the same or different weights) can be connected to
1844the same pipe, which specifies the aggregate rate for the set of queues.
1845.Pp
1846.It Cm weight Ar weight
1847Specifies the weight to be used for flows matching this queue.
1848The weight must be in the range 1..100, and defaults to 1.
1849.El
1850.Pp
1851Finally, the following parameters can be configured for both
1852pipes and queues:
1853.Pp
1854.Bl -tag -width XXXX -compact
1855.Pp
1856.It Cm buckets Ar hash-table-size
1857Specifies the size of the hash table used for storing the
1858various queues.
1859Default value is 64 controlled by the
1860.Xr sysctl 8
1861variable
1862.Em net.inet.ip.dummynet.hash_size ,
1863allowed range is 16 to 65536.
1864.Pp
1865.It Cm mask Ar mask-specifier
1866Packets sent to a given pipe or queue by an
1867.Nm
1868rule can be further classified into multiple flows, each of which is then
1869sent to a different
1870.Em dynamic
1871pipe or queue.
1872A flow identifier is constructed by masking the IP addresses,
1873ports and protocol types as specified with the
1874.Cm mask
1875options in the configuration of the pipe or queue.
1876For each different flow identifier, a new pipe or queue is created
1877with the same parameters as the original object, and matching packets
1878are sent to it.
1879.Pp
1880Thus, when
1881.Em dynamic pipes
1882are used, each flow will get the same bandwidth as defined by the pipe,
1883whereas when
1884.Em dynamic queues
1885are used, each flow will share the parent's pipe bandwidth evenly
1886with other flows generated by the same queue (note that other queues
1887with different weights might be connected to the same pipe).
1888.br
1889Available mask specifiers are a combination of one or more of the following:
1890.Pp
1891.Cm dst-ip Ar mask ,
1892.Cm dst-ip6 Ar mask ,
1893.Cm src-ip Ar mask ,
1894.Cm src-ip6 Ar mask ,
1895.Cm dst-port Ar mask ,
1896.Cm src-port Ar mask ,
1897.Cm flow-id Ar mask ,
1898.Cm proto Ar mask
1899or
1900.Cm all ,
1901.Pp
1902where the latter means all bits in all fields are significant.
1903.Pp
1904.It Cm noerror
1905When a packet is dropped by a dummynet queue or pipe, the error
1906is normally reported to the caller routine in the kernel, in the
1907same way as it happens when a device queue fills up.
1908Setting this
1909option reports the packet as successfully delivered, which can be
1910needed for some experimental setups where you want to simulate
1911loss or congestion at a remote router.
1912.Pp
1913.It Cm plr Ar packet-loss-rate
1914Packet loss rate.
1915Argument
1916.Ar packet-loss-rate
1917is a floating-point number between 0 and 1, with 0 meaning no
1918loss, 1 meaning 100% loss.
1919The loss rate is internally represented on 31 bits.
1920.Pp
1921.It Cm queue Brq Ar slots | size Ns Cm Kbytes
1922Queue size, in
1923.Ar slots
1924or
1925.Cm KBytes .
1926Default value is 50 slots, which
1927is the typical queue size for Ethernet devices.
1928Note that for slow speed links you should keep the queue
1929size short or your traffic might be affected by a significant
1930queueing delay.
1931E.g., 50 max-sized ethernet packets (1500 bytes) mean 600Kbit
1932or 20s of queue on a 30Kbit/s pipe.
1933Even worse effects can result if you get packets from an
1934interface with a much larger MTU, e.g.\& the loopback interface
1935with its 16KB packets.
1936.Pp
1937.It Cm red | gred Ar w_q Ns / Ns Ar min_th Ns / Ns Ar max_th Ns / Ns Ar max_p
1938Make use of the RED (Random Early Detection) queue management algorithm.
1939.Ar w_q
1940and
1941.Ar max_p
1942are floating
1943point numbers between 0 and 1 (0 not included), while
1944.Ar min_th
1945and
1946.Ar max_th
1947are integer numbers specifying thresholds for queue management
1948(thresholds are computed in bytes if the queue has been defined
1949in bytes, in slots otherwise).
1950The
1951.Xr dummynet 4
1952also supports the gentle RED variant (gred).
1953Three
1954.Xr sysctl 8
1955variables can be used to control the RED behaviour:
1956.Bl -tag -width indent
1957.It Em net.inet.ip.dummynet.red_lookup_depth
1958specifies the accuracy in computing the average queue
1959when the link is idle (defaults to 256, must be greater than zero)
1960.It Em net.inet.ip.dummynet.red_avg_pkt_size
1961specifies the expected average packet size (defaults to 512, must be
1962greater than zero)
1963.It Em net.inet.ip.dummynet.red_max_pkt_size
1964specifies the expected maximum packet size, only used when queue
1965thresholds are in bytes (defaults to 1500, must be greater than zero).
1966.El
1967.El
1968.Pp
1969When used with IPv6 data, dummynet currently has several limitations.
1970Information necessary to route link-local packets to an
1971interface is not avalable after processing by dummynet so those packets
1972are dropped in the output path.
1973Care should be taken to insure that link-local packets are not passed to
1974dummynet.
1975.Sh CHECKLIST
1976Here are some important points to consider when designing your
1977rules:
1978.Bl -bullet
1979.It
1980Remember that you filter both packets going
1981.Cm in
1982and
1983.Cm out .
1984Most connections need packets going in both directions.
1985.It
1986Remember to test very carefully.
1987It is a good idea to be near the console when doing this.
1988If you cannot be near the console,
1989use an auto-recovery script such as the one in
1990.Pa /usr/share/examples/ipfw/change_rules.sh .
1991.It
1992Do not forget the loopback interface.
1993.El
1994.Sh FINE POINTS
1995.Bl -bullet
1996.It
1997There are circumstances where fragmented datagrams are unconditionally
1998dropped.
1999TCP packets are dropped if they do not contain at least 20 bytes of
2000TCP header, UDP packets are dropped if they do not contain a full 8
2001byte UDP header, and ICMP packets are dropped if they do not contain
20024 bytes of ICMP header, enough to specify the ICMP type, code, and
2003checksum.
2004These packets are simply logged as
2005.Dq pullup failed
2006since there may not be enough good data in the packet to produce a
2007meaningful log entry.
2008.It
2009Another type of packet is unconditionally dropped, a TCP packet with a
2010fragment offset of one.
2011This is a valid packet, but it only has one use, to try
2012to circumvent firewalls.
2013When logging is enabled, these packets are
2014reported as being dropped by rule -1.
2015.It
2016If you are logged in over a network, loading the
2017.Xr kld 4
2018version of
2019.Nm
2020is probably not as straightforward as you would think.
2021I recommend the following command line:
2022.Bd -literal -offset indent
2023kldload ipfw && \e
2024ipfw add 32000 allow ip from any to any
2025.Ed
2026.Pp
2027Along the same lines, doing an
2028.Bd -literal -offset indent
2029ipfw flush
2030.Ed
2031.Pp
2032in similar surroundings is also a bad idea.
2033.It
2034The
2035.Nm
2036filter list may not be modified if the system security level
2037is set to 3 or higher
2038(see
2039.Xr init 8
2040for information on system security levels).
2041.El
2042.Sh PACKET DIVERSION
2043A
2044.Xr divert 4
2045socket bound to the specified port will receive all packets
2046diverted to that port.
2047If no socket is bound to the destination port, or if the divert module is
2048not loaded, or if the kernel was not compiled with divert socket support,
2049the packets are dropped.
2050.Sh NETWORK ADDRESS TRANSLATION (NAT)
2051The nat configuration command is the following:
2052.Bd -ragged -offset indent
2053.Bk -words
2054.Cm nat
2055.Ar nat_number
2056.Cm config
2057.Ar nat-configuration
2058.Ek
2059.Ed
2060.Pp
2061.
2062The following parameters can be configured:
2063.Bl -tag -width indent
2064.It Cm ip Ar ip_address
2065Define an ip address to use for aliasing.
2066.It Cm if Ar nic
2067Use ip addres of NIC for aliasing, dynamically changing
2068it if NIC's ip address change.
2069.It Cm log
2070Enable logging on this nat instance.
2071.It Cm deny_in
2072Deny any incoming connection from outside world.
2073.It Cm same_ports
2074Try to leave the alias port numbers unchanged from
2075the actual local port numbers.
2076.It Cm unreg_only
2077Traffic on the local network not originating from an
2078unregistered address spaces will be ignored.
2079.It Cm reset
2080Reset table of the packet aliasing engine on address change.
2081.It Cm reverse
2082Reverse the way libalias handles aliasing.
2083.It Cm proxy_only
2084Obey transparent proxy rules only, packet aliasing is not performed.
2085.El
2086.Pp
2087To let the packet continue after being (de)aliased, set the sysctl variable
2088.Em net.inet.ip.fw.one_pass
2089to 0.
2090For more information about aliasing modes, refer to
2091.Xr libalias 3
2092.
2093See Section
2094.Sx EXAMPLES
2095for some examples about nat usage.
2096.Sh REDIRECT AND LSNAT SUPPORT IN IPFW
2097Redirect and LSNAT support follow closely the syntax used in
2098.Xr natd 8
2099.
2100See Section
2101.Sx EXAMPLES
2102for some examples on how to do redirect and lsnat.
2103.Sh SYSCTL VARIABLES
2104A set of
2105.Xr sysctl 8
2106variables controls the behaviour of the firewall and
2107associated modules
2108.Pq Nm dummynet , bridge .
2109These are shown below together with their default value
2110(but always check with the
2111.Xr sysctl 8
2112command what value is actually in use) and meaning:
2113.Bl -tag -width indent
2114.It Em net.inet.ip.dummynet.expire : No 1
2115Lazily delete dynamic pipes/queue once they have no pending traffic.
2116You can disable this by setting the variable to 0, in which case
2117the pipes/queues will only be deleted when the threshold is reached.
2118.It Em net.inet.ip.dummynet.hash_size : No 64
2119Default size of the hash table used for dynamic pipes/queues.
2120This value is used when no
2121.Cm buckets
2122option is specified when configuring a pipe/queue.
2123.It Em net.inet.ip.dummynet.max_chain_len : No 16
2124Target value for the maximum number of pipes/queues in a hash bucket.
2125The product
2126.Cm max_chain_len*hash_size
2127is used to determine the threshold over which empty pipes/queues
2128will be expired even when
2129.Cm net.inet.ip.dummynet.expire=0 .
2130.It Em net.inet.ip.dummynet.red_lookup_depth : No 256
2131.It Em net.inet.ip.dummynet.red_avg_pkt_size : No 512
2132.It Em net.inet.ip.dummynet.red_max_pkt_size : No 1500
2133Parameters used in the computations of the drop probability
2134for the RED algorithm.
2135.It Em net.inet.ip.fw.autoinc_step : No 100
2136Delta between rule numbers when auto-generating them.
2137The value must be in the range 1..1000.
2138.It Em net.inet.ip.fw.curr_dyn_buckets : Em net.inet.ip.fw.dyn_buckets
2139The current number of buckets in the hash table for dynamic rules
2140(readonly).
2141.It Em net.inet.ip.fw.debug : No 1
2142Controls debugging messages produced by
2143.Nm .
2144.It Em net.inet.ip.fw.dyn_buckets : No 256
2145The number of buckets in the hash table for dynamic rules.
2146Must be a power of 2, up to 65536.
2147It only takes effect when all dynamic rules have expired, so you
2148are advised to use a
2149.Cm flush
2150command to make sure that the hash table is resized.
2151.It Em net.inet.ip.fw.dyn_count : No 3
2152Current number of dynamic rules
2153(read-only).
2154.It Em net.inet.ip.fw.dyn_keepalive : No 1
2155Enables generation of keepalive packets for
2156.Cm keep-state
2157rules on TCP sessions.
2158A keepalive is generated to both
2159sides of the connection every 5 seconds for the last 20
2160seconds of the lifetime of the rule.
2161.It Em net.inet.ip.fw.dyn_max : No 8192
2162Maximum number of dynamic rules.
2163When you hit this limit, no more dynamic rules can be
2164installed until old ones expire.
2165.It Em net.inet.ip.fw.dyn_ack_lifetime : No 300
2166.It Em net.inet.ip.fw.dyn_syn_lifetime : No 20
2167.It Em net.inet.ip.fw.dyn_fin_lifetime : No 1
2168.It Em net.inet.ip.fw.dyn_rst_lifetime : No 1
2169.It Em net.inet.ip.fw.dyn_udp_lifetime : No 5
2170.It Em net.inet.ip.fw.dyn_short_lifetime : No 30
2171These variables control the lifetime, in seconds, of dynamic
2172rules.
2173Upon the initial SYN exchange the lifetime is kept short,
2174then increased after both SYN have been seen, then decreased
2175again during the final FIN exchange or when a RST is received.
2176Both
2177.Em dyn_fin_lifetime
2178and
2179.Em dyn_rst_lifetime
2180must be strictly lower than 5 seconds, the period of
2181repetition of keepalives.
2182The firewall enforces that.
2183.It Em net.inet.ip.fw.enable : No 1
2184Enables the firewall.
2185Setting this variable to 0 lets you run your machine without
2186firewall even if compiled in.
2187.It Em net.inet6.ip6.fw.enable : No 1
2188provides the same functionality as above for the IPv6 case.
2189.It Em net.inet.ip.fw.one_pass : No 1
2190When set, the packet exiting from the
2191.Xr dummynet 4
2192pipe or from
2193.Xr ng_ipfw 4
2194node is not passed though the firewall again.
2195Otherwise, after an action, the packet is
2196reinjected into the firewall at the next rule.
2197.It Em net.inet.ip.fw.verbose : No 1
2198Enables verbose messages.
2199.It Em net.inet.ip.fw.verbose_limit : No 0
2200Limits the number of messages produced by a verbose firewall.
2201.It Em net.inet6.ip6.fw.deny_unknown_exthdrs : No 1
2202If enabled packets with unknown IPv6 Extension Headers will be denied.
2203.It Em net.link.ether.ipfw : No 0
2204Controls whether layer-2 packets are passed to
2205.Nm .
2206Default is no.
2207.It Em net.link.bridge.ipfw : No 0
2208Controls whether bridged packets are passed to
2209.Nm .
2210Default is no.
2211.El
2212.Pp
2213.Sh EXAMPLES
2214There are far too many possible uses of
2215.Nm
2216so this Section will only give a small set of examples.
2217.Pp
2218.Ss BASIC PACKET FILTERING
2219This command adds an entry which denies all tcp packets from
2220.Em cracker.evil.org
2221to the telnet port of
2222.Em wolf.tambov.su
2223from being forwarded by the host:
2224.Pp
2225.Dl "ipfw add deny tcp from cracker.evil.org to wolf.tambov.su telnet"
2226.Pp
2227This one disallows any connection from the entire cracker's
2228network to my host:
2229.Pp
2230.Dl "ipfw add deny ip from 123.45.67.0/24 to my.host.org"
2231.Pp
2232A first and efficient way to limit access (not using dynamic rules)
2233is the use of the following rules:
2234.Pp
2235.Dl "ipfw add allow tcp from any to any established"
2236.Dl "ipfw add allow tcp from net1 portlist1 to net2 portlist2 setup"
2237.Dl "ipfw add allow tcp from net3 portlist3 to net3 portlist3 setup"
2238.Dl "..."
2239.Dl "ipfw add deny tcp from any to any"
2240.Pp
2241The first rule will be a quick match for normal TCP packets,
2242but it will not match the initial SYN packet, which will be
2243matched by the
2244.Cm setup
2245rules only for selected source/destination pairs.
2246All other SYN packets will be rejected by the final
2247.Cm deny
2248rule.
2249.Pp
2250If you administer one or more subnets, you can take advantage
2251of the address sets and or-blocks and write extremely
2252compact rulesets which selectively enable services to blocks
2253of clients, as below:
2254.Pp
2255.Dl "goodguys=\*q{ 10.1.2.0/24{20,35,66,18} or 10.2.3.0/28{6,3,11} }\*q"
2256.Dl "badguys=\*q10.1.2.0/24{8,38,60}\*q"
2257.Dl ""
2258.Dl "ipfw add allow ip from ${goodguys} to any"
2259.Dl "ipfw add deny ip from ${badguys} to any"
2260.Dl "... normal policies ..."
2261.Pp
2262The
2263.Cm verrevpath
2264option could be used to do automated anti-spoofing by adding the
2265following to the top of a ruleset:
2266.Pp
2267.Dl "ipfw add deny ip from any to any not verrevpath in"
2268.Pp
2269This rule drops all incoming packets that appear to be coming to the
2270system on the wrong interface.
2271For example, a packet with a source
2272address belonging to a host on a protected internal network would be
2273dropped if it tried to enter the system from an external interface.
2274.Pp
2275The
2276.Cm antispoof
2277option could be used to do similar but more restricted anti-spoofing
2278by adding the following to the top of a ruleset:
2279.Pp
2280.Dl "ipfw add deny ip from any to any not antispoof in"
2281.Pp
2282This rule drops all incoming packets that appear to be coming from another
2283directly connected system but on the wrong interface.
2284For example, a packet with a source address of
2285.Li 192.168.0.0/24
2286, configured on
2287.Li fxp0
2288, but coming in on
2289.Li fxp1
2290would be dropped.
2291.Ss DYNAMIC RULES
2292In order to protect a site from flood attacks involving fake
2293TCP packets, it is safer to use dynamic rules:
2294.Pp
2295.Dl "ipfw add check-state"
2296.Dl "ipfw add deny tcp from any to any established"
2297.Dl "ipfw add allow tcp from my-net to any setup keep-state"
2298.Pp
2299This will let the firewall install dynamic rules only for
2300those connection which start with a regular SYN packet coming
2301from the inside of our network.
2302Dynamic rules are checked when encountering the first
2303.Cm check-state
2304or
2305.Cm keep-state
2306rule.
2307A
2308.Cm check-state
2309rule should usually be placed near the beginning of the
2310ruleset to minimize the amount of work scanning the ruleset.
2311Your mileage may vary.
2312.Pp
2313To limit the number of connections a user can open
2314you can use the following type of rules:
2315.Pp
2316.Dl "ipfw add allow tcp from my-net/24 to any setup limit src-addr 10"
2317.Dl "ipfw add allow tcp from any to me setup limit src-addr 4"
2318.Pp
2319The former (assuming it runs on a gateway) will allow each host
2320on a /24 network to open at most 10 TCP connections.
2321The latter can be placed on a server to make sure that a single
2322client does not use more than 4 simultaneous connections.
2323.Pp
2324.Em BEWARE :
2325stateful rules can be subject to denial-of-service attacks
2326by a SYN-flood which opens a huge number of dynamic rules.
2327The effects of such attacks can be partially limited by
2328acting on a set of
2329.Xr sysctl 8
2330variables which control the operation of the firewall.
2331.Pp
2332Here is a good usage of the
2333.Cm list
2334command to see accounting records and timestamp information:
2335.Pp
2336.Dl ipfw -at list
2337.Pp
2338or in short form without timestamps:
2339.Pp
2340.Dl ipfw -a list
2341.Pp
2342which is equivalent to:
2343.Pp
2344.Dl ipfw show
2345.Pp
2346Next rule diverts all incoming packets from 192.168.2.0/24
2347to divert port 5000:
2348.Pp
2349.Dl ipfw divert 5000 ip from 192.168.2.0/24 to any in
2350.Pp
2351.Ss TRAFFIC SHAPING
2352The following rules show some of the applications of
2353.Nm
2354and
2355.Xr dummynet 4
2356for simulations and the like.
2357.Pp
2358This rule drops random incoming packets with a probability
2359of 5%:
2360.Pp
2361.Dl "ipfw add prob 0.05 deny ip from any to any in"
2362.Pp
2363A similar effect can be achieved making use of dummynet pipes:
2364.Pp
2365.Dl "ipfw add pipe 10 ip from any to any"
2366.Dl "ipfw pipe 10 config plr 0.05"
2367.Pp
2368We can use pipes to artificially limit bandwidth, e.g.\& on a
2369machine acting as a router, if we want to limit traffic from
2370local clients on 192.168.2.0/24 we do:
2371.Pp
2372.Dl "ipfw add pipe 1 ip from 192.168.2.0/24 to any out"
2373.Dl "ipfw pipe 1 config bw 300Kbit/s queue 50KBytes"
2374.Pp
2375note that we use the
2376.Cm out
2377modifier so that the rule is not used twice.
2378Remember in fact that
2379.Nm
2380rules are checked both on incoming and outgoing packets.
2381.Pp
2382Should we want to simulate a bidirectional link with bandwidth
2383limitations, the correct way is the following:
2384.Pp
2385.Dl "ipfw add pipe 1 ip from any to any out"
2386.Dl "ipfw add pipe 2 ip from any to any in"
2387.Dl "ipfw pipe 1 config bw 64Kbit/s queue 10Kbytes"
2388.Dl "ipfw pipe 2 config bw 64Kbit/s queue 10Kbytes"
2389.Pp
2390The above can be very useful, e.g.\& if you want to see how
2391your fancy Web page will look for a residential user who
2392is connected only through a slow link.
2393You should not use only one pipe for both directions, unless
2394you want to simulate a half-duplex medium (e.g.\& AppleTalk,
2395Ethernet, IRDA).
2396It is not necessary that both pipes have the same configuration,
2397so we can also simulate asymmetric links.
2398.Pp
2399Should we want to verify network performance with the RED queue
2400management algorithm:
2401.Pp
2402.Dl "ipfw add pipe 1 ip from any to any"
2403.Dl "ipfw pipe 1 config bw 500Kbit/s queue 100 red 0.002/30/80/0.1"
2404.Pp
2405Another typical application of the traffic shaper is to
2406introduce some delay in the communication.
2407This can significantly affect applications which do a lot of Remote
2408Procedure Calls, and where the round-trip-time of the
2409connection often becomes a limiting factor much more than
2410bandwidth:
2411.Pp
2412.Dl "ipfw add pipe 1 ip from any to any out"
2413.Dl "ipfw add pipe 2 ip from any to any in"
2414.Dl "ipfw pipe 1 config delay 250ms bw 1Mbit/s"
2415.Dl "ipfw pipe 2 config delay 250ms bw 1Mbit/s"
2416.Pp
2417Per-flow queueing can be useful for a variety of purposes.
2418A very simple one is counting traffic:
2419.Pp
2420.Dl "ipfw add pipe 1 tcp from any to any"
2421.Dl "ipfw add pipe 1 udp from any to any"
2422.Dl "ipfw add pipe 1 ip from any to any"
2423.Dl "ipfw pipe 1 config mask all"
2424.Pp
2425The above set of rules will create queues (and collect
2426statistics) for all traffic.
2427Because the pipes have no limitations, the only effect is
2428collecting statistics.
2429Note that we need 3 rules, not just the last one, because
2430when
2431.Nm
2432tries to match IP packets it will not consider ports, so we
2433would not see connections on separate ports as different
2434ones.
2435.Pp
2436A more sophisticated example is limiting the outbound traffic
2437on a net with per-host limits, rather than per-network limits:
2438.Pp
2439.Dl "ipfw add pipe 1 ip from 192.168.2.0/24 to any out"
2440.Dl "ipfw add pipe 2 ip from any to 192.168.2.0/24 in"
2441.Dl "ipfw pipe 1 config mask src-ip 0x000000ff bw 200Kbit/s queue 20Kbytes"
2442.Dl "ipfw pipe 2 config mask dst-ip 0x000000ff bw 200Kbit/s queue 20Kbytes"
2443.Ss LOOKUP TABLES
2444In the following example, we need to create several traffic bandwidth
2445classes and we need different hosts/networks to fall into different classes.
2446We create one pipe for each class and configure them accordingly.
2447Then we create a single table and fill it with IP subnets and addresses.
2448For each subnet/host we set the argument equal to the number of the pipe
2449that it should use.
2450Then we classify traffic using a single rule:
2451.Pp
2452.Dl "ipfw pipe 1 config bw 1000Kbyte/s"
2453.Dl "ipfw pipe 4 config bw 4000Kbyte/s"
2454.Dl "..."
2455.Dl "ipfw table 1 add 192.168.2.0/24 1"
2456.Dl "ipfw table 1 add 192.168.0.0/27 4"
2457.Dl "ipfw table 1 add 192.168.0.2 1"
2458.Dl "..."
2459.Dl "ipfw add pipe tablearg ip from table(1) to any"
2460.Pp
2461Using the
2462.Cm fwd
2463action, the table entries may include hostnames and IP addresses.
2464.Pp
2465.Dl "ipfw table 1 add 192.168.2.0/24 10.23.2.1"
2466.Dl "ipfw table 1 add 192.168.0.0/27 router1.dmz"
2467.Dl "..."
2468.Dl "ipfw add 100 fwd tablearg ip from any to table(1)"
2469.Ss SETS OF RULES
2470To add a set of rules atomically, e.g.\& set 18:
2471.Pp
2472.Dl "ipfw set disable 18"
2473.Dl "ipfw add NN set 18 ...         # repeat as needed"
2474.Dl "ipfw set enable 18"
2475.Pp
2476To delete a set of rules atomically the command is simply:
2477.Pp
2478.Dl "ipfw delete set 18"
2479.Pp
2480To test a ruleset and disable it and regain control if something goes wrong:
2481.Pp
2482.Dl "ipfw set disable 18"
2483.Dl "ipfw add NN set 18 ...         # repeat as needed"
2484.Dl "ipfw set enable 18; echo done; sleep 30 && ipfw set disable 18"
2485.Pp
2486Here if everything goes well, you press control-C before the "sleep"
2487terminates, and your ruleset will be left active.
2488Otherwise, e.g.\& if
2489you cannot access your box, the ruleset will be disabled after
2490the sleep terminates thus restoring the previous situation.
2491.Pp
2492To show rules of the specific set:
2493.Pp
2494.Dl "ipfw set 18 show"
2495.Pp
2496To show rules of the disabled set:
2497.Pp
2498.Dl "ipfw -S set 18 show"
2499.Pp
2500To clear a specific rule counters of the specific set:
2501.Pp
2502.Dl "ipfw set 18 zero NN"
2503.Pp
2504To delete a specific rule of the specific set:
2505.Pp
2506.Dl "ipfw set 18 delete NN"
2507.Ss NAT, REDIRECT AND LSNAT
2508First redirect all the traffic to nat instance 123:
2509.Pp
2510.Dl "ipfw add nat 123 all from any to any"
2511.Pp
2512Then to configure nat instance 123 to alias all the outgoing traffic with ip
2513192.168.0.123, blocking all incoming connections, trying to keep
2514same ports on both sides, clearing aliasing table on address change
2515and keeping a log of traffic/link statistics:
2516.Pp
2517.Dl "ipfw nat 123 config ip 192.168.0.123 log deny_in reset same_ports"
2518.Pp
2519Or to change address of instance 123, aliasing table will be cleared (see
2520reset option):
2521.Pp
2522.Dl "ipfw nat 123 config ip 10.0.0.1"
2523.Pp
2524To see configuration of nat instance 123:
2525.Pp
2526.Dl "ipfw nat 123 show config"
2527.Pp
2528To show logs of all the instances in range 111-999:
2529.Pp
2530.Dl "ipfw nat 111-999 show"
2531.Pp
2532To see configurations of all instances:
2533.Pp
2534.Dl "ipfw nat show config"
2535.Pp
2536Or a redirect rule with mixed modes could looks like:
2537.Pp
2538.Dl "ipfw nat 123 config redirect_addr 10.0.0.1 10.0.0.66"
2539.Dl "			 redirect_port tcp 192.168.0.1:80 500"
2540.Dl "			 redirect_proto udp 192.168.1.43 192.168.1.1"
2541.Dl "			 redirect_addr 192.168.0.10,192.168.0.11"
2542.Dl "			 	    10.0.0.100	# LSNAT"
2543.Dl "			 redirect_port tcp 192.168.0.1:80,192.168.0.10:22"
2544.Dl "			 	    500		# LSNAT"
2545.Pp
2546or it could be splitted in:
2547.Pp
2548.Dl "ipfw nat 1 config redirect_addr 10.0.0.1 10.0.0.66"
2549.Dl "ipfw nat 2 config redirect_port tcp 192.168.0.1:80 500"
2550.Dl "ipfw nat 3 config redirect_proto udp 192.168.1.43 192.168.1.1"
2551.Dl "ipfw nat 4 config redirect_addr 192.168.0.10,192.168.0.11,192.168.0.12"
2552.Dl "				         10.0.0.100"
2553.Dl "ipfw nat 5 config redirect_port tcp"
2554.Dl "			192.168.0.1:80,192.168.0.10:22,192.168.0.20:25 500"
2555.Pp
2556.Sh SEE ALSO
2557.Xr cpp 1 ,
2558.Xr m4 1 ,
2559.Xr altq 4 ,
2560.Xr divert 4 ,
2561.Xr dummynet 4 ,
2562.Xr if_bridge 4 ,
2563.Xr ip 4 ,
2564.Xr ipfirewall 4 ,
2565.Xr ng_ipfw 4 ,
2566.Xr protocols 5 ,
2567.Xr services 5 ,
2568.Xr init 8 ,
2569.Xr kldload 8 ,
2570.Xr reboot 8 ,
2571.Xr sysctl 8 ,
2572.Xr syslogd 8
2573.Sh HISTORY
2574The
2575.Nm
2576utility first appeared in
2577.Fx 2.0 .
2578.Xr dummynet 4
2579was introduced in
2580.Fx 2.2.8 .
2581Stateful extensions were introduced in
2582.Fx 4.0 .
2583.Nm ipfw2
2584was introduced in Summer 2002.
2585.Sh AUTHORS
2586.An Ugen J. S. Antsilevich ,
2587.An Poul-Henning Kamp ,
2588.An Alex Nash ,
2589.An Archie Cobbs ,
2590.An Luigi Rizzo .
2591.Pp
2592.An -nosplit
2593API based upon code written by
2594.An Daniel Boulet
2595for BSDI.
2596.Pp
2597.An -nosplit
2598In-kernel NAT support written by
2599.An Paolo Pisati Aq piso@FreeBSD.org
2600as part of a Summer of Code 2005 project.
2601.Pp
2602Work on
2603.Xr dummynet 4
2604traffic shaper supported by Akamba Corp.
2605.Sh BUGS
2606The syntax has grown over the years and sometimes it might be confusing.
2607Unfortunately, backward compatibility prevents cleaning up mistakes
2608made in the definition of the syntax.
2609.Pp
2610.Em !!! WARNING !!!
2611.Pp
2612Misconfiguring the firewall can put your computer in an unusable state,
2613possibly shutting down network services and requiring console access to
2614regain control of it.
2615.Pp
2616Incoming packet fragments diverted by
2617.Cm divert
2618are reassembled before delivery to the socket.
2619The action used on those packet is the one from the
2620rule which matches the first fragment of the packet.
2621.Pp
2622Packets diverted to userland, and then reinserted by a userland process
2623may lose various packet attributes.
2624The packet source interface name
2625will be preserved if it is shorter than 8 bytes and the userland process
2626saves and reuses the sockaddr_in
2627(as does
2628.Xr natd 8 ) ;
2629otherwise, it may be lost.
2630If a packet is reinserted in this manner, later rules may be incorrectly
2631applied, making the order of
2632.Cm divert
2633rules in the rule sequence very important.
2634.Pp
2635Dummynet drops all packets with IPv6 link-local addresses.
2636.Pp
2637Rules using
2638.Cm uid
2639or
2640.Cm gid
2641may not behave as expected.
2642In particular, incoming SYN packets may
2643have no uid or gid associated with them since they do not yet belong
2644to a TCP connection, and the uid/gid associated with a packet may not
2645be as expected if the associated process calls
2646.Xr setuid 2
2647or similar system calls.
2648.Pp
2649Rule syntax is subject to the command line environment and some patterns
2650may need to be escaped with the backslash character
2651or quoted appropriately.
2652.Pp
2653Due to the architecture of
2654.Xr libalias 3 ,
2655ipfw nat is not compatible with the tcp segmentation offloading
2656(TSO). Thus, to reliably nat your network traffic, please disable TSO
2657on your NICs using
2658.Xr ifconfig 8 .
2659