IPFW(8) FreeBSD System Manager’s Manual IPFW(8)
NAME ipfw -- 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} ipfw disable {firewall | altq | one_pass | debug | verbose | dyn_keepalive} LOOKUP TABLES ipfw table number add addr[/masklen] [value] ipfw table number delete addr[/masklen] ipfw table {number | all} flush ipfw table {number | all} list DUMMYNET CONFIGURATION (TRAFFIC SHAPER AND PACKET SCHEDULER) ipfw {pipe | queue | sched} number config config-options ipfw [-s [field]] {pipe | queue | sched} {delete | list | show} [number ...] IN-KERNEL NAT ipfw [-q] nat number config config-options ipfw [-cfnNqS] [-p preproc [preproc-flags]] pathname DESCRIPTION The ipfw utility is the user interface for controlling the ipfw(4) fire- wall, 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 des- tination 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 rein- jected 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 or 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. 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 com- mands 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. -e When listing and -d is specified, also show expired dynamic rules. -f Do not ask 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. -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 pass- ing 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 ses- sion will be closed and the rest of the ruleset will not be pro- cessed. 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 charac- ter, 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 pre- processor 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 ipfw 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. If the world and the kernel get out of sync the ipfw ABI may break, pre- venting you from being able to add any rules. This can adversely effect 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 A packet is checked against the active ruleset in multiple places in the protocol stack, under control of several sysctl variables. These places and variables are shown below, and it is important to have this picture in mind in order to design a correct ruleset. ^ 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. 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(). Also note that each packet is always checked against the complete rule- set, irrespective of the place where the check occurs, or the source of the packet. 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. It is thus the responsibility of the programmer, if necessary, to write a suitable ruleset to differentiate among the pos- sible places. skipto rules can be 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). SYNTAX In general, each keyword or argument must be provided as a separate com- mand 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). 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" RULE FORMAT The format of firewall rules is the following: [rule_number] [set set_number] [prob match_probability] action [log [logamount number]] [altq queue] [{tag | untag} number] body where the body of the rule specifies which information is used for fil- tering packets, among the following: Layer-2 header fields When available IPv4 and IPv6 Protocol 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, data- gram length, identification, fragment flag (non-zero IP off- set), 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, acknowl- edgment number, window TCP options ICMP types for ICMP packets ICMP6 types for ICMP6 packets User/group ID When the packet can be associ- ated 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 (rout- ing 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. 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. 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. 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. 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. Note: this condition is checked before any other condition, including ones such as keep-state or check-state which might have side effects. log [logamount number] Packets matching a rule with the log keyword will be made avail- able for logging 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. There is no overhead if no bpf(4) is attached to the pseudo interface. 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. 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. tag number When a packet matches a rule with the tag keyword, the numeric tag for the given number in the range 1..65534 will be attached to the packet. The tag acts as an internal marker (it is not sent out over the wire) that can be used to identify these pack- ets later on. This can be used, for example, to provide trust between interfaces and to start doing policy-based filtering. A packet can have multiple tags at the same time. Tags are "sticky", meaning once a tag is applied to a packet by a matching rule it exists until explicit removal. Tags are kept with the packet everywhere within the kernel, but are lost when packet leaves the kernel, for example, on transmitting packet out to the network or sending packet to a divert(4) socket. To check for previously applied tags, use the tagged rule option. To delete previously applied tag, use the untag keyword. Note: since tags are kept with the packet everywhere in ker- nelspace, they can be set and unset anywhere in the kernel net- work subsystem (using the mbuf_tags(9) facility), not only by means of the ipfw(4) tag and untag keywords. For example, there can be a specialized netgraph(4) node doing traffic analyzing and tagging for later inspecting in 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 found, removed from it. Other tags bound to packet, if present, are left untouched. altq queue When a packet matches a rule with the altq keyword, the ALTQ identifier for the given queue (see altq(4)) will be attached. Note that this ALTQ tag is only meaningful for packets going "out" of IPFW, and not being rejected or going to divert sockets. Note that if there is insufficient memory at the time the packet is processed, it will not be tagged, so it is wise to make your ALTQ "default" queue policy account for this. 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 in containing the queue identifiers in the kernel will likely have gone stale and need to be reloaded. Stale queue identifiers will probably result in misclassifica- tion. 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, as the actual rule action is followed always after adding an ALTQ tag. 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 rule. The search terminates. check-state Checks the packet against the dynamic ruleset. If a match is found, execute the action associated with the rule which gener- ated 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. count Update counters for all packets that match rule. The search con- tinues 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. For IPv4, 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 ter- minates if this rule matches. If ipaddr is a local address, then matching packets will be for- warded 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 speci- fied) 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 layer-2 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 is intended for use with transparent proxy servers. To enable fwd a custom kernel needs to be compiled with the option options IPFIREWALL_FORWARD. nat nat_nr | tablearg Pass packet to a nat instance (for network address translation, address redirect, etc.): see the NETWORK ADDRESS TRANSLATION (NAT) 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. 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 termi- nates. 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 termi- nates. skipto number | tablearg Skip all subsequent rules numbered less than number. The search continues with the first rule numbered number or higher. It is possible to use the tablearg keyword with a skipto for a computed skipto, but care should be used, as no destination caching is possible in this case so the rules are always walked to find it, starting from the skipto. call number | tablearg The current rule number is saved in the internal stack and rule- set 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 number of this call rule plus one or higher (the same behaviour as with packets returning from 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 can't be allocated or stack over- flowed/undeflowed. Internally stack for rule numbers is implemented using mbuf_tags(9) facility and currently has size of 16 entries. As mbuf tags are lost when packet leaves the kernel, divert should not be used in subroutines to avoid endless loops and other unde- sired effects. return Takes rule number saved to internal stack by the last call action and returns ruleset processing to the first rule with number greater than number of corresponding call rule. See description of the call action for more details. Note that return rules usually end a ``subroutine'' and thus are unconditional, but ipfw command-line utility currently requires every action except check-state to have body. While it is some- times useful to return only on some packets, usually you want to print just ``return'' for readability. A workaround for this is to use new syntax and -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 be fixed in future releases. tee port Send a copy of packets matching this rule to the divert(4) socket bound to port port. The search continues with the next rule. unreach code Discard packets that match this rule, and try to send 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 search terminates. unreach6 code Discard packets that match this rule, and try 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. The search terminates. netgraph cookie Divert packet into netgraph with given cookie. The search termi- nates. If packet is later returned from netgraph it is either accepted or continues with the next rule, depending on net.inet.ip.fw.one_pass sysctl variable. ngtee cookie A copy of packet is diverted into netgraph, original packet con- tinues with the next rule. See ng_ipfw(4) for more information on netgraph and ngtee actions. setfib fibnum | tablearg The packet is tagged so as to use the FIB (routing table) fibnum in any subsequent forwarding decisions. Initially this is lim- ited to the values 0 through 15, see setfib(1). Processing con- tinues at the next rule. It is possible to use the tablearg key- word with a setfib. If tablearg value is not within compiled FIB range packet fib is set to 0. reass Queue and reassemble ip fragments. If the packet is not frag- mented, counters are updated and processing continues with the next rule. If the packet is 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, else packet is allowed to pass and search terminates. If the packet is a frag- ment in the middle, it is consumed and processing stops immedi- ately. Fragments 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 do not contain port numbers, they should be avoided with the reass rule. Alternatively, direction- based (like in / out ) and source-based (like via ) match pat- terns can be used to select fragments. Usually a simple rule like: # reassemble incoming fragments ipfw add reass all from any to any in is all you need at the beginning of your ruleset. RULE BODY The body of a rule contains zero or more patterns (such as specific source and destination addresses or ports, protocol options, incoming or outgoing interfaces, etc.) that the packet must match in order to be recognised. In general, the patterns are connected by (implicit) and operators -- i.e., all must match in order for the rule to match. Indi- vidual patterns can be prefixed by the not operator to reverse the result of the match, as in ipfw add 100 allow ip from not 1.2.3.4 to any Additionally, sets of alternative match patterns (or-blocks) can be con- structed by putting the patterns in lists enclosed between parentheses ( ) or braces { }, and using the or operator as follows: ipfw add 100 allow ip from { x or not y or z } to any Only one level of parentheses is allowed. Beware that most shells have special meanings for parentheses or braces, so it is advisable to put a backslash \ in front of them to prevent such interpretations. 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 specify that the content of a required field is irrelevant. The rule body has the following format: [proto from src to dst] [options] The first part (proto from src to dst) is for backward compatibility with earlier versions of FreeBSD. In modern FreeBSD any match pattern (including MAC headers, IP protocols, addresses and ports) can be speci- fied in the options section. Rule fields have the following meaning: 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 the following keywords: ip4 | ipv4 Matches IPv4 packets. ip6 | ipv6 Matches IPv6 packets. ip | all Matches any packet. 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 con- venience only but its use is deprecated. 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(number[,value]) | 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(number[,value]) Matches any IPv4 address for which an entry exists in the lookup table number. If an optional 32-bit unsigned value is also specified, an entry will match only if it has this value. See the LOOKUP TABLES section below for more information 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. As an example, 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 is better to resort to the addr/masklen format for contiguous masks, which is more compact and less error-prone. addr-set: addr[/masklen]{list} list: {num | num-num}[,list] Matches all addresses with base address addr (specified as an IP address, a network number, or a hostname) and whose last byte is in the list between braces { } . Note that there must be no spa- ces between braces and numbers (spaces after commas are allowed). Elements of the list can be specified as single entries or ranges. The masklen field is used to limit the size of the set of addresses, and can have any value between 24 and 32. If not specified, it will be assumed as 24. This format is particularly useful to handle sparse address sets within a single rule. Because the matching occurs using a bit- mask, it takes constant time and dramatically reduces the com- plexity of rulesets. As an example, 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 the following IP addresses: 1.2.3.128, 1.2.3.35 to 1.2.3.55, 1.2.3.89 . addr6-list: ip6-addr[,addr6-list] ip6-addr: A host or subnet specified one of the following 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 or a hostname) and mask width of masklen bits. No support for sets of IPv6 addresses is provided because IPv6 addresses are typically random past the initial prefix. ports: {port | port-port}[,ports] For protocols which support port numbers (such as 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 one 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. 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. 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. 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 Rout- ing 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 encapsulated security payload headers (esp). fib fibnum Matches a packet that has been tagged to use the given FIB (rout- ing table) number. 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. frag Matches packets that are fragments and not the first fragment of an IP datagram. Note that these packets will not have the next protocol header (e.g. TCP, UDP) so options that look into these headers cannot match. 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 prisonID Matches all TCP or UDP packets sent by or received for the jail whos prison ID is prisonID. 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 sup- ported 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 IPSEC_FILTERTUNNEL option, 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 with- out IPSEC support. It does not affect rule processing when given and the rules are handled as if with no ipsec flag. iptos spec Matches IPv4 packets whose tos field contains the comma separated list of service types specified in spec. The supported IP types of service are: lowdelay (IPTOS_LOWDELAY), throughput (IPTOS_THROUGHPUT), reliability (IPTOS_RELIABILITY), mincost (IPTOS_MINCOST), congestion (IPTOS_ECN_CE). The absence of a particular type may be denoted with a `!'. ipttl ttl-list Matches IPv4 packets whose time to live is included in ttl-list, which is either a single value or a list of values or ranges specified in the same way as ports. ipversion ver Matches IP packets whose IP version field is ver. keep-state Upon a match, the firewall will create a dynamic rule, whose 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) vari- ables), and the lifetime is refreshed every time a matching packet is found. 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 The firewall will only allow N connections with the same set of parameters as specified in the rule. One or more of source and destination addresses and ports can be specified. Currently, only IPv4 flows are supported. lookup {dst-ip | dst-port | src-ip | src-port | uid | jail} N Search an entry in lookup table N that matches the field speci- fied as argument. 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. { MAC | mac } dst-mac src-mac Match packets with a given dst-mac and src-mac addresses, speci- fied as the any keyword (matching any MAC address), or six groups of hex digits separated by colons, and optionally followed by a mask indicating the significant bits. The mask may be specified using either of the following methods: 1. A slash (/) followed by the number of significant bits. For example, an address with 33 significant bits could be specified as: MAC 10:20:30:40:50:60/33 any 2. An ampersand (&) followed by a bitmask specified as six groups of hex digits separated by colons. For example, an address in which the last 16 bits are significant could be specified 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. Note that the order of MAC addresses (destination first, source second) is the same as on the wire, but the opposite of the one used for IP addresses. mac-type mac-type Matches packets whose Ethernet Type field corresponds to one of those specified as argument. mac-type is specified in 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, 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. recv | xmit | via {ifX | if* | table(number[,value | ipno | any])} Matches packets received, transmitted or going through, respec- tively, the interface specified by exact name (ifX), by device name (if*), by IP address, or through some interface. The via keyword causes the interface to always be checked. If recv or xmit is used instead of via, then only the receive or transmit interface (respectively) is checked. By specifying both, it is possible to 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. 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 avail- able as tablearg value, which in turn can be used as 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) speci- fied 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 tag rule action parameter (see it's description for details on tags). 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. 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: mss (maximum segment size), window (tcp window advertisement), sack (selective ack), ts (rfc1323 timestamp) and cc (rfc1644 t/tcp connection count). The absence of a particular option may be denoted with a `!'. uid user Match all TCP or UDP packets sent by or received for a user. A user may be matched by name or identification number. verrevpath For incoming packets, a routing table lookup is done on the packet's source address. If the interface on which the packet entered the system matches the outgoing interface for the route, the packet matches. If the interfaces do not match up, the packet does not match. All outgoing packets or packets with no incoming interface match. The name and functionality of the option is intentionally similar to the Cisco IOS command: ip verify unicast reverse-path This option can be used to make anti-spoofing rules to reject all packets with source addresses not from this interface. See also the option antispoof. versrcreach For incoming packets, a routing table lookup is done on the packet's source address. If a route to the source address exists, but not the default route or a blackhole/reject route, the packet matches. Otherwise, the packet does not match. All outgoing packets match. The name and functionality of the option is intentionally similar to the Cisco IOS command: ip verify unicast source reachable-via any This option can be used to make anti-spoofing rules to reject all packets whose source address is unreachable. antispoof For incoming packets, the packet's source address is checked if it belongs to a directly connected network. If the network is directly connected, then the interface the packet came on in is compared to the interface the network is connected to. When incoming interface and directly connected interface are not the same, the packet does not match. Otherwise, the packet does match. All outgoing packets match. This option can be used to make anti-spoofing rules to reject all packets that pretend to be from a directly connected network but do not come in through that interface. This option is similar to but more restricted than verrevpath because it engages only on packets with source addresses of directly connected networks instead of all source addresses. LOOKUP TABLES Lookup tables are useful to handle large sparse sets of addresses or other search keys (e.g. ports, jail IDs, interface names). In the rest of this section we will use the term ``address''. There may be up to 4096 different lookup tables, numbered 0 to 4095. Each entry is represented by an addr[/masklen] and will match all addresses with base addr (specified as an IPv4/IPv6 address, a hostname or an unsigned integer) and mask width of masklen bits. If masklen is not 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. Associ- ated with each entry is a 32-bit unsigned value, which can optionally be checked by a rule matching code. When adding an entry, if value is not specified, it defaults to 0. An entry can be added to a table (add), or removed from a table (delete). A table can be examined (list) or flushed (flush). Internally, each table is stored in a Radix tree, the same way as the routing table (see route(4)). Lookup tables currently support only ports, jail IDs, IPv4/IPv6 addresses and interface names. Wildcards is not supported for interface names. 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 number of rules in some configura- tions. If two tables are used in a rule, the result of the second (des- tination) is used. The tablearg argument can be used with the following actions: nat, pipe, queue, divert, tee, netgraph, ngtee, fwd, skipto, setfib, action parameters: tag, untag, rule options: limit, tagged. When used with fwd it is possible to supply table entries with values that are in the form of IP addresses or hostnames. See the EXAMPLES Sec- tion for example usage of tables and the tablearg keyword. 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, and should therefore try keep the ruleset compact between the skipto and the target rules. SETS OF RULES Each rule belongs to one of 32 different sets , numbered 0 to 31. Set 31 is reserved for the default rule. By default, rules are put in set 0, unless you use the set N attribute when entering a new rule. Sets can be individually and atomically enabled or disabled, so this mechanism permits an easy way to store mul- tiple configurations of the firewall and quickly (and atomically) switch between them. The command to enable/disable sets is ipfw set [disable number ...] [enable number ...] where multiple enable or disable sections can be specified. Command exe- cution is atomic on all the sets specified in the command. By default, all sets are enabled. 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. 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 and limit options of rules. Dynamic rules are created when a packet matches a keep-state or 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). 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. Note that no additional attributes other than protocol and IP addresses and ports are checked on dynamic rules. The typical use of dynamic rules is to keep a closed firewall configura- tion, 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 ipfw add allow tcp from my-subnet to any setup keep-state 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 ipfw add allow udp from my-subnet to any keep-state 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. 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. 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 param- eters (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. 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, inter- nally, a dummynet's pipe is implemented exactly as above). 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 signif- icantly 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]{bit/s|Byte/s}. A value of 0 (default) means unlimited bandwidth. The unit must immediately follow the number, as in ipfw pipe 1 config bw 300Kbit/s If a device name is specified instead of a numeric value, as in ipfw 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 effec- tive burst size is calculated as follows: MAX( size , bw * pipe_idle_time). profile filename A file specifying the additional overhead incurred in the trans- mission 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. cumulative probability 1.0 ^ | L +-- loss-level x | ****** | * | ***** | * | ** | * +-------*-------------------> delay The empirical curve may have both vertical and horizontal lines. Vertical lines represent constant delay for a range of probabili- ties. Horizontal lines correspond to a discontinuity in the delay distribution: the pipe will use the largest delay for a given probability. The file format is the following, with whitespace acting as a separator and '#' indicating the beginning a comment: name identifier optional name (listed by "ipfw 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 parame- ter 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 for- mat 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 cho- sen 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 "sam- ples" parameter: ipfw utility will sort and interpolate the curve as needed. Example of a profile file: 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 sched- uler: type {fifo | wf2q+ | rr | qfq} 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 sched- uler). 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 vari- ant of WF2Q+, with similar service guarantees and O(1) processing costs (roughly, 200-250ns per packet). In addition to the type, all parameters allowed for a pipe can also be specified for a scheduler. 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 fur- ther classified into multiple flows, each of which is then sent to a different dynamic pipe or queue. A flow identifier is con- structed 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 match- ing 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 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. The loss rate is internally represented on 31 bits. queue {slots | sizeKbytes} Queue size, in slots or KBytes. Default value is 50 slots, which is the typical queue size for Ethernet devices. Note that for slow speed links you should keep the queue size short or your traffic might be affected by a significant queueing delay. E.g., 50 max- sized ethernet packets (1500 bytes) mean 600Kbit or 20s of queue on a 30Kbit/s pipe. Even worse effects can result if you get packets from an interface with a much larger MTU, e.g. the loopback inter- face 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 that can be specified. red | gred w_q/min_th/max_th/max_p Make use of the RED (Random Early Detection) queue management algo- rithm. w_q and max_p are floating point numbers between 0 and 1 (0 not included), while min_th and max_th are integer numbers specify- ing thresholds for queue management (thresholds are computed in bytes if the queue has been defined in bytes, in slots otherwise). The dummynet also supports the gentle RED variant (gred). Three sysctl(8) variables can be used to 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). 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. Care should be taken to ensure that link-local packets are not passed to dummynet. CHECKLIST Here are some important points to consider when designing your rules: +o Remember that you filter both packets going in and out. Most connec- tions need packets going in both directions. +o 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. +o Do not forget the loopback interface. FINE POINTS +o There are circumstances where fragmented datagrams are uncondition- ally 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. +o 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. +o 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 fol- lowing 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. +o 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. 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. NETWORK ADDRESS TRANSLATION (NAT) ipfw support in-kernel NAT using the kernel version of libalias(3). The nat configuration command is the following: nat nat_number config nat-configuration The following parameters can be configured: ip ip_address Define an ip address to use for aliasing. if nic Use ip address of NIC for aliasing, dynamically changing it if NIC's ip address changes. log Enable logging on this nat instance. deny_in Deny any incoming connection from outside world. 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 an unregistered address spaces will be ignored. reset Reset table of the packet aliasing engine on address change. reverse Reverse the way libalias handles aliasing. proxy_only Obey transparent proxy rules only, packet aliasing is not per- formed. skip_global Skip instance in case of global state lookup (see below). Some specials value can be supplied instead of nat_number: global Looks up translation state in all configured nat instances. If an entry is found, packet is aliased according to that entry. If no entry was found in any of the instances, packet is passed unchanged, and no new entry will be created. See section MULTIPLE INSTANCES in natd(8) for more information. tablearg Uses argument supplied in lookup table. See LOOKUP TABLES section below for more information on lookup tables. To let the packet continue after being (de)aliased, set the sysctl vari- able net.inet.ip.fw.one_pass to 0. For more information about aliasing modes, refer to libalias(3). See Section EXAMPLES for some examples about nat usage. REDIRECT AND LSNAT SUPPORT IN IPFW Redirect and LSNAT support follow closely the syntax used in natd(8). 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 is that sctp nat does not do port translation. Since the local and global side ports will be the same, there is no need to specify both. 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_ta- ble size will only change for new nat instances. See SYSCTL VARIABLES for more info. LOADER TUNABLES Tunables can be set in loader(8) prompt, loader.conf(5) or kenv(1) before ipfw module gets loaded. net.inet.ip.fw.default_to_accept: 0 Defines ipfw last rule behavior. This value overrides options IPFW_DEFAULT_TO_(ACCEPT|DENY) from kernel configuration file. net.inet.ip.fw.tables_max: 128 Defines number of tables available in ipfw. Number cannot exceed 65534. SYSCTL VARIABLES A 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 mes- sages. 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) pack- ets 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). 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 debug- ging) 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 pack- ets 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.expire: 1 Lazily delete dynamic pipes/queue once they have no pending traf- fic. 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.hash_size: 64 Default size of the hash table used for dynamic pipes/queues. This value is used when no buckets option is specified when con- figuring a pipe/queue. net.inet.ip.dummynet.io_fast: 0 If set to a non-zero value, the ``fast'' mode of dummynet opera- tion (see above) is enabled. net.inet.ip.dummynet.io_pkt Number of packets passed to dummynet. net.inet.ip.dummynet.io_pkt_drop Number of packets dropped by dummynet. net.inet.ip.dummynet.io_pkt_fast Number of packets bypassed by the dummynet scheduler. net.inet.ip.dummynet.max_chain_len: 16 Target value for the maximum number of pipes/queues in a hash bucket. The product max_chain_len*hash_size is used to determine the threshold over which empty pipes/queues will be expired even when net.inet.ip.dummynet.expire=0. net.inet.ip.dummynet.red_lookup_depth: 256 net.inet.ip.dummynet.red_avg_pkt_size: 512 net.inet.ip.dummynet.red_max_pkt_size: 1500 Parameters used in the computations of the drop probability for the RED algorithm. net.inet.ip.dummynet.pipe_byte_limit: 1048576 net.inet.ip.dummynet.pipe_slot_limit: 100 The maximum queue size that can be specified in bytes or packets. These limits prevent accidental exhaustion of resources such as mbufs. If you raise these limits, you should make sure the sys- tem is configured so that sufficient resources are available. net.inet.ip.fw.autoinc_step: 100 Delta between rule numbers when auto-generating them. The value must be in the range 1..1000. net.inet.ip.fw.curr_dyn_buckets: net.inet.ip.fw.dyn_buckets The current number of buckets in the hash table for dynamic rules (readonly). net.inet.ip.fw.debug: 1 Controls debugging messages produced by ipfw. net.inet.ip.fw.default_rule: 65535 The default rule number (read-only). By the design of ipfw, the default rule is the last one, so its number can also serve as the highest number allowed for a rule. net.inet.ip.fw.dyn_buckets: 256 The number of buckets in the hash table for dynamic rules. Must be a power of 2, up to 65536. It only takes effect when all dynamic rules have expired, so you are advised to use a flush command to make sure that the hash table is resized. net.inet.ip.fw.dyn_count: 3 Current number of dynamic rules (read-only). net.inet.ip.fw.dyn_keepalive: 1 Enables generation of keepalive packets for keep-state rules on TCP sessions. A keepalive is generated to both sides of the con- nection every 5 seconds for the last 20 seconds of the lifetime of the rule. net.inet.ip.fw.dyn_max: 8192 Maximum number of dynamic rules. When you hit this limit, no more dynamic rules can be installed until old ones expire. net.inet.ip.fw.dyn_ack_lifetime: 300 net.inet.ip.fw.dyn_syn_lifetime: 20 net.inet.ip.fw.dyn_fin_lifetime: 1 net.inet.ip.fw.dyn_rst_lifetime: 1 net.inet.ip.fw.dyn_udp_lifetime: 5 net.inet.ip.fw.dyn_short_lifetime: 30 These variables control the lifetime, in seconds, of dynamic rules. Upon the initial SYN exchange the lifetime is kept short, then increased after both SYN have been seen, then decreased again during the final FIN exchange or when a RST is received. Both dyn_fin_lifetime and dyn_rst_lifetime must be strictly lower than 5 seconds, the period of repetition of keepalives. The firewall enforces that. net.inet.ip.fw.enable: 1 Enables the firewall. Setting this variable to 0 lets you run your machine without firewall even if compiled in. net.inet6.ip6.fw.enable: 1 provides the same functionality as above for the IPv6 case. net.inet.ip.fw.one_pass: 1 When set, the packet exiting from the dummynet pipe or from ng_ipfw(4) node is not passed though the firewall again. Other- wise, after an action, the packet is reinjected into the firewall at the next rule. net.inet.ip.fw.tables_max: 128 Maximum number of tables. net.inet.ip.fw.verbose: 1 Enables verbose messages. net.inet.ip.fw.verbose_limit: 0 Limits the number of messages produced by a verbose firewall. net.inet6.ip6.fw.deny_unknown_exthdrs: 1 If enabled packets with unknown IPv6 Extension Headers will be denied. net.link.ether.ipfw: 0 Controls whether layer-2 packets are passed to ipfw. Default is no. net.link.bridge.ipfw: 0 Controls whether bridged packets are passed to ipfw. Default is no. EXAMPLES There are far too many possible uses of ipfw so this Section will only give a small set of examples. BASIC PACKET FILTERING This command adds an entry which denies all tcp packets from cracker.evil.org to the telnet port of wolf.tambov.su from being for- warded by the host: ipfw add deny tcp from cracker.evil.org to wolf.tambov.su telnet This one disallows any connection from the entire cracker's network to my host: ipfw add deny ip from 123.45.67.0/24 to my.host.org A first and efficient way to limit access (not using dynamic rules) is the use of the following rules: 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 The first rule will be a quick match for normal TCP packets, but it will not match the initial SYN packet, which will be matched by the setup rules only for selected source/destination pairs. All other SYN packets will be rejected by the final deny rule. If you administer one or more subnets, you can take advantage of the address sets and or-blocks and write extremely compact rulesets which selectively enable services to blocks of clients, as below: 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 ... The verrevpath option could be used to do automated anti-spoofing by adding the following to the top of a ruleset: ipfw add deny ip from any to any not verrevpath in This rule drops all incoming packets that appear to be coming to the sys- tem on the wrong interface. For example, a packet with a source address belonging to a host on a protected internal network would be dropped if it tried to enter the system from an external interface. The antispoof option could be used to do similar but more restricted anti-spoofing by adding the following to 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. For exam- ple, a packet with a source address of 192.168.0.0/24, configured on fxp0, but coming in on fxp1 would be dropped. DYNAMIC RULES In order to protect a site from flood attacks involving fake TCP packets, it is 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 connec- tion which start with a regular SYN packet coming from the inside of our network. Dynamic rules are checked when encountering the first check-state or keep-state 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. To limit the number of connections a user can open you can use the fol- lowing 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. 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. 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: ipfw add pipe 10 ip from any to any ipfw pipe 10 config plr 0.05 We can use pipes to artificially limit bandwidth, e.g. on a machine act- ing 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 ipfw 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 outgo- ing packets. Should we want to simulate a bidirectional link with bandwidth limita- tions, 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 ipfw pipe 1 config bw 64Kbit/s queue 10Kbytes ipfw 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 manage- ment algorithm: ipfw add pipe 1 ip from any to any ipfw 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 band- width: ipfw add pipe 1 ip from any to any out ipfw add pipe 2 ip from any to any in ipfw pipe 1 config delay 250ms bw 1Mbit/s ipfw 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 ipfw 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 ipfw pipe 1 config mask src-ip 0x000000ff bw 200Kbit/s queue 20Kbytes ipfw 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 accord- ingly. 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: ipfw pipe 1 config bw 1000Kbyte/s ipfw pipe 4 config bw 4000Kbyte/s ... ipfw table 1 add 192.168.2.0/24 1 ipfw table 1 add 192.168.0.0/27 4 ipfw table 1 add 192.168.0.2 1 ... ipfw add pipe tablearg ip from table(1) to any Using the fwd action, the table entries may include hostnames and IP addresses. ipfw table 1 add 192.168.2.0/24 10.23.2.1 ipfw table 1 add 192.168.0.0/27 router1.dmz ... ipfw add 100 fwd tablearg ip from any to table(1) In the following example per-interface firewall is created: ipfw table 10 add vlan20 12000 ipfw table 10 add vlan30 13000 ipfw table 20 add vlan20 22000 ipfw table 20 add vlan30 23000 .. ipfw add 100 ipfw skipto tablearg ip from any to any recv 'table(10)' in ipfw add 200 ipfw skipto tablearg ip from any to any xmit 'table(10)' out 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" ter- minates, and your ruleset will be left active. Otherwise, e.g. if you cannot access your box, the ruleset will be disabled after the sleep ter- minates thus restoring the previous situation. To show rules of the specific set: ipfw set 18 show 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 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 keep- ing 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 the instances in range 111-999: ipfw nat 111-999 show To see configurations of all instances: ipfw nat show config Or a redirect rule with mixed modes could looks 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 SEE ALSO cpp(1), m4(1), altq(4), divert(4), dummynet(4), if_bridge(4), ip(4), ipfirewall(4), ng_ipfw(4), protocols(5), services(5), init(8), kldload(8), reboot(8), sysctl(8), syslogd(8) 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. 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 developer 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 develop- ers and maintainers are David Hayes and Jason But. For further informa- tion 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. 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, incom- ing 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 appropri- ately. Due to the architecture of libalias(3), ipfw nat is not compatible with the TCP segmentation offloading (TSO). Thus, to reliably nat your net- work 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. FreeBSD 8.3 July 3, 2012 FreeBSD 8.3