IPFW(8)

📦
version 15.1 · 2 versions
, in freebsd 15.1

🌸 IPFW(8) — from the System Manager's Manual 🌸

okay so this is the IPFW(8) page, and it is literally your user interface for the ipfw(4) firewall, the dummynet(4) traffic shaper/packet scheduler, and the in-kernel NAT services. buckle up bestie, it's giving networking main character energy 💅

💖 NAME 💖

ipfw, dnctl — User interface for firewall, traffic shaper, packet scheduler, in-kernel NAT.

✨ SYNOPSIS ✨

🔥 FIREWALL CONFIGURATION 🔥

ipfw [-cq] add rule
ipfw [-acdefnNStT] [set N] {list | show} [rule | first-last ...]
ipfw [-f | -q] [set N] flush
ipfw [-q] [set N] {delete | zero | resetlog} [number ...]

ipfw set [disable number ...] [enable number ...]
ipfw set move [rule] number to number
ipfw set swap number number
ipfw set show

🎀 SYSCTL SHORTCUTS 🎀

ipfw enable
     {firewall | altq | one_pass | debug | verbose | dyn_keepalive | skipto_cache}
ipfw disable
     {firewall | altq | one_pass | debug | verbose | dyn_keepalive | skipto_cache}

🦋 LOOKUP TABLES 🦋

ipfw [set N] table name create create-options
ipfw [set N] table {name | all} destroy
ipfw [set N] table name modify modify-options
ipfw [set N] table name swap name
ipfw [set N] table name add table-key [value]
ipfw [set N] table name add [table-key value ...]
ipfw [set N] table name atomic add [table-key value ...]
ipfw [set N] table name delete [table-key ...]
ipfw [set N] table name lookup addr
ipfw [set N] table name lock
ipfw [set N] table name unlock
ipfw [set N] table {name | all} list
ipfw [set N] table {name | all} info
ipfw [set N] table {name | all} detail
ipfw [set N] table {name | all} flush

🌈 DUMMYNET CONFIGURATION (TRAFFIC SHAPER AND PACKET SCHEDULER) 🌈

dnctl {pipe | queue | sched} number config config-options
dnctl [-s [field]] {pipe | queue | sched} {delete | list | show}
     [number ...]

💜 IN-KERNEL NAT 💜

ipfw [-q] nat number config config-options
ipfw [-q] nat number delete
ipfw nat number show {config | log}

⭐ STATEFUL IPv6/IPv4 NETWORK ADDRESS AND PROTOCOL TRANSLATION ⭐

ipfw [set N] nat64lsn name create create-options
ipfw [set N] nat64lsn name config config-options
ipfw [set N] nat64lsn {name | all} {list | show} [states]
ipfw [set N] nat64lsn {name | all} destroy
ipfw [set N] nat64lsn name stats [reset]

🧚 STATELESS IPv6/IPv4 NETWORK ADDRESS AND PROTOCOL TRANSLATION 🧚

ipfw [set N] nat64stl name create create-options
ipfw [set N] nat64stl name config config-options
ipfw [set N] nat64stl {name | all} {list | show}
ipfw [set N] nat64stl {name | all} destroy
ipfw [set N] nat64stl name stats [reset]

🍓 XLAT464 CLAT IPv6/IPv4 NETWORK ADDRESS AND PROTOCOL TRANSLATION 🍓

ipfw [set N] nat64clat name create create-options
ipfw [set N] nat64clat name config config-options
ipfw [set N] nat64clat {name | all} {list | show}
ipfw [set N] nat64clat {name | all} destroy
ipfw [set N] nat64clat name stats [reset]

👑 IPv6-to-IPv6 NETWORK PREFIX TRANSLATION 👑

ipfw [set N] nptv6 name create create-options
ipfw [set N] nptv6 {name | all} {list | show}
ipfw [set N] nptv6 {name | all} destroy
ipfw [set N] nptv6 name stats [reset]

💀 INTERNAL DIAGNOSTICS 💀

ipfw internal iflist
ipfw internal monitor [filter-comment]
ipfw internal talist
ipfw internal vlist

🫶 LIST OF RULES AND PREPROCESSING 🫶

ipfw [-cfnNqS] [-p preproc [preproc-flags]] pathname

🎀 DESCRIPTION 🎀

okay so the ipfw utility is the user interface for controlling the ipfw(4) firewall, the dummynet(4) traffic shaper/packet scheduler, and the in-kernel NAT services.

A firewall configuration, or ruleset, is made of a list of rules numbered from 1 to 65535. Packets are passed to the firewall from a number of different places in the protocol stack (depending on the source and destination of the packet, it is possible for the firewall to be invoked multiple times on the same packet). The packet passed to the firewall is compared against each of the rules in the ruleset, in rule-number order (multiple rules with the same number are permitted, in which case they are processed in order of insertion). When a match is found, the action corresponding to the matching rule is performed.

Depending on the action and certain system settings, packets can be reinjected into the firewall at some rule after the matching one for further processing.

A ruleset always includes a default rule (numbered 65535) which cannot be modified or deleted, and matches all packets. The action associated with the default rule can be either deny or allow depending on how the kernel is configured.

If the ruleset includes one or more rules with the keep-state, record-state, limit or set-limit option, the firewall will have a stateful behaviour, i.e., upon a match it will create dynamic rules, i.e., rules that match packets with the same 5-tuple (protocol, source and destination addresses and ports) as the packet which caused their creation. Dynamic rules, which have a limited lifetime, are checked at the first occurrence of a check-state, keep-state or limit rule, and are typically used to open the firewall on-demand to legitimate traffic only. Please note, that keep-state and limit imply implicit check-state for all packets (not only these matched by the rule) but record-state and set-limit have no implicit check-state. See the “STATEFUL FIREWALL” and “EXAMPLES” Sections below for more information on the stateful behaviour of ipfw.

All rules (including dynamic ones) have a few associated counters: a packet count, a byte count, a log count and a timestamp indicating the time of the last match. Counters can be displayed or reset with ipfw commands.

Each rule belongs to one of 32 different sets, and there are ipfw commands to atomically manipulate sets, such as enable, disable, swap sets, move all rules in a set to another one, delete all rules in a set. These can be useful to install temporary configurations, or to test them. See Section “SETS OF RULES” for more information on sets.

Rules can be added with the add command; deleted individually or in groups with the delete command, and globally (except those in set 31) with the flush command; displayed, optionally with the content of the counters, using the show and list commands. Finally, counters can be reset with the zero and resetlog commands.

💅 COMMAND OPTIONS 💅

The following general options are available when invoking ipfw:

  • -a — Show counter values when listing rules. The show command implies this option.
  • -b — Only show the action and the comment, not the body of a rule. Implies -c.
  • -c — When entering or showing rules, print them in compact form, i.e., omitting the "ip from any to any" string when this does not carry any additional information.
  • -d — When listing, show dynamic rules in addition to static ones.
  • -D — When listing, show only dynamic states. When deleting, delete only dynamic states.
  • -f — Run without prompting for confirmation for commands that can cause problems if misused, i.e., flush. If there is no tty associated with the process, this is implied. The delete command with this flag ignores possible errors, i.e., nonexistent rule number. And for batched commands execution continues with the next command.
  • -i — When listing a table (see the “LOOKUP TABLES” section below for more information on lookup tables), format values as IP addresses. By default, values are shown as integers.
  • -n — Only check syntax of the command strings, without actually passing them to the kernel.
  • -N — Try to resolve addresses and service names in output.
  • -q — Be quiet when executing the add, nat, zero, resetlog or flush commands; (implies -f). This is useful when updating rulesets by executing multiple ipfw commands in a script (e.g., sh /etc/rc.firewall), or by processing a file with many ipfw rules across a remote login session. It also stops a table add or delete from failing if the entry already exists or is not present.

The reason why this option may be important is that for some of these actions, ipfw may print a message; if the action results in blocking the traffic to the remote client, the remote login session will be closed and the rest of the ruleset will not be processed. Access to the console would then be required to recover.

  • -S — When listing rules, show the set each rule belongs to. If this flag is not specified, disabled rules will not be listed.
  • -s [field] — When listing pipes, sort according to one of the four counters (total or current packets or bytes).
  • -t — When listing, show last match timestamp converted with ctime().
  • -T — When listing, show last match timestamp as seconds from the epoch. This form can be more convenient for postprocessing by scripts.

🦋 LIST OF RULES AND PREPROCESSING 🦋

To ease configuration, rules can be put into a file which is processed using ipfw as shown in the last synopsis line. An absolute pathname must be used. The file will be read line by line and applied as arguments to the ipfw utility.

Optionally, a preprocessor can be specified using -p preproc where pathname is to be piped through. Useful preprocessors include cpp(1) and m4(1). If preproc does not start with a slash (‘/’) as its first character, the usual PATH name search is performed. Care should be taken with this in environments where not all file systems are mounted (yet) by the time ipfw is being run (e.g. when they are mounted over NFS). Once -p has been specified, any additional arguments are passed on to the preprocessor for interpretation. This allows for flexible configuration files (like conditionalizing them on the local hostname) and the use of macros to centralize frequently required arguments like IP addresses.

🌈 TRAFFIC SHAPER CONFIGURATION 🌈

The dnctl pipe, queue and sched commands are used to configure the traffic shaper and packet scheduler. See the “TRAFFIC SHAPER (DUMMYNET) CONFIGURATION” Section below for details.

⚠️ warning bestie ⚠️ If the world and the kernel get out of sync the ipfw ABI may break, preventing you from being able to add any rules. This can adversely affect the booting process. You can use ipfw disable firewall to temporarily disable the firewall to regain access to the network, allowing you to fix the problem.

🌈 PACKET FLOW 🌈

Okay bestie, listen up, because this is genuinely the glow-up moment of the whole page. A packet gets checked against the active ruleset in multiple places in the protocol stack, and it's all under the control of several sysctl variables. Below are those places and those variables, and it is so important that you hold this picture in your head while designing a correct ruleset. Like, actually picture it. Visualize. Manifest. 💅

                    ^    to upper layers    V
                    |                       |
                    +----------->-----------+
                    ^                       V
              [ip(6)_input]           [ip(6)_output]     net.inet(6).ip(6).fw.enable=1
                    |                       |
                    ^                       V
              [ether_demux]        [ether_output_frame]  net.link.ether.ipfw=1
                    |                       |
                    +-->--[bdg_forward]-->--+            net.link.bridge.ipfw=1
                    ^                       V
                    |      to devices       |

The number of times the same packet goes through the firewall can vary between 0 and 4, depending on packet source and destination, and system configuration. Yes, zero. Not every packet gets the full tour, so don't @ me when your rule doesn't fire. 🫶

Note that as packets flow through the stack, headers can be stripped or added to it, and so they may or may not be available for inspection. E.g., incoming packets will include the MAC header when ipfw is invoked from ether_demux(), but the same packets will have the MAC header stripped off when ipfw is invoked from ip_input() or ip6_input(). It's giving "same packet, different outfit." ✨

Also note that each packet is always checked against the complete ruleset, irrespective of the place where the check occurs, or the source of the packet. So if a rule contains some match patterns or actions which are not valid for the place of invocation (e.g. trying to match a MAC header within ip_input or ip6_input), the match pattern will not match — but a not operator in front of such patterns will cause the pattern to always match on those packets. Sneaky, right? 😭 So it's the responsibility of the programmer, if necessary, to write a suitable ruleset to differentiate among the possible places. skipto rules can be super useful here, as an example:

# packets from ether_demux or bdg_forward
ipfw add 10 skipto 1000 all from any to any layer2 in
# packets from ip_input
ipfw add 10 skipto 2000 all from any to any not layer2 in
# packets from ip_output
ipfw add 10 skipto 3000 all from any to any not layer2 out
# packets from ether_output_frame
ipfw add 10 skipto 4000 all from any to any layer2 out

(yes, at the moment there is no way to differentiate between ether_demux and bdg_forward). Tragic, I know. 💀

Also note that only actions allow, deny, netgraph, ngtee and related to dummynet are processed for layer2 frames — all other actions act as if they were allow for such frames. Full set of actions is supported for IP packets without layer2 headers only. For example, divert action does not divert layer2 frames. Not me finding out the hard way 🙈

✍️ SYNTAX ✍️

In general, each keyword or argument must be provided as a separate command line argument, with no leading or trailing spaces. Keywords are case-sensitive, whereas arguments may or may not be case-sensitive depending on their nature (e.g. uid's are, hostnames are not). Fr fr, read that twice. 🎀

Some arguments (e.g., port or address lists) are comma-separated lists of values. In this case, spaces after commas ',' are allowed to make the line more readable. You can also put the entire command (including flags) into a single argument. E.g., the following forms are equivalent:

ipfw -q add deny src-ip 10.0.0.0/24,127.0.0.1/8
ipfw -q add deny src-ip 10.0.0.0/24, 127.0.0.1/8
ipfw "-q add deny src-ip 10.0.0.0/24, 127.0.0.1/8"

Three forms, one vibe. Slay. 💜

🧚 RULE FORMAT 🧚

The format of firewall rules is the following:

[rule_number] [set set_number] [prob match_probability] action
[log [log_opts]] [altq queue] [{tag | untag} number] body

where the body of the rule specifies which information is used for filtering packets, among the following:

  • Layer2 header fields — When available
  • IPv4 and IPv6 Protocol — SCTP, TCP, UDP, ICMP, etc.
  • Source and dest. addresses and ports
  • Direction — See Section “PACKET FLOW”
  • Transmit and receive interface — By name or address
  • Misc. IP header fields — Version, type of service, datagram length, identification, fragmentation flags, Time To Live
  • IP options
  • IPv6 Extension headers — Fragmentation, Hop-by-Hop options, Routing Headers, Source routing rthdr0, Mobile IPv6 rthdr2, IPSec options.
  • IPv6 Flow-ID
  • Misc. TCP header fields — TCP flags (SYN, FIN, ACK, RST, etc.), sequence number, acknowledgment number, window
  • TCP options
  • ICMP types — for ICMP packets
  • ICMP6 types — for ICMP6 packets
  • User/group ID — When the packet can be associated with a local socket.
  • Divert status — Whether a packet came from a divert socket (e.g., natd(8)).
  • Fib annotation state — Whether a packet has been tagged for using a specific FIB (routing table) in future forwarding decisions.

Note that some of the above information, e.g. source MAC or IP addresses and TCP/UDP ports, can be easily spoofed, so filtering on those fields alone might not guarantee the desired results. Bestie. Do not trust the header. The header lies. 😤

rule_number

Each rule is associated with a rule_number in the range 1..65535, with the latter reserved for the default rule. Rules are checked sequentially by rule number. Multiple rules can have the same number, in which case they are checked (and listed) according to the order in which they have been added. Order matters, always has. 👑

If a rule is entered without specifying a number, the kernel will assign one in such a way that the rule becomes the last one before the default rule. Automatic rule numbers are assigned by incrementing the last non-default rule number by the value of the sysctl variable net.inet.ip.fw.autoinc_step which defaults to 100. If this is not possible (e.g. because we would go beyond the maximum allowed rule number), the number of the last non-default value is used instead. It's giving "auto-assigned but make it sensible." ✨

set set_number

Each rule is associated with a set_number in the range 0..31. Sets can be individually disabled and enabled, so this parameter is of fundamental importance for atomic ruleset manipulation. It can be also used to simplify deletion of groups of rules. If a rule is entered without specifying a set number, set 0 will be used. 🎀

Set 31 is special in that it cannot be disabled, and rules in set 31 are not deleted by the ipfw flush command (but you can delete them with the ipfw delete set 31 command). Set 31 is also used for the default rule. Set 31 said "I'm the main character" and honestly? It ate. 💅

prob match_probability

A match is only declared with the specified probability (floating point number between 0 and 1). This can be useful for a number of applications such as random packet drop or (in conjunction with dummynet) to simulate the effect of multiple paths leading to out-of-order packet delivery. Chaos as a feature, we love to see it. 🌸

Note: this condition is checked before any other condition, including ones such as keep-state or check-state which might have side effects. So it's checked first-first. The way that this one jumps the queue 😭

log [logamount number]

Packets matching a rule with the log keyword will be made available for logging. Unless per-rule log destination is specified by logdst logdst_spec option (see below), packets are logged in two ways: if the sysctl variable net.inet.ip.fw.verbose is set to 0 (default), one can use bpf(4) attached to the ipfw0 pseudo interface. This pseudo interface can be created manually after a system boot by using the following command:

# ifconfig ipfw0 create

Or, automatically at boot time by adding the following line to the rc.conf(5) file:

firewall_logif="YES"

There is zero overhead when no bpf(4) is attached to the pseudo interface. Zero. None. Free. 🦋

If net.inet.ip.fw.verbose is set to 1, packets will be logged to syslogd(8) with a LOG_SECURITY facility up to a maximum of logamount packets. If no logamount is specified, the limit is taken from the sysctl variable net.inet.ip.fw.verbose_limit. In both cases, a value of 0 means unlimited logging. Go off, log everything. 🔥

Once the limit is reached, logging can be re-enabled by clearing the logging counter or the packet counter for that entry, see the resetlog command.

Note: logging is done after all other packet matching conditions have been successfully verified, and before performing the final action (accept, deny, etc.) on the packet. So it's the last check, first action. Timeline is wild.

log [logamount number] logdst logdst_spec

logdst_spec is a comma-separated list of log destinations for logging packets matching the rule. Destinations supported are:

  • syslog — Logs a packet to syslogd(8) with a LOG_SECURITY facility.
  • ipfw0 — Logs a packet to the ipfw0 pseudo interface.
  • rtsock — Logs a packet to the route(4) socket. See the comments of ipfw_log_rtsock() in ipfw source code for more information on the message's structure.
Note: logamount limits a number of logging events rather than packets being logged. I.e. A packet matching a rule with
...  log logamount 100 logdst syslog,ipfw0 ...

will log upto 50 packets.

Math girlie moment: two destinations, one hundred events, fifty packets. The math is mathing. 🧮💖

tag number

Okay so when a packet matches a rule that has the tag keyword, it gets a numeric tag attached — the number has to be in the range 1..65534. This tag is like an internal lil marker (it is not sent out over the wire, bestie, it stays inside) that you can use later to identify those exact packets. Use case? Say you wanna provide trust between interfaces and start doing policy-based filtering — tag is your girl. A packet can rock multiple tags at the same time 💅.

Tags are "sticky", which means once a matching rule applies one, it sticks around until it's explicitly removed. Tags are kept with the packet everywhere within the kernel, but 💀 they are lost when the packet leaves the kernel — for example when it's transmitted out to the network, or sent to a divert(4) socket.

To check for tags that were already applied, use the tagged rule option. To delete a tag that was already applied, use the untag keyword.

Note: since tags are kept with the packet everywhere in kernelspace, they can be set and unset anywhere in the kernel network subsystem (using the mbuf_tags(9) facility) — not only via the ipfw(4) tag and untag keywords. Like, there can totally be a specialized netgraph(4) node doing traffic analyzing and tagging for later inspecting in the firewall 🧚

untag number

When a packet matches a rule with the untag keyword, the tag with the number number is searched among the tags attached to this packet and, if it's found, it gets removed from the packet. Any other tags bound to the packet, if present, are left untouched 🫶

setmark value | tablearg

When a packet matches a rule with the setmark keyword, a 32-bit numeric mark gets assigned to the packet. The mark is basically an extension to the tags, but here's the catch: the mark is preserved for a packet within a single ipfw ruleset traversal only, and it's lost when the packet is checked against the active ruleset next time (peep the “PACKET FLOW” section) or when it leaves ipfw context (e.g. accepted, diverted, bridged or routed) 😭

Unlike tags, the mark can be matched as a lookup table key, or compared with a bitwise mask applied against another value. Each packet can only have one mark, so setmark always overwrites the previous mark value. No hoarding marks, bestie.

The initial mark value is 0. To check the current mark value, use the mark rule option. Mark value can be entered as decimal or hexadecimal (if it's prefixed by 0x), and they are always printed as hexadecimal. Periodt ✨

altq queue

When a packet matches a rule with the altq keyword, the ALTQ identifier for the given queue gets attached (see altq(4)). Heads up though: this ALTQ tag is only meaningful for packets going "out" of IPFW, and not for ones being rejected or going to divert sockets. Also, if there isn't enough memory at the time the packet is processed, it will not be tagged — so it's wise to make your ALTQ "default" queue policy account for that 🙈

If multiple altq rules match a single packet, only the first one adds the ALTQ classification tag. In doing so, traffic may be shaped by using count altq queue rules for classification early in the ruleset, then later applying the filtering decision. For example, check-state and keep-state rules may come later and provide the actual filtering decisions in addition to the fallback ALTQ tag 👑

You must run pfctl(8) to set up the queues before IPFW will be able to look them up by name. And if the ALTQ disciplines are rearranged, the rules containing the queue identifiers in the kernel will likely have gone stale and need to be reloaded. Stale queue identifiers will probably result in misclassification, ngl.

All system ALTQ processing can be turned on or off via ipfw enable altq and ipfw disable altq. The usage of net.inet.ip.fw.one_pass is irrelevant to ALTQ traffic shaping, because the actual rule action is followed always after adding an ALTQ tag. Slay.

🌈 RULE ACTIONS 🌈

A rule can be associated with one of the following actions, which will be executed when the packet matches the body of the rule 💖

allow | accept | pass | permit

Allow packets that match the rule. The search terminates ✋

check-state [:flowname | :any]

Checks the packet against the dynamic ruleset. If a match is found, execute the action associated with the rule which generated this dynamic rule; otherwise, move to the next rule. Check-state rules do not have a body 🫥

If no check-state rule is found, the dynamic ruleset is checked at the first keep-state or limit rule. The :flowname is a symbolic name assigned to the dynamic rule by the keep-state opcode. The special flowname :any can be used to ignore states flowname when matching. The :default keyword is a special name used for compatibility with old rulesets 👵

count

Update counters for all packets that match the rule. The search continues with the next rule 🧮

deny | drop

Discard packets that match this rule. The search terminates 💀

divert port

Divert packets that match this rule to the divert(4) socket bound to port port. The search terminates 🌀

fwd | forward ipaddr | tablearg[,port]

Change the next-hop on matching packets to ipaddr, which can be an IP address or a host name. The next hop can also be supplied by the last table looked up for the packet by using the tablearg keyword instead of an explicit address. The search terminates if this rule matches ✨

If ipaddr is a local address, then matching packets will be forwarded to port (or the port number in the packet if one is not specified in the rule) on the local machine.

If ipaddr is not a local address, then the port number (if specified) is ignored, and the packet will be forwarded to the remote address, using the route as found in the local routing table for that IP.

A fwd rule will not match layer2 packets (those received on ether_input, ether_output, or bridged) 🙅

The fwd action does not change the contents of the packet at all. In particular, the destination address remains unmodified, so packets forwarded to another system will usually be rejected by that system unless there is a matching rule on that system to capture them. For packets forwarded locally, the local address of the socket will be set to the original destination address of the packet. This makes the netstat(1) entry look rather weird, but it's intended for use with transparent proxy servers 👀

nat nat_nr | global | tablearg

Pass packet to a nat instance (for network address translation, address redirect, etc.): see the “NETWORK ADDRESS TRANSLATION (NAT)” Section for further information 🔁

nat64lsn name

Pass packet to a stateful NAT64 instance (for IPv6/IPv4 network address and protocol translation): see the “IPv6/IPv4 NETWORK ADDRESS AND PROTOCOL TRANSLATION” Section for further information 🌐

nat64stl name

Pass packet to a stateless NAT64 instance (for IPv6/IPv4 network address and protocol translation): see the “IPv6/IPv4 NETWORK ADDRESS AND PROTOCOL TRANSLATION” Section for further information 🌐

nat64clat name

Pass packet to a CLAT NAT64 instance (for client-side IPv6/IPv4 network address and protocol translation): see the “IPv6/IPv4 NETWORK ADDRESS AND PROTOCOL TRANSLATION” Section for further information 🌐

nptv6 name

Pass packet to a NPTv6 instance (for IPv6-to-IPv6 network prefix translation): see the “IPv6-to-IPv6 NETWORK PREFIX TRANSLATION (NPTv6)” Section for further information 🦋

pipe pipe_nr

Pass packet to a dummynet "pipe" (for bandwidth limitation, delay, etc.). See the “TRAFFIC SHAPER (DUMMYNET) CONFIGURATION” Section for further information. The search terminates; however, on exit from the pipe and if the sysctl(8) variable net.inet.ip.fw.one_pass is not set, the packet is passed again to the firewall code starting from the next rule 🚰

queue queue_nr

Pass packet to a dummynet "queue" (for bandwidth limitation using WF2Q+) 🎯

reject

(Deprecated). Synonym for unreach host. Like, it's old, bestie — noted and moved on 🫡

reset

Discard packets that match this rule, and if the packet is a TCP packet, try to send a TCP reset (RST) notice. The search terminates 🚫

reset6

Discard packets that match this rule, and if the packet is a TCP packet, try to send a TCP reset (RST) notice. The search terminates 🚫

skipto number | tablearg

Skip all subsequent rules numbered less than number. The search continues with the first rule numbered number or higher. It's possible to use the tablearg keyword with a skipto for a computed skipto. Skipto may work either in O(log(N)) or in O(1), depending on the amount of memory and/or sysctl variables. See the “SYSCTL VARIABLES” section for more details 🦘

call number | tablearg

The current rule number is saved in the internal stack, and ruleset processing continues with the first rule numbered number or higher. If later a rule with the return action is encountered, the processing returns to the first rule with the number of this call rule plus one or higher (the same behaviour as with packets returning from a divert(4) socket after a divert action). This could be used to make somewhat like an assembly language "subroutine" calls to rules with common checks for different interfaces, etc. 🧩

Rule with any number could be called — not just forward jumps as with skipto. So, to prevent endless loops in case of mistakes, both call and return actions don't do any jumps and simply go to the next rule if memory cannot be allocated or the stack overflowed/underflowed 🌀

Internally the stack for rule numbers is implemented using the mbuf_tags(9) facility and currently has size of 16 entries. As mbuf tags are lost when the packet leaves the kernel, divert should not be used in subroutines, to avoid endless loops and other undesired effects 😭

return 🎀

okay so this one grabs the rule number that the last call action stashed on the internal stack, and then yeets ruleset processing right back to the first rule with a number greater than the number of the matching call rule. the whole deets live in the call description, go read that bestie 💖

ngl, return rules usually wrap up a "subroutine", so like, they're unconditional — but the ipfw command-line utility currently makes every action except check-state have a body. sometimes you do only wanna return on some packets, but usually you just wanna print plain return for readability. the workaround? new syntax + the -c switch:

# Add a rule without actual body
ipfw add 2999 return via any

# List rules without "from any to any" part
ipfw -c list

this cosmetic annoyance may get fixed in future releases. we'll see 🙈

tee port 🫶

sends a copy of every packet matching this rule off to the divert(4) socket bound to port. and the search keeps going with the next rule, no drama.

unreach code [mtu] 💀

discards packets matching this rule and tries to fire off an ICMP unreachable notice with code code, where code is a number from 0 to 255, or one of these aliases: net, host, protocol, port, needfrag, srcfail, net-unknown, host-unknown, isolated, net-prohib, host-prohib, tosnet, toshost, filter-prohib, host-precedence or precedence-cutoff. the needfrag code can take an optional mtu parameter — if you specify it, the MTU value gets dropped into the generated ICMP packet. and then the search terminates. periodt.

unreach6 code 🧚

same vibe but ICMPv6: discards matching packets and tries to send an ICMPv6 unreachable notice with code code, where code is a number from 0, 1, 3 or 4, or one of these aliases: no-route, admin-prohib, address or port. search terminates.

diverts the packet into netgraph with the given cookie. search terminates. if the packet comes back from netgraph later, it's either accepted or continues with the next rule — depends on the net.inet.ip.fw.one_pass sysctl variable. iykyk ✨

a copy of the packet gets diverted into netgraph, and the original packet just carries on with the next rule. see ng_ipfw(4) for more info on netgraph and ngtee actions.

setfib fibnum | tablearg 💜

tags the packet so it uses the FIB (routing table) fibnum in any subsequent forwarding decisions. in the current implementation this is limited to values 0 through 15, see setfib(2). processing continues at the next rule. you can use the tablearg keyword with setfib — but if the tablearg value isn't within the compiled range of fibs, the packet's fib gets set to 0. no cap.

setdscp DSCP | number | tablearg 👑

sets the specified DiffServ codepoint for an IPv4/IPv6 packet. processing continues at the next rule. supported values are:

cs0 (000000), cs1 (001000), cs2 (010000), cs3 (011000), cs4 (100000), cs5 (101000), cs6 (110000), cs7 (111000), af11 (001010), af12 (001100), af13 (001110), af21 (010010), af22 (010100), af23 (010110), af31 (011010), af32 (011100), af33 (011110), af41 (100010), af42 (100100), af43 (100110), va (101100), ef (101110), be (000000).

plus you can just slap in a DSCP value by number (0..63). and yes, the tablearg keyword works with setdscp too — if the tablearg value isn't within the 0..63 range, the lower 6 bits of the supplied value get used. math is math 💅

tcp-setmss mss 🔥

sets the Maximum Segment Size (MSS) in the TCP segment to mss. heads up: the kernel module ipfw_pmod needs to be loaded, or your kernel needs options IPFIREWALL_PMOD, to use this action at all. this command does not change a packet if the original MSS value is lower than the specified value. both TCP over IPv4 and over IPv6 are supported. and regardless of whether the packet matched the tcp-setmss rule or not, the search continues with the next rule.

reass 🌈

queue and reassemble IPv4 fragments. if the packet isn't fragmented, counters get updated and processing continues with the next rule. if it's the last logical fragment, the packet is reassembled and — if net.inet.ip.fw.one_pass is set to 0 — processing continues with the next rule; otherwise it's allowed to pass and the search terminates. and if it's a fragment sitting in the middle of a logical group of fragments, it gets consumed and processing stops immediately. brutal but fair 💀

fragment handling can be tuned via net.inet.ip.maxfragpackets and net.inet.ip.maxfragsperpacket, which limit — respectively — the maximum number of processable fragments (default: 800) and the maximum number of fragments per packet (default: 16).

NOTA BENE: since fragments don't contain port numbers, they should be avoided with the reass rule. alternatively, direction-based (like in / out) and source-based (like via) match patterns can be used to select fragments. 🙈

usually a simple rule like this:

# reassemble incoming fragments
ipfw add reass all from any to any in

is literally all you need at the beginning of your ruleset. iconic.

abort ⭐

discards packets that match this rule, and if the packet is an SCTP packet, tries to send an SCTP packet containing an ABORT chunk. the search terminates.

abort6 ⭐

discards packets that match this rule, and if the packet is an SCTP packet, tries to send an SCTP packet containing an ABORT chunk. the search terminates.

🎀 RULE BODY 🎀

so the body of a rule is zero or more patterns (specific source and destination addresses or ports, protocol options, incoming or outgoing interfaces, etc.) that the packet has to match to even get recognised. in general the patterns are connected by (implicit) and operators — i.e. all of them must match for the rule to match. individual patterns can be prefixed by the not operator to flip the result, like:

ipfw add 100 allow ip from not 1.2.3.4 to any

and you can also build sets of alternative match patterns (or-blocks) by putting the patterns in lists wrapped in parentheses ( ) or braces { }, using the or operator like this:

ipfw add 100 allow ip from { x or not y or z } to any

only one level of parentheses is allowed, bestie. and beware — most shells give parentheses and braces special meanings, so it's advisable to slap a backslash \ in front of them so the shell doesn't do its own thing 😭

the body of a rule must in general include a source and destination address specifier. the keyword any can be used in various places to say "the content of this required field is irrelevant, don't worry about it".

the rule body has this format:

[proto from src to dst] [options]

the first part (proto from src to dst) is there for backward compatibility with earlier versions of FreeBSD. in modern FreeBSD any match pattern (including MAC headers, IP protocols, addresses and ports) can be specified in the options section. glow-up fr fr ✨

rule fields mean the following:

proto: protocol | { protocol or ... } 💖

protocol: [not] protocol-name | protocol-number

an IP protocol specified by number or name (for a complete list see /etc/protocols), or one of these keywords:

  • ip4 | ipv4 — matches IPv4 packets.
  • ip6 | ipv6 — matches IPv6 packets.
  • ip | all — matches any packet.

heads up: the ipv6 in proto option will be treated as inner protocol. and the ipv4 is not available in proto option. 🙈

the { protocol or ... } format (an or-block) is provided for convenience only, but its use is deprecated. consider this your gentle warning 🫶

src and dst: {addr | { addr or ... }} [[not] ports] 🦋

an address (or a list, see below) optionally followed by ports specifiers. the second format (or-block with multiple addresses) is provided for convenience only and its use is discouraged.

addr: [not] {any | me | me6 | table-ref | addr-list | addr-set} 🌸

  • any — matches any IP address.
  • me — matches any IP address configured on an interface in the system.
  • me6 — matches any IPv6 address configured on an interface in the system. the address list is evaluated at the time the packet is analysed.

table-ref: 👑

a table lookup can be specified in one of the following ways:

  • table(name) — matches any IPv4 or IPv6 address for which an entry exists in the lookup table number.
  • table(name,value) — matches any IPv4 or IPv6 address for which an entry exists in the lookup table number and the 32-bit unsigned value specified matches the entry value.
  • table(name,value-type=value) — matches any IPv4 or IPv6 address for which an entry exists in the lookup table number and the 32-bit unsigned value specified matches the corresponding value-type field for the record found.

see the "LOOKUP TABLES" section below for more info on lookup tables 💜

addr-list: ip-addr[,addr-list] 🎀

ip-addr: 💅

a host or subnet address specified in one of the following ways:

  • numeric-ip | hostname — matches a single IPv4 address, specified as dotted-quad or a hostname. hostnames are resolved at the time the rule is added to the firewall list.
  • addr/masklen — matches all addresses with base addr (specified as an IP address, a network number, or a hostname) and mask width of masklen bits. example era: 1.2.3.4/25 or 1.2.3.0/25 will match all IP numbers from 1.2.3.0 to 1.2.3.127.
  • addr:mask — matches all addresses with base addr (specified as an IP address, a network number, or a hostname) and the mask of mask, specified as a dotted quad. as an example, 1.2.3.4:255.0.255.0 or 1.0.3.0:255.0.255.0 will match 1.*.3.*. this form is advised only for non-contiguous masks — it's better to resort to the addr/masklen format for contiguous masks, which is more compact and less error-prone. trust, it's giving cleaner 💖
addr-set: addr[/masklen]{list}

list: {num | num-num}[,list]

Okay so addr-set is that girl for sparse address sets — it matches every address whose base is addr (an IP address, a network number, or a hostname) and whose last byte is in the list between the braces { }. No cap: there must be no spaces between the braces and the numbers (spaces after commas are allowed, bestie). Elements of the list can be single entries or ranges. The masklen field limits how big the set of addresses can be, and it can be any value between 24 and 32. If you don’t specify it, it’s assumed to be 24. This format is lowkey iconic for handling sparse address sets inside a single rule. Because the matching happens using a bitmask, it takes constant time and dramatically reduces ruleset complexity. Example time: an address specified as 1.2.3.4/24{128,35-55,89} or 1.2.3.0/24{128,35-55,89} will match these IP addresses: 1.2.3.128, 1.2.3.35 to 1.2.3.55, and 1.2.3.89. slay ✨

addr6-list: ip6-addr[,addr6-list]

ip6-addr:

A host or subnet is specified in one of these ways:

  • numeric-ip | hostname — Matches a single IPv6 address as allowed by inet_pton(3), or a hostname. Hostnames are resolved at the time the rule is added to the firewall list.
  • addr/masklen — Matches all IPv6 addresses with base addr (specified as allowed by inet_pton(3) or a hostname) and mask width of masklen bits.
  • addr/mask — Matches all IPv6 addresses with base addr (specified as allowed by inet_pton(3) or a hostname) and the mask of mask, specified as allowed by inet_pton(3). For example, fe::640:0:0/ffff::ffff:ffff:0:0 will match fe:*:*:*:0:640:*:*. This form is advised only for non-contiguous masks. It’s better to use the addr/masklen format for contiguous masks — it’s more compact and less error-prone.

No support for sets of IPv6 addresses is provided, because IPv6 addresses are typically random past the initial prefix. iykyk 💅

ports: {port | port-port}[,ports]

For protocols that support port numbers (such as SCTP, TCP and UDP), optional ports may be specified as one or more ports or port ranges, separated by commas but no spaces, and an optional not operator. The - notation specifies a range of ports (including boundaries). Service names from /etc/services may be used instead of numeric port values. The length of the port list is limited to 30 ports or ranges, though you can specify larger ranges by using an or-block in the options section of the rule.

A backslash (\) can be used to escape the dash (-) character in a service name (from a shell, the backslash must be typed twice to avoid the shell itself interpreting it as an escape character).

ipfw add count tcp from any ftp\\-data-ftp to any

Fragmented packets which have a non-zero offset (i.e., not the first fragment) will never match a rule which has one or more port specifications. See the frag option for details on matching fragmented packets. important caveat, bestie — don’t forget it. 😭

✨ RULE OPTIONS (MATCH PATTERNS) ✨

Additional match patterns can be used within rules. Zero or more of these so-called options can be present in a rule, optionally prefixed by the not operand, and possibly grouped into or-blocks. vibe check: stack ’em, negate ’em, group ’em. 🌸

The following match patterns can be used (listed in alphabetical order):

  • // this is a comment. — Inserts the specified text as a comment in the rule. Everything following // is considered as a comment and stored in the rule. You can have comment-only rules, which are listed as having a count action followed by the comment.
  • bridged — Alias for layer2.
  • defer-immediate-action, defer-action — A rule with this option will not perform normal action upon a match. This option is intended to be used with record-state or keep-state as the dynamic rule, created but ignored on match, will work as intended. Rules with both record-state and defer-immediate-action create a dynamic rule and continue with the next rule without actually performing the action part of this rule. When the rule is later activated via the state table, the action is performed as usual.
  • diverted — Matches only packets generated by a divert socket.
  • diverted-loopback — Matches only packets coming from a divert socket back into the IP stack input for delivery.
  • diverted-output — Matches only packets going from a divert socket back outward to the IP stack output for delivery.
  • dst-ip ip-address — Matches IPv4 packets whose destination IP is one of the address(es) specified as argument.
  • {dst-ip6 | dst-ipv6} ip6-address — Matches IPv6 packets whose destination IP is one of the address(es) specified as argument.
  • dst-port ports — Matches IP packets whose destination port is one of the port(s) specified as argument.
  • established — Matches TCP packets that have the RST or ACK bits set.
  • ext6hdr header — Matches IPv6 packets containing the extended header given by header. Supported headers are: Fragment, (frag), Hop-to-hop options (hopopt), any type of Routing Header (route), Source routing Routing Header Type 0 (rthdr0), Mobile IPv6 Routing Header Type 2 (rthdr2), Destination options (dstopt), IPSec authentication headers (ah), and IPsec encapsulating security payload headers (esp).
  • fib fibnum — Matches a packet that has been tagged to use the given FIB (routing table) number.
  • flow table-ref — Search for the flow entry in lookup table specified by table-ref. If not found, the match fails. Otherwise, the match succeeds and tablearg is set to the value extracted from the table. This option can be useful to quickly dispatch traffic based on certain packet fields. See the “LOOKUP TABLES” section below for more information on lookup tables.
  • flow-id labels — Matches IPv6 packets containing any of the flow labels given in labels. labels is a comma separated list of numeric flow labels.
  • dst-mac table-ref — Search for the destination MAC address entry in lookup table specified by table-ref. If not found, the match fails. Otherwise, the match succeeds and tablearg is set to the value extracted from the table.
  • src-mac table-ref — Search for the source MAC address entry in lookup table specified by table-ref. If not found, the match fails. Otherwise, the match succeeds and tablearg is set to the value extracted from the table.
  • frag spec — Matches IPv4 packets whose ip_off field contains the comma separated list of IPv4 fragmentation options specified in spec. The recognized options are: df (don't fragment), mf (more fragments), rf (reserved fragment bit), offset (non-zero fragment offset). The absence of a particular option may be denoted with a !. An empty list of options defaults to matching on non-zero fragment offset. Such a rule would match every datagram that isn’t the first fragment, for both IPv4 and IPv6. This is backward compatibility with older rulesets.
  • gid group — Matches all TCP or UDP packets sent by or received for a group. A group may be specified by name or number.
  • jail jail — Matches all TCP or UDP packets sent by or received for the jail whose ID or name is jail.
  • icmptypes types — Matches ICMP packets whose ICMP type is in the list types. The list may be specified as any combination of individual types (numeric) separated by commas. Ranges are not allowed. The supported ICMP types are: echo reply (0), destination unreachable (3), source quench (4), redirect (5), echo request (8), router advertisement (9), router solicitation (10), time-to-live exceeded (11), IP header bad (12), timestamp request (13), timestamp reply (14), information request (15), information reply (16), address mask request (17) and address mask reply (18).
  • icmp6types types — Matches ICMP6 packets whose ICMP6 type is in the list of types. The list may be specified as any combination of individual types (numeric) separated by commas. Ranges are not allowed.
  • in, out — Matches incoming or outgoing packets, respectively. in and out are mutually exclusive (in fact, out is implemented as not in).
  • ipid id-list — Matches IPv4 packets whose ip_id field has value included in id-list, which is either a single value or a list of values or ranges specified in the same way as ports.
  • iplen len-list — Matches IP packets whose total length, including header and data, is in the set len-list, which is either a single value or a list of values or ranges specified in the same way as ports.
  • ipoptions spec — Matches packets whose IPv4 header contains the comma separated list of options specified in spec. The supported IP options are: ssrr (strict source route), lsrr (loose source route), rr (record packet route) and ts (timestamp). The absence of a particular option may be denoted with a !.
  • ipprecedence precedence — Matches IPv4 packets whose precedence field is equal to precedence.
  • ipsec — Matches packets that have IPSEC history associated with them (i.e., the packet comes encapsulated in IPSEC, the kernel has IPSEC support, and can correctly decapsulate it). Note that specifying ipsec is different from specifying proto ipsec as the latter will only look at the specific IP protocol field, irrespective of IPSEC kernel support and the validity of the IPSEC data. Further note that this flag is silently ignored in kernels without IPSEC support. It does not affect rule processing when given and the rules are handled as if with no ipsec flag. 🫶
  • iptos spec — okay so this one checks IPv4 packets whose tos field contains the comma-separated list of service types you put in spec. 💅 The supported IP types of service are lowdelay (IPTOS_LOWDELAY), throughput (IPTOS_THROUGHPUT), reliability (IPTOS_RELIABILITY), mincost (IPTOS_MINCOST), and congestion (IPTOS_ECN_CE). You can say “nah, not this one” by putting a ! in front of the type — the absence of a particular type may be denoted with a !. ✨
  • dscp spec[,spec] — matches IPv4/IPv6 packets whose DS field value is contained in the spec mask. You can pass multiple values as a comma-separated list, and each value can be one of the keywords used in the setdscp action or an exact number. 🎀
  • ipttl ttl-list — matches IPv4 packets whose time to live is included in ttl-list. That list can be a single value, or a list of values/ranges written the same way as ports. ⏳
  • ipversion ver — matches IP packets whose IP version field is ver. 🦋
  • keep-state [:flowname] — upon a match, the firewall creates a dynamic rule. Its default behaviour is to match bidirectional traffic between source and destination IP/port using the same protocol. 👑 The rule has a limited lifetime (controlled by a set of sysctl(8) variables), and that lifetime is refreshed every time a matching packet is found. The :flowname assigns additional addresses, ports, and protocol parameters to the dynamic rule. It can be used for more accurate matching by a check-state rule. The :default keyword is a special name used for compatibility with old rulesets. 💖
  • layer2 — matches only layer2 packets, i.e., those passed to ipfw from ether_demux() and ether_output_frame(). 🧚
  • limit {src-addr | src-port | dst-addr | dst-port} N [:flowname] — the firewall will only allow N connections with the same set of parameters you specified in the rule. You can specify one or more of source and destination addresses and ports. 🚦
  • lookup {dst-ip | dst-ip4 | dst-ip6 | dst-port | dst-mac | src-ip | src-ip4 | src-ip6 | src-port | src-mac | uid | jail | dscp | mark | rulenum} [:bitmask] name — searches for an entry in lookup table name that matches the field you specified as the argument. If it’s not found, the match fails. If it is found, the match succeeds and tablearg is set to the value extracted from the table. 🔎
    • If you add an optional bitmask, the field’s value gets altered by a bitwise AND with the bitmask, and the resulting value is what gets searched instead of the original one. The bitmask accepts these formats:
      1. Dotted-quad form, e.g. 127.88.34.0. This form works for IPv4 lookups as well as all numeric lookup types.
      2. A 32-bit number, e.g. 0xf00baa1 or 255. This form also works for IPv4 lookups as well as all numeric lookup types.
      3. An IPv6 address when specified along with the dst-ip6 or src-ip6 field. If you use it, the rule will match IPv6 packets only. Example: src-ip6:afff:ff00:ffff:ffff:0:0:0:0f0f.
      4. An Ethernet MAC address when specified along with the dst-mac or src-mac field. E.g. 00:11:22:33:44:55.
    • The bitmask can not be specified for dst-ip or src-ip, because these field specifiers look up both IPv4 and IPv6 addresses. 🚫
    • This option can be useful for quickly dispatching traffic based on certain packet fields. The bitmask lets you implement wildcard lookups by inserting a masked prefix into the table and applying the bitmask on each lookup.
    • See the “LOOKUP TABLES” section below for more info on lookup tables. 📚
  • { MAC | mac } dst-mac src-mac — matches packets with the given dst-mac and src-mac addresses. You can specify them as the any keyword (which matches any MAC address), or as six groups of hex digits separated by colons, optionally followed by a mask that says which bits are significant. The mask can be specified using either of these methods: 🎀
    1. A slash (/) followed by the number of significant bits. For example, an address with 33 significant bits can be written as:
MAC 10:20:30:40:50:60/33 any
  1. An ampersand (&) followed by a bitmask specified as six groups of hex digits separated by colons. For example, an address where the last 16 bits are significant can be written as:
MAC 10:20:30:40:50:60&00:00:00:00:ff:ff any
  • Note that the ampersand character has a special meaning in many shells and should generally be escaped. 😭
Heads up, bestie: the order of MAC addresses is destination first, source second — same as on the wire, but the opposite of the order used for IP addresses. 💅
  • mac-type mac-type — matches packets whose Ethernet Type field corresponds to one of the values you specify as the argument. mac-type is written the same way as port numbers (i.e., one or more comma-separated single values or ranges). You can use symbolic names for known values such as vlan, ipv4, and ipv6. Values can be entered as decimal or hexadecimal (if prefixed by 0x), and they are always printed as hexadecimal (unless the -N option is used, in which case symbolic resolution will be attempted). 🌈
  • proto protocol — matches packets with the corresponding IP protocol. 🦋
  • record-state — upon a match, the firewall creates a dynamic rule as if keep-state was specified. However, this option does not imply an implicit check-state, in contrast to keep-state. 🔥
  • recv, xmit, via {ifX | ifmask | table-ref | ipno | any} — matches packets received, transmitted, or going through (respectively) the interface specified by exact name (ifX), by device mask (ifmask), by IP address, or through some interface. 📡
    • An interface name can be matched against ifmask with fnmatch(3) according to the rules used by the shell (f.e. tun*). See also the “EXAMPLES” section.
    • A lookup table specified by table-ref can be used to match an interface by its kernel ifindex. See the “LOOKUP TABLES” section below for more info on lookup tables.
    • The via keyword causes the interface to always be checked. If you use recv or xmit instead of via, then only the receive or transmit interface (respectively) is checked. By specifying both, you can match packets based on both receive and transmit interface, e.g.:
ipfw add deny ip from any to any out recv ed0 xmit ed1
  • The recv interface can be tested on either incoming or outgoing packets, while the xmit interface can only be tested on outgoing packets. So out is required (and in is invalid) whenever xmit is used. ⚠️
  • A packet might not have a receive or transmit interface: packets originating from the local host have no receive interface, while packets destined for the local host have no transmit interface. 🫶
  • set-limit {src-addr | src-port | dst-addr | dst-port} N — works like limit but does not have an implicit check-state attached to it. 🚫
  • setup — matches TCP packets that have the SYN bit set but no ACK bit. This is the short form of tcpflags syn,!ack. 🔥
  • sockarg — matches packets that are associated to a local socket and for which the SO_USER_COOKIE socket option has been set to a non-zero value. As a side effect, the value of the option is made available as a tablearg value, which in turn can be used as a skipto or pipe number. 🍪
  • src-ip ip-address — matches IPv4 packets whose source IP is one of the address(es) specified as an argument. 💖
  • src-ip6 ip6-address — matches IPv6 packets whose source IP is one of the address(es) specified as an argument. 💜
  • src-port ports — matches IP packets whose source port is one of the port(s) specified as argument. 🔌
  • tagged tag-list — matches packets whose tags are included in tag-list, which is either a single value or a list of values or ranges specified in the same way as ports. Tags can be applied to the packet using the tag rule action parameter (see its description for details on tags). 🏷️
  • mark value[:bitmask] | tablearg[:bitmask] — matches packets whose mark is equal to value with an optional bitmask applied to it. You can also use tablearg instead of an explicit value to match a value supplied by the last table lookup. Both value and bitmask can be entered as decimal or hexadecimal (if prefixed by 0x), and they are always printed as hexadecimal. ✨
  • tcpack ack — TCP packets only. Match if the TCP header acknowledgment number field is set to ack. ✅
  • tcpdatalen tcpdatalen-list — matches TCP packets whose length of TCP data is tcpdatalen-list, which is either a single value or a list of values or ranges specified in the same way as ports. 📦
  • tcpflags spec — TCP packets only. Match if the TCP header contains the comma-separated list of flags specified in spec. The supported TCP flags are fin, syn, rst, psh, ack, and urg. The absence of a particular flag may be denoted with a !. A rule which contains a tcpflags specification can never match a fragmented packet which has a non-zero offset. See the frag option for details on matching fragmented packets. 🚩
  • tcpmss tcpmss-list — matches TCP packets whose MSS (maximum segment size) value is set to tcpmss-list, which is either a single value or a list of values or ranges specified in the same way as ports. 📏
  • tcpseq seq — TCP packets only. Match if the TCP header sequence number field is set to seq. 🔢
  • tcpwin tcpwin-list — matches TCP packets whose header window field is set to tcpwin-list, which is either a single value or a list of values or ranges specified in the same way as ports. 🪟
  • tcpoptions spec — TCP packets only. Match if the TCP header contains the comma-separated list of options specified in spec. The supported TCP options are:

…the tcp option parade wraps up with mss (maximum segment size), window (tcp window advertisement), sack (selective ack), ts (rfc1323 timestamp) and cc (rfc1644 t/tcp connection count) 🎀 and bestie, plot twist: the absence of a particular option may be denoted with a ! — so you can literally say "not this one" 💅

  • uid user — matches all TCP or UDP packets sent by or received for a user. You can match a user by name or by identification number, whichever you vibe with 🫶
  • verrevpath — for incoming packets, a routing table lookup is done on the packet's source address. If the interface the packet entered the system on matches the outgoing interface for that route → the packet matches ✨. If the interfaces don't line up → it does not match 💀. All outgoing packets, or packets with no incoming interface, match.

The name and functionality are intentionally similar to the Cisco IOS command:

ip verify unicast reverse-path

Use this to build anti-spoofing rules that reject all packets whose source addresses aren't from this interface. Also peep the antispoof option below 👀

  • versrcreach — for incoming packets, a routing table lookup happens on the packet's source address. If a route to the source address exists — but not the default route, and not a blackhole/reject route — then the packet matches. Otherwise, nope, no match 🙈. All outgoing packets match.

The name and functionality are intentionally similar to the Cisco IOS command:

ip verify unicast source reachable-via any

Iconic for anti-spoofing rules that reject every packet whose source address is unreachable 🔥

  • antispoof — for incoming packets, the packet's source address gets checked to see if it belongs to a directly connected network. If the network is directly connected, then the interface the packet came in on is compared against the interface that network is connected to. When the incoming interface and the directly connected interface are not the same → the packet does not match. Otherwise → it matches 💖. All outgoing packets match.

Use this for anti-spoofing rules that reject all packets pretending to be from a directly connected network while not actually coming in through that interface 🙅‍♀️ This one is similar to verrevpath but more restricted, because it only engages on packets whose source addresses belong to directly connected networks, instead of all source addresses. No cap, it's the pickier queen 👑

🔍 LOOKUP TABLES 🔍

Lookup tables are lowkey the moment for handling large sparse sets of addresses or other search keys (think: ports, jail IDs, interface names) 🦋 For the rest of this section we're gonna use the term "key" — periodt.

Table name needs to match this spec: table-name. Tables with the same name can be created in different sets. However, rules link to the tables in set 0 by default 😳 This behavior can be controlled by the net.inet.ip.fw.tables_sets variable. Peep the "SETS OF RULES" section for more info. And bestie, there may be up to 65535 different lookup tables 🌈

The following table types are supported:

table-type: addr | iface | number | flow | mac

table-key: addr[/masklen] | iface-name | number | flow-spec

flow-spec: flow-field[,flow-spec]

flow-field: src-ip | proto | src-port | dst-ip | dst-port
  • addr — matches IPv4 or IPv6 address. Each entry is represented by an addr[/masklen] and will match all addresses with base addr (specified as an IPv4/IPv6 address, or a hostname) and mask width of masklen bits ✨ If masklen isn't specified, it defaults to 32 for IPv4 and 128 for IPv6. When looking up an IP address in a table, the most specific entry will match 🎯
  • iface — matches interface names. Each entry is represented by a string treated as an interface name. Wildcards are not supported 🚫
  • number — matches protocol ports, uids/gids or jail IDs. Each entry is represented by a 32-bit unsigned integer. Ranges are not supported 🙈
  • flow — matches packet fields specified by flow type suboptions against table entries 🌊
  • mac — matches MAC address. Each entry is represented by an addr[/masklen] and will match all addresses with base addr and mask width of masklen bits. If masklen isn't specified, it defaults to 48. When looking up a MAC address in a table, the most specific entry will match 💅
Tables require explicit creation via create before use. No spontaneous tables, bestie — you gotta manifest them first 💫

The following creation options are supported:

create-options: create-option | create-options

create-option: type table-type | valtype value-mask | algo algo-desc |
        limit number | locked | missing | or-flush
  • type — table key type.
  • valtype — table value mask.
  • algo — table algorithm to use (see below 👇).
  • limit — maximum number of items that may be inserted into table.
  • locked — restrict any table modifications 🔒
  • missing — do not fail if the table already exists and has exactly the same options as the new one.
  • or-flush — flush the existing table with the same name instead of returning an error. Implies missing, so the existing table must be compatible with the new one 🫶

Some of these options may be modified later via the modify keyword. The following options can be changed:

modify-options: modify-option | modify-options

modify-option: limit number
  • limit — alter the maximum number of items that may be inserted into table.

Additionally, a table can be locked or unlocked using the lock or unlock commands 🔐

Tables of the same type can be swapped with each other using the swap name command. Swap may fail if tables have limits set and the data exchange would result in limits being hit 💀 The operation is performed atomically — it's all-or-nothing, main character energy ✨

One or more entries can be added to a table at once using the add command. Addition of all items is performed atomically. By default, an error in adding one entry does not influence the addition of other entries. However, a non-zero error code is returned in that case 😭 A special atomic keyword may be specified before add to indicate an all-or-none add request 💅

One or more entries can be removed from a table at once using the delete command. By default, an error in removing one entry does not influence the removal of other entries. However, a non-zero error code is returned in that case 🙃

It may be possible to check what entry will be found for a particular table-key using the lookup table-key command. This functionality is optional and may be unsupported in some algorithms — no promises, bestie 🧚

The following operations can be performed on one or all tables:

  • list — list all entries 📋
  • flush — removes all entries 🧹
  • info — shows generic table information ℹ️
  • detail — shows generic table information and algo-specific data 🔍

The following lookup algorithms are supported:

algo-desc: algo-name | algo-name algo-data

algo-name: addr: radix | addr: hash | iface: array | number: array |
        flow: hash | mac: radix
  • addr: radix — separate Radix trees for IPv4 and IPv6, the same way as the routing table (see route(4)). Default choice for addr type 🌳
  • addr:hash — separate auto-growing hashes for IPv4 and IPv6. Accepts entries with the same mask length specified initially via addr:hash masks=/v4,/v6 algorithm creation options. Assumes /32 and /128 masks by default. Search removes host bits (according to mask) from the supplied address and checks the resulting key in the appropriate hash. Mostly optimized for /64 and byte-ranged IPv6 masks ⚡
  • iface:array — array storing sorted indexes for entries which are presented in the system. Optimized for very fast lookup 🚀
  • number:array — array storing sorted u32 numbers 🔢
  • flow:hash — auto-growing hash storing flow entries. Search calculates the hash on required packet fields and searches for matching entries in the selected bucket 🌊
  • mac: radix — Radix tree for MAC address 🌲

The tablearg feature provides the ability to use a value, looked up in the table, as the argument for a rule action, action parameter or rule option 😱 This can significantly reduce the number of rules in some configurations — less rules, more slay 💅 If two tables are used in a rule, the result of the second (destination) is used.

Each record may hold one or more values according to value-mask. This mask is set on table creation via the valtype option. The following value types are supported:

value-mask: value-type[,value-mask]

value-type: skipto | pipe | fib | nat | dscp | tag | divert |
        netgraph | limit | ipv4 | ipv6 | mark
  • skipto — rule number to jump to.
  • pipe — Pipe number to use.
  • fib — fib number to match/set.
  • nat — nat number to jump to.
  • dscp — dscp value to match/set.
  • tag — tag number to match/set.
  • divert — port number to divert traffic to.
  • netgraph — hook number to move packet to.
  • limit — maximum number of connections.
  • ipv4 — IPv4 nexthop to fwd packets to.
  • ipv6 — IPv6 nexthop to fwd packets to.
  • mark — mark value to match/set.

The tablearg argument can be used with the following actions: nat, pipe, queue, divert, tee, netgraph, ngtee, fwd, skipto, setfib, setmark, action parameters: tag, untag, rule options: limit, tagged, mark.

⚠️ Heads up bestie: when used with the skipto action, the user should be aware that the code will walk the ruleset up to a rule equal to, or past, the given number. So like… it's a stroll, not a teleport 👀

See the "EXAMPLES" Section for example usage of tables and the tablearg keyword 🎀

🎀 SETS OF RULES 🎀

Okay so every rule or table belongs to one of 32 different sets, numbered 0 to 31. Set 31 is reserved for the default rule, periodt. 👑

By default, rules or tables are put in set 0, unless you use the set N attribute when adding a new rule or table. Sets can be individually and atomically enabled or disabled, so this mechanism permits an easy way to store multiple configurations of the firewall and quickly (and atomically) switch between them. Lowkey iconic behavior. 💅

By default, tables from set 0 are referenced when adding rule with table opcodes regardless of rule set. This behavior can be changed by setting net.inet.ip.fw.tables_sets variable to 1. Rule's set will then be used for table references. ✨

The command to enable/disable sets is:

ipfw set [disable number ...] [enable number ...]

where multiple enable or disable sections can be specified. Command execution is atomic on all the sets specified in the command. By default, all sets are enabled. Slay, atomicity. 🫶

When you disable a set, its rules behave as if they do not exist in the firewall configuration, with only one exception:

dynamic rules created from a rule before it had been disabled will still be active until they expire. In order to delete dynamic rules you have to explicitly delete the parent rule which generated them. 😭

The set number of rules can be changed with the command:

ipfw set move {rule rule-number | old-set} to new-set

Also, you can atomically swap two rulesets with the command:

ipfw set swap first-set second-set

See the EXAMPLES Section on some possible uses of sets of rules. iykyk. 🦋

💖 STATEFUL FIREWALL 💖

Stateful operation is a way for the firewall to dynamically create rules for specific flows when packets that match a given pattern are detected. Support for stateful operation comes through the check-state, keep-state, record-state, limit and set-limit options of rules. ✨

Dynamic rules are created when a packet matches a keep-state, record-state, limit or set-limit rule, causing the creation of a dynamic rule which will match all and only packets with a given protocol between a src-ip/src-port dst-ip/dst-port pair of addresses (src and dst are used here only to denote the initial match addresses, but they are completely equivalent afterwards). Rules created by keep-state option also have a :flowname taken from it. This name is used in matching together with addresses, ports and protocol. Dynamic rules will be checked at the first check-state, keep-state or limit occurrence, and the action performed upon a match will be the same as in the parent rule. No cap, that's the whole flow. 🌸

Note that no additional attributes other than protocol and IP addresses and ports and :flowname are checked on dynamic rules. That's the vibe check, bestie. 💀

The typical use of dynamic rules is to keep a closed firewall configuration, but let the first TCP SYN packet from the inside network install a dynamic rule for the flow so that packets belonging to that session will be allowed through the firewall:

ipfw add check-state :OUTBOUND
ipfw add allow tcp from my-subnet to any setup keep-state
:OUTBOUND
ipfw add deny tcp from any to any

A similar approach can be used for UDP, where an UDP packet coming from the inside will install a dynamic rule to let the response through the firewall:

ipfw add check-state :OUTBOUND
ipfw add allow udp from my-subnet to any keep-state :OUTBOUND
ipfw add deny udp from any to any

Dynamic rules expire after some time, which depends on the status of the flow and the setting of some sysctl variables. See Section “SYSCTL VARIABLES” for more details. For TCP sessions, dynamic rules can be instructed to periodically send keepalive packets to refresh the state of the rule when it is about to expire. Obsessed with a self-refreshing rule ngl. 🎀

See Section “EXAMPLES” for more examples on how to use dynamic rules. 🧚

🦋 TRAFFIC SHAPER (DUMMYNET) CONFIGURATION 🦋

ipfw is also the user interface for the dummynet traffic shaper, packet scheduler and network emulator, a subsystem that can artificially queue, delay or drop packets emulating the behaviour of certain network links or queueing systems. It's giving network simulation main character energy. 🌈

dummynet operates by first using the firewall to select packets using any match pattern that can be used in ipfw rules. Matching packets are then passed to either of two different objects, which implement the traffic regulation:

  • pipe — A pipe emulates a link with given bandwidth and propagation delay, driven by a FIFO scheduler and a single queue with programmable queue size and packet loss rate. Packets are appended to the queue as they come out from ipfw, and then transferred in FIFO order to the link at the desired rate.
  • queue — A queue is an abstraction used to implement packet scheduling using one of several packet scheduling algorithms. Packets sent to a queue are first grouped into flows according to a mask on the 5-tuple. Flows are then passed to the scheduler associated to the queue, and each flow uses scheduling parameters (weight and others) as configured in the queue itself. A scheduler in turn is connected to an emulated link, and arbitrates the link's bandwidth among backlogged flows according to weights and to the features of the scheduling algorithm in use.

In practice, pipes can be used to set hard limits to the bandwidth that a flow can use, whereas queues can be used to determine how different flows share the available bandwidth. Highkey useful. 🔥

A graphical representation of the binding of queues, flows, schedulers and links is below. 💜

                              (flow_mask|sched_mask)  sched_mask
                      +---------+   weight Wx  +-------------+
                      |         |->-[flow]-->--|             |-+
                 -->--| QUEUE x |   ...        |             | |
                      |         |->-[flow]-->--| SCHEDuler N | |
                      +---------+              |             | |
                          ...                  |             +--[LINK N]-->--
                      +---------+   weight Wy  |             | +--[LINK N]-->--
                      |         |->-[flow]-->--|             | |
                 -->--| QUEUE y |   ...        |             | |
                      |         |->-[flow]-->--|             | |
                      +---------+              +-------------+ |
                                                 +-------------+

It is important to understand the role of the SCHED_MASK and FLOW_MASK, which are configured through the commands:

ipfw sched N config mask SCHED_MASK ...

and

ipfw queue X config mask FLOW_MASK ....

The SCHED_MASK is used to assign flows to one or more scheduler instances, one for each value of the packet's 5-tuple after applying SCHED_MASK. As an example, using src-ip 0xffffff00 creates one instance for each /24 destination subnet. 🦋

The FLOW_MASK, together with the SCHED_MASK, is used to split packets into flows. As an example, using src-ip 0x000000ff together with the previous SCHED_MASK makes a flow for each individual source address. In turn, flows for each /24 subnet will be sent to the same scheduler instance. 🌸

The above diagram holds even for the pipe case, with the only restriction that a pipe only supports a SCHED_MASK, and forces the use of a FIFO scheduler (these are for backward compatibility reasons; in fact, internally, a dummynet's pipe is implemented exactly as above). No cap, pipes are lowkey queues in a trench coat. 💅

There are two modes of dummynet operation: “normal” and “fast”. The “normal” mode tries to emulate a real link: the dummynet scheduler ensures that the packet will not leave the pipe faster than it would on the real link with a given bandwidth. The “fast” mode allows certain packets to bypass the dummynet scheduler (if packet flow does not exceed pipe's bandwidth). This is the reason why the “fast” mode requires less CPU cycles per packet (on average) and packet latency can be significantly lower in comparison to a real link with the same bandwidth. The default mode is “normal”. The “fast” mode can be enabled by setting the net.inet.ip.dummynet.io_fast sysctl(8) variable to a non-zero value. ⚡

🧚 PIPE, QUEUE AND SCHEDULER CONFIGURATION 🧚

The pipe, queue and scheduler configuration commands are the following:

pipe number config pipe-configuration

queue number config queue-configuration

sched number config sched-configuration

The following parameters can be configured for a pipe:

  • bw bandwidth | device — Bandwidth, measured in [K|M|G]{bit/s|Byte/s}.

A value of 0 (default) means unlimited bandwidth. The unit must immediately follow the number, as in:

dnctl pipe 1 config bw 300Kbit/s

If a device name is specified instead of a numeric value, as in:

dnctl pipe 1 config bw tun0

then the transmit clock is supplied by the specified device. At the moment only the tun(4) device supports this functionality, for use in conjunction with ppp(8). 🫶

  • delay ms-delay — Propagation delay, measured in milliseconds. The value is rounded to the next multiple of the clock tick (typically 10ms, but it is a good practice to run kernels with options HZ=1000 to reduce the granularity to 1ms or less). The default value is 0, meaning no delay. ⏳
  • burst size — If the data to be sent exceeds the pipe's bandwidth limit (and the pipe was previously idle), up to size bytes of data are allowed to bypass the dummynet scheduler, and will be sent as fast as the physical link allows. Any additional data will be transmitted at the rate specified by the pipe bandwidth. The burst size depends on how long the pipe has been idle; the effective burst size is calculated as follows: MAX( size , bw * pipe_idle_time). 💥
  • profile filename — A file specifying the additional overhead incurred in the transmission of a packet on the link.

Some link types introduce extra delays in the transmission of a packet, e.g., because of MAC level framing, contention on the use of the channel, MAC level retransmissions and so on. From our point of view, the channel is effectively unavailable for this extra time, which is constant or variable depending on the link type. Additionally, packets may be dropped after this time (e.g., on a wireless link after too many retransmissions). We can model the additional delay with an empirical curve that represents its distribution. 🌸

okay bestie, the empirical curve is giving data-visualization main character energy:

                           cumulative probability
                           1.0 ^
                               |
                           L   +-- loss-level          x
                               |                 ******
                               |                *
                               |           *****
                               |          *
                               |        **
                               |       *
                               +-------*------------------->
                                           delay

Vertical lines are constant delay for a range of probabilities. Horizontal lines are a discontinuity in the delay distribution: the pipe will use the largest delay for a given probability. No cap.

The profile file format is literally this, with whitespace acting as a separator and # marking the start of a comment:

  • name identifier — optional name (listed by dnctl pipe show) to identify the delay distribution.
  • bw value — the bandwidth used for the pipe. If not specified here, it must be present explicitly as a configuration parameter for the pipe.
  • loss-level L — the probability above which packets are lost. (0.0 <= L <= 1.0, default 1.0 i.e., no loss.)
  • samples N — the number of samples used in the internal representation of the curve (2..1024; default 100).
  • delay prob | prob delay — one of these two lines is mandatory and defines the format of the following lines with data points.
  • XXX YYY — 2 or more lines representing points in the curve, with either delay or probability first, according to the chosen format. The unit for delay is milliseconds. Data points do not need to be sorted. Also, the number of actual lines can be different from the value of the samples parameter: ipfw utility will sort and interpolate the curve as needed.

Example of a profile file, and yes it’s cute:

name    bla_bla_bla
samples 100
loss-level    0.86
prob    delay
0       200     # minimum overhead is 200ms
0.5     200
0.5     300
0.8     1000
0.9     1300
1       1300
#configuration file end

The following parameters can be configured for a queue:

  • pipe pipe_nr — connects a queue to the specified pipe. Multiple queues (with the same or different weights) can be connected to the same pipe, which specifies the aggregate rate for the set of queues.
  • weight weight — specifies the weight to be used for flows matching this queue. The weight must be in the range 1..100, and defaults to 1.

The following case-insensitive parameters can be configured for a scheduler:

  • type {fifo | wf2q+ | rr | qfq | fq_codel | fq_pie} — specifies the scheduling algorithm to use.
    • fifo — is just a FIFO scheduler (which means that all packets are stored in the same queue as they arrive to the scheduler). FIFO has O(1) per-packet time complexity, with very low constants (estimate 60-80ns on a 2GHz desktop machine) but gives no service guarantees.
    • wf2q+ — implements the WF2Q+ algorithm, which is a Weighted Fair Queueing algorithm which permits flows to share bandwidth according to their weights. Note that weights are not priorities; even a flow with a minuscule weight will never starve. WF2Q+ has O(log N) per-packet processing cost, where N is the number of flows, and is the default algorithm used by previous versions dummynet's queues.
    • rr — implements the Deficit Round Robin algorithm, which has O(1) processing costs (roughly, 100-150ns per packet) and permits bandwidth allocation according to weights, but with poor service guarantees.
    • qfq — implements the QFQ algorithm, which is a very fast variant of WF2Q+, with similar service guarantees and O(1) processing costs (roughly, 200-250ns per packet).
    • fq_codel — implements the FQ-CoDel (FlowQueue-CoDel) scheduler/AQM algorithm, which uses a modified Deficit Round Robin scheduler to manage two lists of sub-queues (old sub-queues and new sub-queues) for providing brief periods of priority to lightweight or short burst flows. By default, the total number of sub-queues is 1024. FQ-CoDel's internal, dynamically created sub-queues are controlled by separate instances of CoDel AQM.
    • fq_pie — implements the FQ-PIE (FlowQueue-PIE) scheduler/AQM algorithm, which similar to fq_codel but uses per sub-queue PIE AQM instance to control the queue delay.

fq_codel inherits AQM parameters and options from codel (see below), and fq_pie inherits AQM parameters and options from pie (see below). Additionally, both of fq_codel and fq_pie have shared scheduler parameters which are:

  • quantum m — m specifies the quantum (credit) of the scheduler. m is the number of bytes a queue can serve before being moved to the tail of old queues list. The default is 1514 bytes, and the maximum acceptable value is 9000 bytes.
  • limit m — m specifies the hard size limit (in unit of packets) of all queues managed by an instance of the scheduler. The default value of m is 10240 packets, and the maximum acceptable value is 20480 packets.
  • flows m — m specifies the total number of flow queues (sub-queues) that fq_* creates and manages. By default, 1024 sub-queues are created when an instance of the fq_{codel/pie} scheduler is created. The maximum acceptable value is 65536.
Note that any token after fq_codel or fq_pie is considered a parameter for fq_{codel/pie}. So, ensure all scheduler configuration options not related to fq_{codel/pie} are written before fq_codel/fq_pie tokens.

In addition to the type, all parameters allowed for a pipe can also be specified for a scheduler. Periodt.

Finally, the following parameters can be configured for both pipes and queues:

  • buckets hash-table-size — specifies the size of the hash table used for storing the various queues. Default value is 64 controlled by the sysctl(8) variable net.inet.ip.dummynet.hash_size, allowed range is 16 to 65536.
  • mask mask-specifier — packets sent to a given pipe or queue by an ipfw rule can be further classified into multiple flows, each of which is then sent to a different dynamic pipe or queue. A flow identifier is constructed by masking the IP addresses, ports and protocol types as specified with the mask options in the configuration of the pipe or queue. For each different flow identifier, a new pipe or queue is created with the same parameters as the original object, and matching packets are sent to it.

Thus, when dynamic pipes are used, each flow will get the same bandwidth as defined by the pipe, whereas when dynamic queues are used, each flow will share the parent's pipe bandwidth evenly with other flows generated by the same queue (note that other queues with different weights might be connected to the same pipe). Available mask specifiers are a combination of one or more of the following: dst-ip mask, dst-ip6 mask, src-ip mask, src-ip6 mask, dst-port mask, src-port mask, flow-id mask, proto mask or all, where the latter means all bits in all fields are significant.

  • noerror — when a packet is dropped by a dummynet queue or pipe, the error is normally reported to the caller routine in the kernel, in the same way as it happens when a device queue fills up. Setting this option reports the packet as successfully delivered, which can be needed for some experimental setups where you want to simulate loss or congestion at a remote router.
  • plr packet-loss-rate / plr K,p,H,r — packet loss rate. Argument packet-loss-rate is a floating-point number between 0 and 1, with 0 meaning no loss, 1 meaning 100% loss.

When invoked with four arguments, the simple Gilbert-Elliott channel model with two states (Good and Bad) is used.

                                           r
                                  .----------------.
                                  v                |
                            .------------.   .------------.
                            |     G      |   |     B      |
                            |  drop (K)  |   |  drop (H)  |
                            '------------'   '------------'
                                  |                ^
                                  '----------------'
                                           p

This has the associated probabilities (K and H) for the loss probability. This is different from the literature, where this model is described with probabilities of successful transmission k and h. However, converting from literature is easy:

K = 1 - k ; H = 1 - h

This is to retain consistency within the interface and allow the quick re-use of loss probability when giving only a single argument. In addition the state change probabilities (p and r) are given. All of the above probabilities are internally represented on 31 bits.

  • queue {slots | sizeKbytes} — how big is the queue, bestie? You can say it in slots or in KBytes. Default is 50 slots, which is literally the typical queue size for Ethernet devices 🫶. But heads up: on slow links you gotta keep the queue short or your traffic gets wrecked by a big ol' queueing delay 😭. Like, 50 max-sized Ethernet packets (1500 bytes) means 600Kbit or 20s of queue on a 30Kbit/s pipe 💀. And it gets even worse if packets come in from an interface with a much larger MTU — e.g. the loopback interface with its 16KB packets 🙈. The sysctl(8) variables net.inet.ip.dummynet.pipe_byte_limit and net.inet.ip.dummynet.pipe_slot_limit control the maximum lengths you're allowed to specify, so that's the ceiling, periodt.
  • red | gred w_q/min_th/max_th/max_p [ecn] — okay so this turns on the RED (Random Early Detection) queue management algorithm ✨. w_q and max_p are floating point numbers between 0 and 1 (inclusive), while min_th and max_th are integer numbers that specify thresholds for queue management — and those thresholds get computed in bytes if the queue was defined in bytes, in slots otherwise 🎀. The two parameters can also be the same value if you need that. dummynet also supports the gentle RED variant (gred) and ECN (Explicit Congestion Notification) as optional 💅. Three sysctl(8) variables control the RED behaviour:
    • net.inet.ip.dummynet.red_lookup_depth — specifies the accuracy in computing the average queue when the link is idle (defaults to 256, must be greater than zero) ⭐
    • net.inet.ip.dummynet.red_avg_pkt_size — specifies the expected average packet size (defaults to 512, must be greater than zero) 💖
    • net.inet.ip.dummynet.red_max_pkt_size — specifies the expected maximum packet size, only used when queue thresholds are in bytes (defaults to 1500, must be greater than zero) 🦋
  • codel [target time] [interval time] [ecn | noecn] — make use of the CoDel (Controlled-Delay) queue management algorithm 👑. time is interpreted as milliseconds by default, but you can specify seconds (s), milliseconds (ms) or microseconds (us) instead. CoDel drops or marks (ECN) packets depending on packet sojourn time in the queue. target time (5ms by default) is the minimum acceptable persistent queue delay that CoDel allows 🌸. CoDel does not drop packets directly after the sojourn time goes higher than target time — it waits for interval time (100ms default) before dropping 🎀. interval time should be set to the maximum RTT for all expected connections. ecn enables (disabled by default) packet marking (instead of dropping) for ECN-enabled TCP flows when queue delay becomes high ✨.
Ngl this one's important: any token after codel is considered a parameter for CoDel 😭. So make sure all your pipe/queue configuration options are written before the codel token, okay?

The sysctl(8) variables net.inet.ip.dummynet.codel.target and net.inet.ip.dummynet.codel.interval can be used to set CoDel default parameters 💜.

  • pie [target time] [tupdate time] [alpha n] [beta n] [max_burst time] [max_ecnth n] [ecn | noecn] [capdrop | nocapdrop] [drand | nodrand] [onoff] [dre | ts] — make use of the PIE (Proportional Integral controller Enhanced) queue management algorithm 🌈. PIE drops or marks packets depending on a calculated drop probability during the en-queue process, with the aim of achieving high throughput while keeping queue delay low 🔥. At regular time intervals of tupdate time (15ms by default) a background process (re)calculates the probability based on queue delay deviations from target time (15ms by default) and queue delay trends. PIE approximates current queue delay by using a departure rate estimation method, or (optionally) by using a packet timestamp method similar to CoDel ⭐. time is interpreted as milliseconds by default but seconds (s), milliseconds (ms) or microseconds (us) can be specified instead. The other PIE parameters and options are as follows:
    • alpha n — n is a floating point number between 0 and 7 which specifies the weight of queue delay deviations that is used in drop probability calculation. 0.125 is the default 💅
    • beta n — n is a floating point number between 0 and 7 which specifies is the weight of queue delay trend that is used in drop probability calculation. 1.25 is the default 🎀
    • max_burst time — the maximum period of time that PIE does not drop/mark packets. 150ms is the default and 10s is the maximum value 🦋
    • max_ecnth n — even when ECN is enabled, PIE drops packets instead of marking them when drop probability becomes higher than ECN probability threshold max_ecnth n, the default is 0.1 (i.e 10%) and 1 is the maximum value ✨
    • ecn | noecn — enable or disable ECN marking for ECN-enabled TCP flows. Disabled by default 💖
    • capdrop | nocapdrop — enable or disable cap drop adjustment. Cap drop adjustment is enabled by default 🌸
    • drand | nodrand — enable or disable drop probability de-randomisation. De-randomisation eliminates the problem of dropping packets too close or too far. De-randomisation is enabled by default 🧚
    • onoff — enable turning PIE on and off depending on queue load. If this option is enabled, PIE turns on when over 1/3 of queue becomes full. This option is disabled by default 👑
    • dre | ts — calculate queue delay using departure rate estimation dre or timestamps ts. dre is used by default 💜
Same vibe check as codel, bestie: any token after pie is considered a parameter for PIE. So ensure all pipe/queue configuration options are written before the pie token 😭. sysctl(8) variables can be used to control the pie default parameters — see the “SYSCTL VARIABLES” section for more details 📖.

Okay, and now the sad part 😭 — when used with IPv6 data, dummynet currently has several limitations. Information necessary to route link-local packets to an interface is not available after processing by dummynet, so those packets are dropped in the output path 💀. So be careful and make sure link-local packets are not passed to dummynet, fr fr.

🎀 CHECKLIST 🎀

Here are some important points to consider when designing your rules — think of it as your pre-flight vibe check ✨:

  • Remember that you filter both packets going in and out. Most connections need packets going in both directions 🫶
  • Remember to test very carefully. It is a good idea to be near the console when doing this 💻. If you cannot be near the console, use an auto-recovery script such as the one in /usr/share/examples/ipfw/change_rules.sh 🙈
  • Do not forget the loopback interface 👑 (she matters!)

🦋 FINE POINTS 🦋

  • There are circumstances where fragmented datagrams are unconditionally dropped 💀. TCP packets are dropped if they do not contain at least 20 bytes of TCP header, UDP packets are dropped if they do not contain a full 8 byte UDP header, and ICMP packets are dropped if they do not contain 4 bytes of ICMP header — enough to specify the ICMP type, code, and checksum 🎀. These packets are simply logged as “pullup failed” since there may not be enough good data in the packet to produce a meaningful log entry 😭
  • Another type of packet is unconditionally dropped: a TCP packet with a fragment offset of one 🙅‍♀️. This is a valid packet, but it only has one use — to try to circumvent firewalls 💀. When logging is enabled, these packets are reported as being dropped by rule -1
  • If you are logged in over a network, loading the kld(4) version of ipfw is probably not as straightforward as you would think 😅. The following command line is recommended:
kldload ipfw && \
ipfw add 32000 allow ip from any to any

Along the same lines, doing an

ipfw flush

in similar surroundings is also a bad idea 🚨 (like, genuinely do not)

  • The ipfw filter list may not be modified if the system security level is set to 3 or higher (see init(8) for information on system security levels) 🔒

✨ PACKET DIVERSION ✨

A divert(4) socket bound to the specified port will receive all packets diverted to that port 📬. But if no socket is bound to the destination port, or if the divert module is not loaded, or if the kernel was not compiled with divert socket support — the packets are dropped 💀. That's it, that's the whole tea 🍓.

💅 NETWORK ADDRESS TRANSLATION (NAT) 💅

Okay bestie, so ipfw does in-kernel NAT using the kernel version of libalias(3). The way this works: the kernel module ipfw_nat needs to be loaded, OR your kernel needs options IPFIREWALL_NAT — that's the whole vibe check for being able to use NAT at all. 💜

The nat configuration command goes like this:

nat nat_number config nat-configuration

And here's everything you can configure, no cap:

  • ip ip_address — set an IP address to use for aliasing. That's it, that's the whole job.
  • if nic — use the IP address of nic for aliasing, and dynamically changing it if nic's IP address changes. Lowkey iconic, it keeps up with you. ✨
  • log — enable logging on this nat instance.
  • deny_in — deny any incoming connection from the outside world. Gatekeeping, but make it firewall. 🚪
  • same_ports — try to leave the alias port numbers unchanged from the actual local port numbers.
  • unreg_only — traffic on the local network not originating from a RFC 1918 unregistered address space gets ignored. Not from the registered range? Not invited. 🙈
  • unreg_cgn — like unreg_only, but includes the RFC 6598 (Carrier Grade NAT) address range too.
  • reset — reset the table of the packet aliasing engine whenever the address changes.
  • reverse — reverse the way libalias handles aliasing. Flip it, reverse it, slay.
  • proxy_only — obey transparent proxy rules only; packet aliasing is not performed. Read carefully bestie, this one changes the whole game.
  • skip_global — skip this instance in case of a global state lookup (see below).
  • port_range lower-upper — set the aliasing ports between the ranges given. Upper port has to be greater than lower. Not a suggestion, fr fr.
  • udp_eim — when enabled, UDP packets use endpoint-independent mapping (EIM) from RFC 4787 (that's the "full cone" NAT of RFC 3489). All packets from the same internal address:port get mapped to the same NAT address:port, regardless of their destination address:port. If filtering rules allow, and if deny_in is unset, any other external address:port can also send to the internal address:port through its mapped NAT address:port. This is more compatible with applications, and can reduce the need for port forwarding, but less scalable as each NAT address:port can only be concurrently used by at most one internal address:port.

When disabled, UDP packets use endpoint-dependent mapping (EDM) — the "symmetric" NAT. Each connection from a particular internal address:port to different external addresses:ports is mapped to a random and unpredictable NAT address:port. Two appplications behind EDM NATs can only connect to each other by port forwarding on the NAT, or tunnelling through an in-between server. 💀

So, some special values can be supplied instead of nat_number in nat rule actions, and they're honestly kind of main character energy:

  • global — looks up translation state in all configured nat instances. If an entry is found, the packet is aliased according to that entry. If no entry was found in any of the instances, the packet is passed unchanged, and no new entry will be created. See section "MULTIPLE INSTANCES" in natd(8) for more information.
  • tablearg — uses the argument supplied in the lookup table. See the "LOOKUP TABLES" section below for more information on lookup tables. 🫶

To let the packet continue after being (de)aliased, set the sysctl variable net.inet.ip.fw.one_pass to 0. For more information about aliasing modes, refer to libalias(3). And see Section "EXAMPLES" for some examples of nat usage.

⭐ Also, quick heads up: to delete a specific nat configuration instance, use this command: nat nat_number delete

🎀 REDIRECT AND LSNAT SUPPORT IN IPFW 🎀

Redirect and LSNAT support follow closely the syntax used in natd(8). That's the whole section, no notes bestie. 💖 See Section "EXAMPLES" for some examples on how to do redirect and lsnat.

🦋 SCTP NAT SUPPORT 🦋

SCTP nat can be configured in a similar manner to TCP through the ipfw command line tool. The main difference? SCTP nat does not do port translation. Since the local and global side ports will be the same, there's no need to specify both — like, why say it twice, you know? Ports are redirected as follows:

nat nat_number config if nic redirect_port sctp
ip_address [,addr_list] {[port | port-port] [,ports]}

Most SCTP nat configuration can be done in real-time through the sysctl(8) interface. All may be changed dynamically, though the hash_table size will only change for new nat instances. That part's important — existing instances stay on their old size. See "SYSCTL VARIABLES" for more info. 🌸

✨ IPv6/IPv4 NETWORK ADDRESS AND PROTOCOL TRANSLATION ✨

💖 Stateful translation 💖

Okay bestie, so ipfw lowkey supports in-kernel IPv6/IPv4 network address and protocol translation. Stateful NAT64 translation lets IPv6-only clients reach IPv4 servers using unicast TCP, UDP, or ICMP protocols. One or more IPv4 addresses assigned to a stateful NAT64 translator get shared among several IPv6-only clients. When stateful NAT64 is used with DNS64, usually no changes are needed in the IPv6 client or the IPv4 server. The kernel module ipfw_nat64 should be loaded, or the kernel should have options IPFIREWALL_NAT64, so you can use the stateful NAT64 translator. 💅

Stateful NAT64 uses a bunch of memory for several types of objects. When an IPv6 client initiates a connection, the NAT64 translator creates a host entry in the states table. Each host entry uses a preallocated IPv4 alias entry. Each alias entry has a number of ports group entries allocated on demand. Ports group entries contain connection state entries. There are several options to control limits and lifetime for these objects. 🫶

NAT64 translator follows RFC7915 when it does ICMPv6/ICMP translation, and unsupported message types will be silently dropped. IPv6 needs several ICMPv6 message types to be explicitly allowed for correct operation.

🚨 Make sure that ND6 neighbor solicitation (ICMPv6 type 135) and neighbor advertisement (ICMPv6 type 136) messages will not be handled by translation rules.

After translation, NAT64 translator by default sends packets through the corresponding netisr queue. That means the translator host should be configured as an IPv4 and IPv6 router. It also means a packet is handled by the firewall twice: first the original packet is handled and consumed by the translator, and then it is handled again as a translated packet. This behavior can be changed by the sysctl variable net.inet.ip.fw.nat64_direct_output. Also, the translated packet can be tagged using the tag rule action, and then matched by the tagged opcode to avoid loops and extra overhead. ✨

The stateful NAT64 configuration command is this:

nat64lsn name create create-options

The following parameters can be configured:

  • prefix4 ipv4_prefix/plen — the IPv4 prefix with mask defines the pool of IPv4 addresses used as source address after translation. Stateful NAT64 module translates the IPv6 source address of the client to one IPv4 address from this pool. Note that incoming IPv4 packets that don't have a corresponding state entry in the states table will be dropped by the translator. Make sure that translation rules handle packets destined to the configured prefix.
  • prefix6 ipv6_prefix/length — the IPv6 prefix defines IPv4-embedded IPv6 addresses used by the translator to represent IPv4 addresses. This IPv6 prefix should be configured in DNS64. The translator implementation follows RFC6052, which restricts the length of prefixes to one of following: 32, 40, 48, 56, 64, or 96. The Well-Known IPv6 Prefix 64:ff9b:: must be 96 bits long. The special ::/length prefix can be used to handle several IPv6 prefixes with one NAT64 instance. The NAT64 instance will determine a destination IPv4 address from the prefix length.
  • states_chunks number — the number of states chunks in a single ports group. Each ports group by default can keep 64 state entries in a single chunk. The above value affects the maximum number of states that can be associated with a single IPv4 alias address and port. The value must be a power of 2, and up to 128.
  • host_del_age seconds — the number of seconds until the host entry for an IPv6 client will be deleted and all its resources will be released due to inactivity. Default value is 3600.
  • pg_del_age seconds — the number of seconds until a ports group with unused state entries will be released. Default value is 900.
  • tcp_syn_age seconds — the number of seconds while a state entry for a TCP connection with only SYN sent will be kept. If TCP connection establishing is not finished, the state entry will be deleted. Default value is 10.
  • tcp_est_age seconds — the number of seconds while a state entry for an established TCP connection will be kept. Default value is 7200.
  • tcp_close_age seconds — the number of seconds while a state entry for a closed TCP connection will be kept. Keeping state entries for closed connections is needed because IPv4 servers typically keep closed connections in a TIME_WAIT state for several minutes. Since the translator's IPv4 addresses are shared among all IPv6 clients, new connections from the same addresses and ports may be rejected by the server, because these connections are still in a TIME_WAIT state. Keeping them in the translator's state table protects from such rejects. Default value is 180.
  • udp_age seconds — the number of seconds while the translator keeps a state entry waiting for a reply to the sent UDP datagram. Default value is 120.
  • icmp_age seconds — the number of seconds while the translator keeps a state entry waiting for a reply to the sent ICMP message. Default value is 60.
  • log — turn on logging of all handled packets via BPF through the ipfwlog0 interface. ipfwlog0 is a pseudo interface and can be created after boot manually with the ifconfig command. Note that it has a different purpose than the ipfw0 interface. The translator sends additional information to BPF with each packet. With tcpdump you are able to see each handled packet before and after translation. 👀
  • -log — turn off logging of all handled packets via BPF.
  • allow_private — turn on processing private IPv4 addresses. By default, IPv6 packets with destinations mapped to private address ranges defined by RFC1918 are not processed.
  • -allow_private — turn off private address handling in the nat64 instance.

To inspect a states table of stateful NAT64, this command can be used:

nat64lsn name show states

Stateless NAT64 translator doesn't use a states table for translation and converts IPv4 addresses to IPv6 and vice versa solely based on the mappings taken from configured lookup tables. Since a states table isn't used by the stateless translator, it can be configured to pass IPv4 clients to IPv6-only servers. 🦋

The stateless NAT64 configuration command is this:

nat64stl name create create-options

The following parameters can be configured:

  • prefix6 ipv6_prefix/length — the IPv6 prefix defines IPv4-embedded IPv6 addresses used by the translator to represent IPv4 addresses. This IPv6 prefix should be configured in DNS64.
  • table4 table46 — the lookup table table46 contains mapping for how IPv4 addresses should be translated to IPv6 addresses.
  • table6 table64 — the lookup table table64 contains mapping for how IPv6 addresses should be translated to IPv4 addresses.
  • log — turn on logging of all handled packets via BPF through the ipfwlog0 interface.
  • -log — turn off logging of all handled packets via BPF.
  • allow_private — turn on processing private IPv4 addresses. By default, IPv6 packets with destinations mapped to private address ranges defined by RFC1918 are not processed.
  • -allow_private — turn off private address handling in the nat64 instance.
🌸 Note that the behavior of the stateless translator with respect to not-matched packets differs from the stateful translator. If the corresponding addresses were not found in the lookup tables, the packet will not be dropped and the search continues.

🦋 XLAT464 CLAT translation 🦋

XLAT464 CLAT NAT64 translator implements client-side stateless translation as defined in RFC6877 and is very similar to the stateless NAT64 translator explained above. Instead of lookup tables, it uses a one-to-one mapping between IPv4 and IPv6 addresses using configured prefixes. This mode can be used as a replacement of DNS64 service for applications that are not using it (e.g. VoIP), allowing them to access IPv4-only Internet over IPv6-only networks with help of a remote NAT64 translator. 💜

The CLAT NAT64 configuration command is this:

nat64clat name create create-options

The following parameters can be configured:

  • clat_prefix ipv6_prefix/length — the IPv6 prefix defines IPv4-embedded IPv6 addresses used by the translator to represent source IPv4 addresses.
  • plat_prefix ipv6_prefix/length — the IPv6 prefix defines IPv4-embedded IPv6 addresses used by the translator to represent destination IPv4 addresses. This IPv6 prefix should be configured on a remote NAT64 translator.
  • log — turn on logging of all handled packets via BPF through the ipfwlog0 interface.
  • -log — turn off logging of all handled packets via BPF.
  • allow_private — turn on processing private IPv4 addresses. By default, a nat64clat instance will not process IPv4 packets with a destination address from private ranges as defined in RFC1918.
  • -allow_private — turn off private address handling in the nat64clat instance.
⭐ Note that the behavior of the CLAT translator with respect to not-matched packets differs from the stateful translator. If the corresponding addresses were not matched against the configured prefixes, the packet will not be dropped and the search continues.

💖 IPv6-to-IPv6 NETWORK PREFIX TRANSLATION (NPTv6) 💖

Okay so ipfw supports in-kernel IPv6-to-IPv6 network prefix translation, exactly as described in RFC6296. The kernel module ipfw_nptv6 should be loaded, or your kernel should have options IPFIREWALL_NPTV6, to be able to use the NPTv6 translator.

The NPTv6 configuration command is literally this:

nptv6 name create create-options

The following parameters can be configured:

  • int_prefix ipv6_prefix — the IPv6 prefix used in the internal network. The NPTv6 module translates the source address when it matches this prefix.
  • ext_prefix ipv6_prefix — the IPv6 prefix used in the external network. The NPTv6 module translates the destination address when it matches this prefix.
  • ext_if nic — the NPTv6 module will use the first global IPv6 address from interface nic as the external prefix. This can be useful when the IPv6 prefix of the external network is dynamically obtained. The ext_prefix and ext_if options are mutually exclusive.
  • prefixlen length — the length of the specified IPv6 prefixes. It must be in the range from 8 to 64.
⚠️ Bestie, note that the prefix translation rules are silently ignored when IPv6 packet forwarding is disabled. To enable packet forwarding, set the sysctl variable net.inet6.ip6.forwarding to 1.

To let the packet continue after being translated, set the sysctl variable net.inet.ip.fw.one_pass to 0.

🌸 LOADER TUNABLES 🌸

Tunables can be set in the loader(8) prompt, loader.conf(5) or kenv(1) before the ipfw module gets loaded.

  • net.inet.ip.fw.enable: 1 — enables the firewall. Setting this variable to 0 lets you run your machine without a firewall even if it was compiled in.
  • net.inet6.ip6.fw.enable: 1 — provides the same functionality as above for the IPv6 case.
  • net.link.ether.ipfw: 0 — controls whether layer2 packets are passed to ipfw. Default is no.
  • net.inet.ip.fw.default_to_accept: 0 — defines ipfw last rule behavior. This value overrides options IPFW_DEFAULT_TO_(ACCEPT|DENY) from the kernel configuration file.
  • net.inet.ip.fw.tables_max: 128 — defines the number of tables available in ipfw. The number cannot exceed 65534.

✨ SYSCTL VARIABLES ✨

A whole set of sysctl(8) variables controls the behaviour of the firewall and associated modules (dummynet, bridge, sctp nat). These are shown below together with their default value (but always check with the sysctl(8) command what value is actually in use) and meaning:

  • net.inet.ip.alias.sctp.accept_global_ootb_addip: 0 — defines how the nat responds to receipt of global OOTB ASCONF-AddIP:
    • 0 — No response (unless a partially matching association exists - ports and vtags match but global address does not)
    • 1 — nat will accept and process all OOTB global AddIP messages.
Option 1 should never be selected as this forms a security risk. An attacker can establish multiple fake associations by sending AddIP messages.
  • net.inet.ip.alias.sctp.chunk_proc_limit: 5 — defines the maximum number of chunks in an SCTP packet that will be parsed for a packet that matches an existing association. This value is enforced to be greater or equal than net.inet.ip.alias.sctp.initialising_chunk_proc_limit. A high value is a DoS risk, yet setting too low a value may result in important control chunks in the packet not being located and parsed.
  • net.inet.ip.alias.sctp.error_on_ootb: 1 — defines when the nat responds to any Out-of-the-Blue (OOTB) packets with ErrorM packets. An OOTB packet is a packet that arrives with no existing association registered in the nat and is not an INIT or ASCONF-AddIP packet:
    • 0 — ErrorM is never sent in response to OOTB packets.
    • 1 — ErrorM is only sent to OOTB packets received on the local side.
    • 2 — ErrorM is sent to the local side and on the global side ONLY if there is a partial match (ports and vtags match but the source global IP does not). This value is only useful if the nat is tracking global IP addresses.
    • 3 — ErrorM is sent in response to all OOTB packets on both the local and global side (DoS risk).
Heads-up: the line above says the default is 1, but the paragraph says “At the moment the default is 0” — both are literally what this page says, so keep both in mind, bestie 💀. At the moment the default is 0, since the ErrorM packet is not yet supported by most SCTP stacks. When it is supported, and if not tracking global addresses, we recommend setting this value to 1 to allow multi-homed local hosts to function with the nat. To track global addresses, we recommend setting this value to 2 to allow global hosts to be informed when they need to (re)send an ASCONF-AddIP. Value 3 should never be chosen (except for debugging) as the nat will respond to all OOTB global packets (a DoS risk).
  • net.inet.ip.alias.sctp.hashtable_size: 2003 — size of hash tables used for nat lookups (100 < prime_number > 1000001). This value sets the hash table size for any future created nat instance and therefore must be set prior to creating a nat instance. The table sizes may be changed to suit specific needs. If there will be few concurrent associations, and memory is scarce, you may make these smaller. If there will be many thousands (or millions) of concurrent associations, you should make these larger. A prime number is best for the table size. The sysctl update function will adjust your input value to the next highest prime number.
  • net.inet.ip.alias.sctp.holddown_time: 0 — hold association in table for this many seconds after receiving a SHUTDOWN-COMPLETE. This allows endpoints to correct shutdown gracefully if a shutdown_complete is lost and retransmissions are required.
  • net.inet.ip.alias.sctp.init_timer: 15 — timeout value while waiting for (INIT-ACK|AddIP-ACK). This value cannot be 0.
  • net.inet.ip.alias.sctp.initialising_chunk_proc_limit: 2 — defines the maximum number of chunks in an SCTP packet that will be parsed when no existing association exists that matches that packet. Ideally this packet will only be an INIT or ASCONF-AddIP packet. A higher value may become a DoS risk as malformed packets can consume processing resources.
  • net.inet.ip.alias.sctp.param_proc_limit: 25 — defines the maximum number of parameters within a chunk that will be parsed in a packet. As for other similar sysctl variables, larger values pose a DoS risk.
  • net.inet.ip.alias.sctp.log_level: 0 — level of detail in the system log messages (0 - minimal, 1 - event, 2 - info, 3 - detail, 4 - debug, 5 - max debug). May be a good option in high loss environments.
  • net.inet.ip.alias.sctp.shutdown_time: 15 — timeout value while waiting for SHUTDOWN-COMPLETE. This value cannot be 0.
  • net.inet.ip.alias.sctp.track_global_addresses: 0 — enables/disables global IP address tracking within the nat and places an upper limit on the number of addresses tracked for each association:
    • 0 — Global tracking is disabled
    • >1 — Enables tracking, the maximum number of addresses tracked for each association is limited to this value
This variable is fully dynamic, the new value will be adopted for all newly arriving associations, existing associations are treated as they were previously. Global tracking will decrease the number of collisions within the nat at a cost of increased processing load, memory usage, complexity, and possible nat state problems in complex networks with multiple nats. We recommend not tracking global IP addresses, this will still result in a fully functional nat.
  • net.inet.ip.alias.sctp.up_timer: 300 — timeout value to keep an association up with no traffic. This value cannot be 0.
  • net.inet.ip.dummynet.codel.interval: 100000 — default codel AQM interval in microseconds. The value must be in the range 1..5000000.
  • net.inet.ip.dummynet.codel.target: 5000 — default codel AQM target delay time in microseconds (the minimum acceptable persistent queue delay). The value must be in the range 1..5000000.
  • net.inet.ip.dummynet.expire: 1 — lazily delete dynamic pipes/queue once they have no pending traffic. You can disable this by setting the variable to 0, in which case the pipes/queues will only be deleted when the threshold is reached.
  • net.inet.ip.dummynet.fqcodel.flows: 1024 — defines the default total number of flow queues (sub-queues) that fq_codel creates and manages. The value must be in the range 1..65536.
  • net.inet.ip.dummynet.fqcodel.interval: 100000 — default fq_codel scheduler/AQM interval in microseconds. The value must be in the range 1..5000000.
  • net.inet.ip.dummynet.fqcodel.limit: 10240 — the default hard size limit (in unit of packet) of all queues managed by an instance of the fq_codel scheduler. The value must be in the range 1..20480.
  • net.inet.ip.dummynet.fqcodel.quantum: 1514 — the default quantum (credit) of the fq_codel in unit of byte. The value must be in the range 1..9000.
  • net.inet.ip.dummynet.fqcodel.target: 5000 — default fq_codel scheduler/AQM target delay time in microseconds (the minimum acceptable persistent queue delay). The value must be in the range 1..5000000.
  • net.inet.ip.dummynet.fqpie.alpha: 125 — the default alpha parameter (scaled by 1000) for fq_pie scheduler/AQM. The value must be in the range 1..7000.
  • net.inet.ip.dummynet.fqpie.beta: 1250 — the default beta parameter (scaled by 1000) for fq_pie scheduler/AQM. The value must be in the range 1..7000.
  • net.inet.ip.dummynet.fqpie.flows: 1024 — defines the default total number of flow queues (sub-queues) that fq_pie creates and manages. The value must be in the range 1..65536.
  • net.inet.ip.dummynet.fqpie.limit: 10240 — the default hard size limit (in unit of packet) of all queues managed by an instance of the fq_pie scheduler. The value must be in the range 1..20480.
  • net.inet.ip.dummynet.fqpie.max_burst (default 150000) — the default max stretch of microseconds where the fq_pie scheduler/AQM just… doesn't drop or mark anything. Has to be in the range 1..10000000.
  • net.inet.ip.dummynet.fqpie.max_ecnth (default 99) — default maximum ECN probability threshold (scaled by 1000) for the fq_pie scheduler/AQM. Range check: 1..7000, or it's a no.
  • net.inet.ip.dummynet.fqpie.quantum (default 1514) — default quantum (aka credit) of fq_pie, measured in bytes. Must land in 1..9000.
  • net.inet.ip.dummynet.fqpie.target (default 15000) — default target delay for fq_pie, in microseconds. Range: 1..5000000.
  • net.inet.ip.dummynet.fqpie.tupdate (default 15000) — default tupdate for fq_pie, also in microseconds. Also 1..5000000.
  • net.inet.ip.dummynet.hash_size (default 64) — default size of the hash table used for dynamic pipes/queues. This is the value it reaches for when you configure a pipe/queue and didn't pass the buckets option. iykyk.
  • net.inet.ip.dummynet.io_fast (default 0) — set it to anything non-zero and the "fast" mode of dummynet operation (as described above) gets switched on. ✨
  • net.inet.ip.dummynet.io_pkt — count of packets handed over to dummynet.
  • net.inet.ip.dummynet.io_pkt_drop — count of packets dummynet yeeted into the void.
  • net.inet.ip.dummynet.io_pkt_fast — count of packets the dummynet scheduler straight up bypassed.
  • net.inet.ip.dummynet.max_chain_len (default 16) — the target value for the biggest number of pipes/queues you can cram into one hash bucket. The product max_chain_len*hash_size is what determines the threshold above which empty pipes/queues get expired — even when net.inet.ip.dummynet.expire=0. 💀
  • net.inet.ip.dummynet.red_lookup_depth (default 256)
  • net.inet.ip.dummynet.red_avg_pkt_size (default 512)
  • net.inet.ip.dummynet.red_max_pkt_size (default 1500) — these three are the parameters used when computing the drop probability for the RED algorithm. bestie math, but make it network 🔥
  • net.inet.ip.dummynet.pie.alpha (default 125) — the default alpha parameter (scaled by 1000) for pie AQM. Value has to be in 1..7000.
  • net.inet.ip.dummynet.pie.beta (default 1250) — the default beta parameter (scaled by 1000) for pie AQM. Also 1..7000.
  • net.inet.ip.dummynet.pie.max_burst (default 150000) — default max period of microseconds where pie AQM refuses to drop/mark packets. Range 1..10000000.
  • net.inet.ip.dummynet.pie.max_ecnth (default 99) — default maximum ECN probability threshold (scaled by 1000) for pie AQM. 1..7000 only.
  • net.inet.ip.dummynet.pie.target (default 15000) — default target delay of pie AQM, in microseconds. 1..5000000.
  • net.inet.ip.dummynet.pie.tupdate (default 15000) — default tupdate of pie AQM, in microseconds. 1..5000000.
  • net.inet.ip.dummynet.pipe_byte_limit (default 1048576)
  • net.inet.ip.dummynet.pipe_slot_limit (default 100) — the maximum queue size you're allowed to specify, in bytes or in packets. These limits exist so you don't accidentally go and exhaust resources like mbufs. 👀 If you raise them, bestie, you better make sure the system is configured so the resources are actually there. No cap.
  • net.inet.ip.fw.autoinc_step (default 100) — the delta between rule numbers when they're auto-generated. Must be in 1..1000.
  • net.inet.ip.fw.curr_dyn_buckets (default net.inet.ip.fw.dyn_buckets) — the current number of buckets in the hash table for dynamic rules. Read-only, so look don't touch. 🙈
  • net.inet.ip.fw.debug (default 1) — controls the debugging messages ipfw spits out.
  • net.inet.ip.fw.default_rule (default 65535) — the default rule number, and it's read-only. By ipfw's whole design the default rule is the last one, so its number doubles as the highest number a rule is allowed to have. 👑
  • net.inet.ip.fw.dyn_buckets (default 256) — number of buckets in the hash table for dynamic rules. Has to be a power of 2, up to 65536. It only takes effect once all dynamic rules have expired, so you're advised to use a flush command to make sure the hash table actually gets resized. 💅
  • net.inet.ip.fw.dyn_count (default 3) — current number of dynamic rules. Read-only.
  • net.inet.ip.fw.dyn_keepalive (default 1) — turns on generating keepalive packets for keep-state rules on TCP sessions. A keepalive gets sent to both sides of the connection every 5 seconds, for the last 20 seconds of the rule's lifetime. 🫶
  • net.inet.ip.fw.dyn_max (default 8192) — maximum number of dynamic rules. Hit this ceiling and no more dynamic rules can be installed until some old ones expire. Sorry babe.
  • net.inet.ip.fw.dyn_ack_lifetime (default 300)
  • net.inet.ip.fw.dyn_syn_lifetime (default 20)
  • net.inet.ip.fw.dyn_fin_lifetime (default 1)
  • net.inet.ip.fw.dyn_rst_lifetime (default 1)
  • net.inet.ip.fw.dyn_udp_lifetime (default 5)
  • net.inet.ip.fw.dyn_short_lifetime (default 30) — these six control the lifetime (in seconds) of dynamic rules. On the initial SYN exchange the lifetime is kept short, then it grows after both SYN have been seen, then it shrinks again during the final FIN exchange or when a RST shows up. 🦋 Important: both dyn_fin_lifetime and dyn_rst_lifetime MUST be strictly lower than 5 seconds (that's the keepalive repetition period). The firewall enforces it, so don't even try it.
  • net.inet.ip.fw.dyn_keep_states (default 0) — keeps dynamic states around when a rule/set gets deleted. The states get relinked to the default rule (65535). This is handly (yes, handly) for ruleset reloads. Off by default.
  • net.inet.ip.fw.one_pass (default 1) — when it's set, a packet exiting the dummynet pipe or the ng_ipfw(4) node does not get passed through the firewall again. Otherwise, after an action, the packet gets reinjected into the firewall at the next rule. ⭐
  • net.inet.ip.fw.tables_max (default 128) — maximum number of tables.
  • net.inet.ip.fw.verbose (default 1) — turns on verbose messages.
  • net.inet.ip.fw.verbose_limit (default 0) — caps how many messages a verbose firewall is allowed to produce.
  • net.inet6.ip6.fw.deny_unknown_exthdrs (default 1) — if it's enabled, packets carrying unknown IPv6 Extension Headers get denied. 🚫
  • net.link.bridge.ipfw (default 0) — controls whether bridged packets get passed to ipfw. The default is no.
  • net.inet.ip.fw.nat64_debug (default 0) — controls the debugging messages produced by the ipfw_nat64 module.
  • net.inet.ip.fw.nat64_direct_output (default 0) — controls which output method the ipfw_nat64 module uses:
    • 0 — a packet gets handled by ipfw twice. First the original packet is handled by ipfw and consumed by the ipfw_nat64 translator. Then the translated packet is queued via netisr back into input processing again. 🔁
    • 1 — a packet gets handled by ipfw only once, and after translation it's pushed directly out to the outgoing interface. 💨

🔍 INTERNAL DIAGNOSTICS

There are a handful of commands that might come in useful when you want to understand the current state of certain subsystems inside the kernel module. Heads up though: these commands give you debugging output, and that output may change without notice. So don't build your whole personality on it. 😭

Right now, the following commands are available as internal sub-options:

  • iflist — lists every interface that ipfw is currently tracking, along with each one's in-kernel status. 👑
  • monitor [filter-comment] — captures messages from the route(4) socket that were logged using rules with log logdst rtsock opcode. You can optionally pass a filter-comment so it only shows you the messages logged by rules that have that specific rule comment. It's giving "search your group chat for one specific message" energy.
  • talist — lists all the table lookup algorithms that are currently available. ✨

🎀 EXAMPLES

There are far too many possible uses of ipfw to ever list them all, so this section is just gonna hand you a small set of examples. Snack-sized. 🍓

BASIC PACKET FILTERING

This command adds an entry that denies all tcp packets from cracker.evil.org to the telnet port of wolf.tambov.su, so they never get forwarded by the host:

ipfw add deny tcp from cracker.evil.org to wolf.tambov.su telnet

This one blocks any connection from the entire cracker network to my host:

ipfw add deny ip from 123.45.67.0/24 to my.host.org

A first and honestly pretty efficient way to limit access (without using dynamic rules) is reaching for these rules right here:

ipfw add allow tcp from any to any established
ipfw add allow tcp from net1 portlist1 to net2 portlist2 setup
ipfw add allow tcp from net3 portlist3 to net3 portlist3 setup
...
ipfw add deny tcp from any to any

So the first rule is a quick win for normal TCP packets, but it will NOT match the initial SYN packet — that one gets matched by the setup rules only for the selected source/destination pairs. Every other SYN packet gets rejected by that final deny rule. Ate. 💅

If you admin one or more subnets, you can absolutely take advantage of address sets and or-blocks and write extremely compact rulesets that selectively enable services to whole blocks of clients, like this:

goodguys="{ 10.1.2.0/24{20,35,66,18} or 10.2.3.0/28{6,3,11} }"
badguys="10.1.2.0/24{8,38,60}"

ipfw add allow ip from ${goodguys} to any
ipfw add deny ip from ${badguys} to any
... normal policies ...

Allow any transit packets coming from single vlan 10 and heading out to vlans 100-1000:

ipfw add 10 allow out recv vlan10 \
{ xmit vlan1000 or xmit "vlan[1-9]??" }

The verrevpath option is literally how you do automated anti-spoofing — just drop this at the top of a ruleset:

ipfw add deny ip from any to any not verrevpath in

This rule drops all incoming packets that look like they showed up to the system on the wrong interface. Example: a packet whose source address belongs to a host on a protected internal network would get dropped if it tried to enter the system from an external interface. Bye. 🦋

The antispoof option does something similar but more restricted — stick this at the top of a ruleset:

ipfw add deny ip from any to any not antispoof in

This rule drops all incoming packets that appear to be coming from another directly connected system but on the wrong interface. So like, a packet with a source address of 192.168.0.0/24, configured on fxp0, but arriving on fxp1 — dropped. 🫶

The setdscp option is the move for (re)marking user traffic; put the following in the appropriate place in the ruleset:

ipfw add setdscp be ip from any to any dscp af11,af21

SELECTIVE MIRRORING

Okay so bestie, if your network has a network traffic analyzer connected to your host directly via a dedicated interface or remotely via RSPAN vlan, you can selectively mirror some Ethernet layer2 frames to the analyzer. No cap, it's giving private investigator vibes. 🕵️‍♀️

First, make sure your firewall is already configured and runs. Then, enable layer2 processing if not already enabled:

             sysctl net.link.ether.ipfw=1

Next, load needed additional kernel modules:

             kldload ng_ether ng_ipfw

Optionally, make system load these modules automatically at startup:

             sysrc kld_list+="ng_ether ng_ipfw"

Next, configure ng_ipfw(4) kernel module to transmit mirrored copies of layer2 frames out via vlan900 interface:

             ngctl connect ipfw: vlan900: 1 lower

Think of "1" here as of "mirroring instance index" and vlan900 is its destination. You can have arbitrary number of instances. Refer to ng_ipfw(4) for details. 💅

At last, actually start mirroring of selected frames using "instance 1". For frames incoming from em0 interface:

             ipfw add ngtee 1 ip from any to 192.168.0.1 layer2 in recv em0

For frames outgoing to em0 interface:

             ipfw add ngtee 1 ip from any to 192.168.0.1 layer2 out xmit em0

For both incoming and outgoing frames while flowing through em0:

             ipfw add ngtee 1 ip from any to 192.168.0.1 layer2 via em0
⚠️ Make sure you do not perform mirroring for already duplicated frames or kernel may hang as there is no safety net. Fr fr, don't do it. 💀

DYNAMIC RULES

In order to protect a site from flood attacks involving fake TCP packets, it's safer to use dynamic rules:

             ipfw add check-state
             ipfw add deny tcp from any to any established
             ipfw add allow tcp from my-net to any setup keep-state

This will let the firewall install dynamic rules only for those connection which start with a regular SYN packet coming from the inside of our network. Dynamic rules are checked when encountering the first occurrence of a check-state, keep-state or limit rule. A check-state rule should usually be placed near the beginning of the ruleset to minimize the amount of work scanning the ruleset. Your mileage may vary. ✨

For more complex scenarios with dynamic rules record-state and defer-action can be used to precisely control creation and checking of dynamic rules. Example of usage of these options are provided in “NETWORK ADDRESS TRANSLATION (NAT)” Section.

To limit the number of connections a user can open you can use the following type of rules:

             ipfw add allow tcp from my-net/24 to any setup limit src-addr 10
             ipfw add allow tcp from any to me setup limit src-addr 4

The former (assuming it runs on a gateway) will allow each host on a /24 network to open at most 10 TCP connections. The latter can be placed on a server to make sure that a single client does not use more than 4 simultaneous connections. Slay, boundaries. 💅

🚨 BEWARE: stateful rules can be subject to denial-of-service attacks by a SYN-flood which opens a huge number of dynamic rules. The effects of such attacks can be partially limited by acting on a set of sysctl(8) variables which control the operation of the firewall. So like, be aware bestie.

Here is a good usage of the list command to see accounting records and timestamp information:

             ipfw -at list

or in short form without timestamps:

             ipfw -a list

which is equivalent to:

             ipfw show

Next rule diverts all incoming packets from 192.168.2.0/24 to divert port 5000:

             ipfw divert 5000 ip from 192.168.2.0/24 to any in

TRAFFIC SHAPING

The following rules show some of the applications of ipfw and dummynet for simulations and the like.

This rule drops random incoming packets with a probability of 5%:

             ipfw add prob 0.05 deny ip from any to any in

A similar effect can be achieved making use of dummynet pipes:

             dnctl add pipe 10 ip from any to any
             dnctl pipe 10 config plr 0.05

We can use pipes to artificially limit bandwidth, e.g. on a machine acting as a router, if we want to limit traffic from local clients on 192.168.2.0/24 we do:

             ipfw add pipe 1 ip from 192.168.2.0/24 to any out
             dnctl pipe 1 config bw 300Kbit/s queue 50KBytes

note that we use the out modifier so that the rule is not used twice. Remember in fact that ipfw rules are checked both on incoming and outgoing packets. 💡

Should we want to simulate a bidirectional link with bandwidth limitations, the correct way is the following:

             ipfw add pipe 1 ip from any to any out
             ipfw add pipe 2 ip from any to any in
             dnctl pipe 1 config bw 64Kbit/s queue 10Kbytes
             dnctl pipe 2 config bw 64Kbit/s queue 10Kbytes

The above can be very useful, e.g. if you want to see how your fancy Web page will look for a residential user who is connected only through a slow link. You should not use only one pipe for both directions, unless you want to simulate a half-duplex medium (e.g. AppleTalk, Ethernet, IRDA). It is not necessary that both pipes have the same configuration, so we can also simulate asymmetric links. 🌈

Should we want to verify network performance with the RED queue management algorithm:

             ipfw add pipe 1 ip from any to any
             dnctl pipe 1 config bw 500Kbit/s queue 100 red 0.002/30/80/0.1

Another typical application of the traffic shaper is to introduce some delay in the communication. This can significantly affect applications which do a lot of Remote Procedure Calls, and where the round-trip-time of the connection often becomes a limiting factor much more than bandwidth:

             ipfw add pipe 1 ip from any to any out
             ipfw add pipe 2 ip from any to any in
             dnctl pipe 1 config delay 250ms bw 1Mbit/s
             dnctl pipe 2 config delay 250ms bw 1Mbit/s

Per-flow queueing can be useful for a variety of purposes. A very simple one is counting traffic:

             ipfw add pipe 1 tcp from any to any
             ipfw add pipe 1 udp from any to any
             ipfw add pipe 1 ip from any to any
             dnctl pipe 1 config mask all

The above set of rules will create queues (and collect statistics) for all traffic. Because the pipes have no limitations, the only effect is collecting statistics. Note that we need 3 rules, not just the last one, because when ipfw tries to match IP packets it will not consider ports, so we would not see connections on separate ports as different ones. 🎯

A more sophisticated example is limiting the outbound traffic on a net with per-host limits, rather than per-network limits:

             ipfw add pipe 1 ip from 192.168.2.0/24 to any out
             ipfw add pipe 2 ip from any to 192.168.2.0/24 in
             dnctl pipe 1 config mask src-ip 0x000000ff bw 200Kbit/s queue
             20Kbytes
             dnctl pipe 2 config mask dst-ip 0x000000ff bw 200Kbit/s queue
             20Kbytes

LOOKUP TABLES

In the following example, we need to create several traffic bandwidth classes and we need different hosts/networks to fall into different classes. We create one pipe for each class and configure them accordingly. Then we create a single table and fill it with IP subnets and addresses. For each subnet/host we set the argument equal to the number of the pipe that it should use. Then we classify traffic using a single rule:

             dnctl pipe 1 config bw 1000Kbyte/s
             dnctl pipe 4 config bw 4000Kbyte/s
             ...
             ipfw table T1 create type addr
             ipfw table T1 add 192.168.2.0/24 1
             ipfw table T1 add 192.168.0.0/27 4
             ipfw table T1 add 192.168.0.2 1
             ...
             ipfw add pipe tablearg ip from 'table(T1)' to any

Using the fwd action, the table entries may include hostnames and IP addresses.

             ipfw table T2 create type addr valtype ipv4
             ipfw table T2 add 192.168.2.0/24 10.23.2.1
             ipfw table T2 add 192.168.0.0/27 router1.dmz
             ...
             ipfw add 100 fwd tablearg ip from any to 'table(T2)'

In the following example per-interface firewall is created:

             ipfw table IN create type iface valtype skipto,fib
             ipfw table IN add vlan20 12000,12
             ipfw table IN add vlan30 13000,13
             ipfw table OUT create type iface valtype skipto
             ipfw table OUT add vlan20 22000
             ipfw table OUT add vlan30 23000
             ..
             ipfw add 100 setfib tablearg ip from any to any recv 'table(IN)'
             in
             ipfw add 200 skipto tablearg ip from any to any recv 'table(IN)'
             in
             ipfw add 300 skipto tablearg ip from any to any xmit 'table(OUT)'
             out

The following example illustrate usage of flow tables:

             ipfw table fl create type flow:src-ip,proto,dst-ip,dst-port
             ipfw table fl add 2001:db8:77::88,tcp,2001:db8:77::99,80 11
             ipfw table fl add 10.0.0.1,udp,10.0.0.2,53 12
             ..
             ipfw add 100 allow ip from any to any flow 'table(fl,11)' recv ix0

The following example illustrate masked table lookups to aid uniform client distribution among multiple NAT instances:

             # Configure NAT instances
             ipfw nat 10 config ip 192.0.2.0
             ipfw nat 11 config ip 192.0.2.1
             ipfw nat 12 config ip 192.0.2.2
             ipfw nat 13 config ip 192.0.2.3

             ipfw table mynats create type addr valtype nat
             # Map external NAT address to NAT instance
             ipfw table mynats add 192.0.2.0 10
             ipfw table mynats add 192.0.2.1 11
             ipfw table mynats add 192.0.2.2 12
             ipfw table mynats add 192.0.2.3 13

             # Map last 2 bits of client's IP address to NAT instance
             ipfw table mynats add 0.0.0.0 10
             ipfw table mynats add 0.0.0.1 11
             ipfw table mynats add 0.0.0.2 12
             ipfw table mynats add 0.0.0.3 13

             # In -> Out NAT, zero out all bits in a client's IP exept
             # 2 least significant prior to table lookup
             ipfw add nat tablearg ip from 10.0.0.0/24 to any
                                     lookup src-ip4:0.0.0.3 mynats
             # Out -> In NAT
             ipfw add nat tablearg ip from any to 192.0.2.0/30
                                     lookup dst-ip mynats

SETS OF RULES

To add a set of rules atomically, e.g. set 18:

             ipfw set disable 18
             ipfw add NN set 18 ...         # repeat as needed
             ipfw set enable 18

To delete a set of rules atomically the command is simply:

             ipfw delete set 18

To test a ruleset and disable it and regain control if something goes wrong:

             ipfw set disable 18
             ipfw add NN set 18 ...         # repeat as needed
             ipfw set enable 18; echo done; sleep 30 && ipfw set disable 18

Here if everything goes well, you press control-C before the "sleep" terminates, and your ruleset will be left active. Otherwise, e.g. if you cannot access your box, the ruleset will be disabled after the sleep terminates thus restoring the previous situation. So like, it's a safety net, bestie. 🛡️

okay so to show rules of the specific set, you literally run:

ipfw set 18 show

and to show rules of the disabled set:

ipfw -S set 18 show

to clear a specific rule counters of the specific set:

ipfw set 18 zero NN

and to delete a specific rule of the specific set:

ipfw set 18 delete NN

🌈 NAT, REDIRECT AND LSNAT 🌈

first, redirect all the traffic to nat instance 123:

ipfw add nat 123 all from any to any

then to configure nat instance 123 to alias all the outgoing traffic with ip 192.168.0.123, blocking all incoming connections, trying to keep same ports on both sides, clearing aliasing table on address change and keeping a log of traffic/link statistics:

ipfw nat 123 config ip 192.168.0.123 log deny_in reset same_ports

or to change address of instance 123, aliasing table will be cleared (see reset option):

ipfw nat 123 config ip 10.0.0.1

to see configuration of nat instance 123:

ipfw nat 123 show config

to show logs of all instances:

ipfw nat show log

to see configurations of all instances:

ipfw nat show config

or a redirect rule with mixed modes could look like:

ipfw nat 123 config redirect_addr 10.0.0.1 10.0.0.66
                         redirect_port tcp 192.168.0.1:80 500
                         redirect_proto udp 192.168.1.43 192.168.1.1
                         redirect_addr 192.168.0.10,192.168.0.11
                                    10.0.0.100  # LSNAT
                         redirect_port tcp 192.168.0.1:80,192.168.0.10:22
                                    500         # LSNAT

or it could be split in:

ipfw nat 1 config redirect_addr 10.0.0.1 10.0.0.66
ipfw nat 2 config redirect_port tcp 192.168.0.1:80 500
ipfw nat 3 config redirect_proto udp 192.168.1.43 192.168.1.1
ipfw nat 4 config redirect_addr 192.168.0.10,192.168.0.11,192.168.0.12
                                         10.0.0.100
ipfw nat 5 config redirect_port tcp
                        192.168.0.1:80,192.168.0.10:22,192.168.0.20:25 500

sometimes you may want to mix NAT and dynamic rules. it could be achieved with record-state and defer-action options. the problem is, you need to create dynamic rule before NAT and check it after NAT actions (or vice versa) to have consistent addresses and ports. rule with keep-state option will trigger activation of existing dynamic state, and action of such rule will be performed as soon as rule is matched. in case of NAT and allow rule packet need to be passed to NAT, not allowed as soon is possible.

there is example of set of rules to achieve this. bear in mind that this is example only and it is not very useful by itself.

on way out, after all checks place this rules:

ipfw add allow record-state defer-action
ipfw add nat 1

and on way in there should be something like this:

ipfw add nat 1
ipfw add check-state

please note, that first rule on way out doesn't allow packet and doesn't execute existing dynamic rules. all it does, create new dynamic rule with allow action, if it is not created yet. later, this dynamic rule is used on way in by check-state rule.

🍓 CONFIGURING CODEL, PIE, FQ-CODEL and FQ-PIE AQM 🍓

codel and pie AQM can be configured for dummynet pipe or queue.

to configure a pipe with codel AQM using default configuration for traffic from 192.168.0.0/24 and 1Mbits/s rate limit, we do:

dnctl pipe 1 config bw 1mbits/s codel
ipfw add 100 pipe 1 ip from 192.168.0.0/24 to any

to configure a queue with codel AQM using different configurations parameters for traffic from 192.168.0.0/24 and 1Mbits/s rate limit, we do:

dnctl pipe 1 config bw 1mbits/s
dnctl queue 1 config pipe 1 codel target 8ms interval 160ms ecn
ipfw add 100 queue 1 ip from 192.168.0.0/24 to any

to configure a pipe with pie AQM using default configuration for traffic from 192.168.0.0/24 and 1Mbits/s rate limit, we do:

dnctl pipe 1 config bw 1mbits/s pie
ipfw add 100 pipe 1 ip from 192.168.0.0/24 to any

to configure a queue with pie AQM using different configuration parameters for traffic from 192.168.0.0/24 and 1Mbits/s rate limit, we do:

dnctl pipe 1 config bw 1mbits/s
dnctl queue 1 config pipe 1 pie target 20ms tupdate 30ms ecn
ipfw add 100 queue 1 ip from 192.168.0.0/24 to any

fq_codel and fq_pie AQM can be configured for dummynet schedulers.

to configure fq_codel scheduler using different configurations parameters for traffic from 192.168.0.0/24 and 1Mbits/s rate limit, we do:

dnctl pipe 1 config bw 1mbits/s
dnctl sched 1 config pipe 1 type fq_codel
dnctl queue 1 config sched 1
ipfw add 100 queue 1 ip from 192.168.0.0/24 to any

to change fq_codel default configuration for a sched such as disable ECN and change the target to 10ms, we do:

dnctl sched 1 config pipe 1 type fq_codel target 10ms noecn

similar to fq_codel, to configure fq_pie scheduler using different configurations parameters for traffic from 192.168.0.0/24 and 1Mbits/s rate limit, we do:

dnctl pipe 1 config bw 1mbits/s
dnctl sched 1 config pipe 1 type fq_pie
dnctl queue 1 config sched 1
ipfw add 100 queue 1 ip from 192.168.0.0/24 to any

the configurations of fq_pie sched can be changed in a similar way as for fq_codel

👑 SEE ALSO 👑

💜 HISTORY 💜

the ipfw utility first appeared in FreeBSD 2.0. dummynet was introduced in FreeBSD 2.2.8. stateful extensions were introduced in FreeBSD 4.0. ipfw2 was introduced in Summer 2002.

🎀 AUTHORS 🎀

  • Ugen J. S. Antsilevich
  • Poul-Henning Kamp
  • Alex Nash
  • Archie Cobbs
  • Luigi Rizzo
  • Rasool Al-Saadi

API based upon code written by Daniel Boulet for BSDI.

Dummynet has been introduced by Luigi Rizzo in 1997-1998.

some early work (1999-2000) on the dummynet traffic shaper supported by Akamba Corp.

the ipfw core (ipfw2) has been completely redesigned and reimplemented by Luigi Rizzo in summer 2002. further actions and options have been added by various developers over the years.

in-kernel NAT support written by Paolo Pisati <piso@FreeBSD.org> as part of a Summer of Code 2005 project.

SCTP nat support has been developed by The Centre for Advanced Internet Architectures (CAIA) <http://www.caia.swin.edu.au>. the primary developers and maintainers are David Hayes and Jason But. for further information visit: ⟨http://www.caia.swin.edu.au/urp/SONATA⟩

delay profiles have been developed by Alessandro Cerri and Luigi Rizzo, supported by the European Commission within Projects Onelab and Onelab2.

CoDel, PIE, FQ-CoDel and FQ-PIE AQM for Dummynet have been implemented by The Centre for Advanced Internet Architectures (CAIA) in 2016, supported by The Comcast Innovation Fund. the primary developer is Rasool Al-Saadi.

💀 BUGS 💀

the syntax has grown over the years and sometimes it might be confusing. unfortunately, backward compatibility prevents cleaning up mistakes made in the definition of the syntax.

!!! WARNING !!! misconfiguring the firewall can put your computer in an unusable state, possibly shutting down network services and requiring console access to regain control of it.

incoming packet fragments diverted by divert are reassembled before delivery to the socket. the action used on those packet is the one from the rule which matches the first fragment of the packet.

packets diverted to userland, and then reinserted by a userland process may lose various packet attributes. the packet source interface name will be preserved if it is shorter than 8 bytes and the userland process saves and reuses the sockaddr_in (as does natd(8)); otherwise, it may be lost. if a packet is reinserted in this manner, later rules may be incorrectly applied, making the order of divert rules in the rule sequence very important.

Dummynet drops all packets with IPv6 link-local addresses.

rules using uid or gid may not behave as expected. in particular, incoming SYN packets may have no uid or gid associated with them since they do not yet belong to a TCP connection, and the uid/gid associated with a packet may not be as expected if the associated process calls setuid(2) or similar system calls.

rule syntax is subject to the command line environment and some patterns may need to be escaped with the backslash character or quoted appropriately.

due to the architecture of libalias(3), ipfw nat is not compatible with the TCP segmentation offloading (TSO). thus, to reliably nat your network traffic, please disable TSO on your NICs using ifconfig(8).

ICMP error messages are not implicitly matched by dynamic rules for the respective conversations. to avoid failures of network error detection and path MTU discovery, ICMP error messages may need to be allowed explicitly through static rules.

rules using call and return actions may lead to confusing behaviour if ruleset has mistakes, and/or interaction with other subsystems (netgraph, dummynet, etc.) is used. one possible case for this is packet leaving ipfw in subroutine on the input pass, while later on output encountering unpaired return first. as the call stack is kept intact after input pass, packet will suddenly return to the rule number used on input pass, not on output one. order of processing should be checked carefully to avoid such mistakes.

and that’s the whole page bestie 💖 source/version is Debian, the date is March 1, 2026, and the tag is IPFW(8) — periodt 🎀✨

📦 where this page comes from