Packet filter configuration file
The pf packet filter modifies, drops or passes packets according to rules or definitions specified in pf.conf. The default location of this file is /etc/pf.conf.
The pf.conf file can include the following types of statements:
With the exception of macros and tables, the types of statements should be grouped and appear in pf.conf in the order shown above, as this matches the operation of the underlying packet filtering engine. By default, pfctl enforces this order (see set require-order below).
Much as for cpp or m4, you can define macros that will later be expanded in context. Macro names must start with a letter, and may contain letters, digits and underscores. Macro names may not be reserved words (for example, pass, in, out). Macros aren't expanded inside quotes.
For example:
ext_if = "kue0" all_ifs = "{" $ext_if lo0 "}" pass out on $ext_if from any to any keep state pass in on $ext_if proto tcp from any to any port 25 keep state
Tables are named structures that can hold a collection of addresses and networks. Lookups against tables in pf are relatively fast, making a single rule with tables much more efficient, in terms of processor usage and memory consumption, than a large number of rules that differ only in IP address (either created explicitly or automatically by rule expansion).
You can use tables as the source or destination of filter rules, scrub rules or translation rules such as nat or rdr (see below for details on the various rule types) You can also use tables for the redirect address of nat and rdr rules and in the routing options of filter rules, but only for round-robin pools.
You can define tables with any of the following pfctl mechanisms. As with macros, reserved words may not be used as table names.
Tables may be defined with the following attributes:
For example:
table <private> const { 10/8, 172.16/12, 192.168/16 } table <badhosts> persist block on fxp0 from { <private>, <badhosts> } to any
creates a table called private, to hold RFC 1918 private network blocks, and a table called badhosts, which is initially empty. A filter rule is set up to block all traffic coming from addresses listed in either table.
The private table can't have its contents changed, and the badhosts table will exist even when no active filter rules refer to it. Addresses may later be added to the badhosts table, so that traffic from these hosts can be blocked by using:
# pfctl -t badhosts -Tadd 204.92.77.111
A table can also be initialized with an address list specified in one or more external files, using the following syntax:
table <spam> persist file "/etc/spammers" file "/etc/openrelays" block on fxp0 from <spam> to any
The files /etc/spammers and /etc/openrelays list IP addresses, one per line. Any lines beginning with a # are treated as comments and ignored. In addition to being specified by IP address, hosts may also be specified by their hostname. When the resolver is called to add a hostname to a table, all resulting IPv4 and IPv6 addresses are placed into the table. IP addresses can also be entered in a table by specifying a valid interface name or the self keyword, in which case all addresses assigned to the interface(s) are added to the table.
You can tune pf for various situations by using the set command:
When a packet matches a stateful connection, the seconds to live for the connection is updated to that of the proto.modifier that corresponds to the connection state. Each packet that matches this state resets the TTL. Tuning these values may improve the performance of the firewall, at the risk of dropping valid idle connections.
ICMP and UDP are handled in a fashion similar to TCP, but with a much more limited set of states:
Other protocols are handled similarly to UDP:
Timeout values can be reduced adaptively as the number of state table entries grows.
You can define these values both globally and for each rule. When used on a per-rule basis, the values relate to the number of states created by the rule, otherwise to the total number of states. For example:
set timeout tcp.first 120 set timeout tcp.established 86400 set timeout { adaptive.start 6000, adaptive.end 12000 } set limit states 10000
With 9000 state table entries, the timeout values are scaled to 50% (tcp.first 60, tcp.established 43200).
pfctl -s info
In this example, pf collects statistics on the interface named dc0:
set loginterface dc0
You can disable the log interface by using:
set loginterface none
set limit states 20000
sets the maximum number of entries in the memory pool used by state table entries (generated by keep state rules) to 20000. This command:
set limit frags 20000
sets the maximum number of entries in the memory pool used for fragment reassembly (generated by scrub rules) to 20000. This command:
set limit src-nodes 2000
sets the maximum number of entries in the memory pool used for tracking source IP addresses (generated by the sticky-address and source-track options) to 2000.
You can combine them:
set limit { states 20000, frags 20000, src-nodes 2000 }
For example:
set optimization aggressive
For example:
set block-policy return
For example:
set state-policy if-bound
There may be non-trivial and non-obvious implications to an out-of-order rule set. Consider carefully before disabling the order enforcement. |
set fingerprints "/etc/pf.os.devel"
Setting this option may leave a small period of time where the fingerprints referenced by the currently active rule set are inconsistent until the new rule set finishes loading.
set skip on lo0
Traffic normalization is used to sanitize packet content in such a way that there are no ambiguities in packet interpretation on the receiving side. The normalizer does IP fragment reassembly to prevent attacks that confuse intrusion detection systems by sending overlapping IP fragments. Packet normalization is invoked with the scrub directive, which has the following options:
Unfortunately some operating systems also generate their dont-fragment packets with a IP identification field of zero. Clearing the dont-fragment bit on packets with a zero IP ID may cause deleterious results if an upstream router later fragments the packet. Using the random-id modifier (see below) is recommended in combination with the no-df modifier to ensure unique IP identifiers.
For example:
scrub in on $ext_if all fragment reassemble
The no option prefixed to a scrub rule causes matching packets to remain unscrubbed, much in the same way as drop quick works in the packet filter (see below). This mechanism should be used when it is necessary to exclude specific packets from broader scrub rules.
Packets can be assigned to queues for the purpose of bandwidth control. At least two declarations are required to configure queues, and later any packet filtering rule can reference the defined queues by name. During the filtering component of pf.conf, the last referenced queue name is where any packets from pass rules will be queued, while for block rules it specifies where any resulting ICMP or TCP RST packets should be queued. The scheduler defines the algorithm used to decide which packets get delayed, dropped, or sent out immediately.
The currently supported schedulers are:
The cbq scheduler achieves both partitioning and sharing of link bandwidth by hierarchically structured classes. Each class has its own queue and is assigned its share of bandwidth. A child class can borrow bandwidth from its parent class, as long as excess bandwidth is available (see the option borrow, below).
The hfsc scheduler supports both link-sharing and guaranteed realtime services. It employs a service-curve-based QoS model, and its unique feature is an ability to decouple delay and bandwidth allocation.
The interfaces on which queueing should be activated are declared using the altq on declaration, which has the following keywords:
In the following example, the interface dc0 should queue up to 5 Mbit/s in four second-level queues using Class Based Queueing. Those four queues will be shown in a later example:
altq on dc0 cbq bandwidth 5Mb queue { std, http, mail, ssh }
Once interfaces are activated for queueing using the altq directive, you can define a sequence of queue directives. The name associated with a queue must match a queue defined in the altq directive (e.g. mail), or — except for the priq scheduler — in a parent queue declaration. You can use the following keywords:
The scheduler can get additional parameters with scheduler( parameters ). The parameters are as follows:
The CBQ scheduler supports an additional option:
The HFSC scheduler supports some additional options (sc is an acronym for service curve):
The format for service curve specifications is (m1, d, m2). The m2 variable controls the bandwidth assigned to the queue, while m1 and d are optional and can be used to control the initial bandwidth assignment. For the first d milliseconds, the queue gets the bandwidth given as m1, and afterward, the value given in m2.
Furthermore, with CBQ and HFSC, child queues can be specified as in an altq declaration, thus building a tree of queues using a part of their parent's bandwidth.
Packets can be assigned to queues based on filter rules by using the queue keyword. Normally only one queue is specified; when a second one is specified, it's used instead for packets that have a TOS of lowdelay and for TCP ACKs with no data payload.
To continue the previous example, the examples below specify the four referenced queues, plus a few child queues. The queues may then be referenced by filtering rules (see “Packet filtering,” below).
queue std bandwidth 10% cbq(default) queue http bandwidth 60% priority 2 cbq(borrow red) \ { employees, developers } queue developers bandwidth 75% cbq(borrow) queue employees bandwidth 15% queue mail bandwidth 10% priority 0 cbq(borrow ecn) queue ssh bandwidth 20% cbq(borrow) { ssh_interactive, ssh_bulk } queue ssh_interactive bandwidth 50% priority 7 cbq(borrow) queue ssh_bulk bandwidth 50% priority 0 cbq(borrow) block return out on dc0 inet all queue std pass out on dc0 inet proto tcp from $developerhosts to any port 80 \ keep state queue developers pass out on dc0 inet proto tcp from $employeehosts to any port 80 \ keep state queue employees pass out on dc0 inet proto tcp from any to any port 22 \ keep state queue(ssh_bulk, ssh_interactive) pass out on dc0 inet proto tcp from any to any port 25 \ keep state queue mail
Translation rules modify either the source or destination address of the packets associated with a stateful connection. A stateful connection is automatically created to track packets matching such a rule as long as they aren't blocked by the filtering section of pf.conf. The translation engine modifies the specified address and/or port in the packet, recalculates IP, TCP and UDP checksums as necessary, and passes it to the packet filter for evaluation.
Since translation occurs before filtering the filter engine will see packets as they look after any addresses and ports have been translated. Filter rules will therefore have to filter based on the translated address and port number. Packets that match a translation rule are automatically passed only if the pass modifier is given; otherwise they're still subject to block and pass rules.
The state entry created permits pf to keep track of the original address for traffic associated with that state and correctly direct return traffic for that connection.
Various types of translation are possible with pf:
rdr ... port 2000:2999 -> ... port 4000
redirects ports 2000 to 2999 (inclusive) to port 4000. This command:
rdr ... port 2000:2999 -> ... port 4000:*
redirects port 2000 to 4000, 2001 to 4001, ..., 2999 to 4999.
In addition to modifying the address, some translation rules may modify source or destination ports for TCP or UDP connections; implicitly in the case of nat rules and explicitly in the case of rdr rules. Port numbers are never translated with a binat rule.
For each packet processed by the translator, the translation rules are evaluated in sequential order, from first to last. The first matching rule decides what action is taken.
The no option prefixed to a translation rule causes packets to remain untranslated, much in the same way as drop quick works in the packet filter (see below). If no rule matches the packet, the packet is passed to the filter engine unmodified.
Translation rules apply only to packets that pass through the specified interface, and if no interface is specified, translation is applied to packets on all interfaces. For instance, redirecting port 80 on an external interface to an internal web server works only for connections originating from the outside. Connections to the address of the external interface from local hosts aren't redirected, since such packets don't actually pass through the external interface. Redirections can't reflect packets back through the interface they arrive on; they can only be redirected to hosts connected to different interfaces or to the firewall itself.
Note that redirecting external incoming connections to the loopback address, as in:
rdr on ne3 inet proto tcp to port 8025 -> 127.0.0.1 port 25
effectively allows an external host to connect to daemons bound solely to the loopback address, circumventing the traditional blocking of such connections on a real interface. Unless this effect is desired, any of the local non-loopback addresses should be used as redirection target instead, which allows external connections only to daemons bound to this address or not bound to any address.
See “Translation examples,” below.
The pf pseudo-device can block and pass packets based on attributes of their layer-3 (see IP and IPv6 in the Neutrino Library Reference) and layer-4 (see ICMP, ICMP6, TCP, and UDP) headers. In addition, packets may also be assigned to queues for the purpose of bandwidth control.
For each packet processed by the packet filter, the filter rules are evaluated in sequential order, from first to last. The last matching rule decides what action is taken.
You can use the following actions in the filter:
Options returning ICMP packets currently have no effect if pf operates on a bridge, as the code to support this feature hasn't yet been implemented.
If no rule matches the packet, the default action is to pass it. To block everything by default and pass only packets that match explicit rules, use:
block all
as the first filter rule.
See “Filter examples,” below.
The rule parameters specify the packets to which a rule applies. A packet always comes in on, or goes out through, one interface. Most parameters are optional. If a parameter is specified, the rule applies only to packets with matching attributes. Certain parameters can be expressed as lists, in which case pfctl generates all needed rule combinations.
Interface names can have modifiers appended:
Host names may also have the :0 option appended to restrict the name resolution to the first of each v4 and v6 address found.
Host-name resolution and interface-to-address translation are done when the rule set is loaded. When the address of an interface (or host name) changes (under DHCP or PPP, for instance), the rule set must be reloaded for the change to be reflected in io-pkt.
Surrounding the interface name (and optional modifiers) in parentheses changes this behavior: the rule is automatically updated whenever the interface changes its address. The rule set doesn't need to be reloaded. This is especially useful with nat.
You can specify ports either by number or by name. For example, port 80 can be specified as www. For a list of all port name-to-number mappings used by pfctl, see the file /etc/services.
You can use these operators to specify ports and ranges of ports:
Operator | Meaning |
---|---|
= | Equal to |
!= | Unequal to |
< | Less than |
<= | Less than or equal to |
> | Greater than |
>= | Greater than or equal to |
: | Range, including boundaries |
>< | Range, excluding boundaries |
<> | Except range |
The ><, <> and : operators are binary; they take two arguments. For instance:
This: | Means: |
---|---|
port 2000:2004 | All ports ≥ 2000 and ≤ 2004; hence ports 2000, 2001, 2002, 2003 and 2004 |
port 2000 >< 2004 | All ports > 2000 and < 2004; hence ports 2001, 2002 and 2003 |
port 2000 <> 2004 | All ports < 2000 or > 2004; hence ports 1-1999 and 2005-65535 |
The operating system of the source host can be specified in the case of TCP rules with the OS modifier. See “Operating system fingerprinting” for more information.
The host, port and OS specifications are optional, as in the following examples:
pass in all pass in from any to any pass in proto tcp from any port <= 1024 to any pass in proto tcp from any to any port 25 pass in proto tcp from 10.0.0.0/8 port > 1024 \ to ! 10.1.2.3 port != ssh pass in proto tcp from any os "OpenBSD" flags S/SA pass in proto tcp from route "DTAG"
All packets, both outgoing and incoming, of one connection are associated with the same user and group. Only TCP and UDP packets can be associated with users; for other protocols these parameters are ignored.
User and group refer to the effective (as opposed to the real) IDs, in case the socket is created by a setuid or setgid process. User and group IDs are stored when a socket is created; when a process creates a listening socket as root (for instance, by binding to a privileged port) and subsequently changes to another user ID (to drop privileges), the credentials will remain root.
You can specify user and group IDs as either numbers or names. The syntax is similar to the one for ports. The value unknown matches packets of forwarded connections. You can use unknown only with the operators = and !=. Other constructions, such as user >= unknown, are invalid. Forwarded packets with unknown user and group ID match only rules that explicitly compare against unknown with the operators = or !=. For instance, user >= 0 doesn't match forwarded packets. The following example allows only selected users to open outgoing connections:
block out proto { tcp, udp } all pass out proto { tcp, udp } all \ user { < 1000, dhartmei } keep state
This: | Means: |
---|---|
flags S/S | Flag SYN is set. The other flags are ignored. |
flags S/SA | Out of SYN and ACK, exactly SYN may be set. SYN, SYN+PSH and SYN+RST match, but SYN+ACK, ACK and ACK+RST don't. This is more restrictive than the previous example. |
flags /SFRA | If the first set isn't specified, it defaults to none. All of SYN, FIN, RST and ACK must be unset. |
You can use the following macros in labels:
Macro | Meaning |
---|---|
$if | The interface |
$srcaddr | The source IP address |
$dstaddr | The destination IP address |
$srcport | The source port specification |
$dstport | The destination port specification |
$proto | The protocol name |
$nr | The rule number |
For example:
ips = "{ 1.2.3.4, 1.2.3.5 }" pass in proto tcp from any to $ips \ port > 1023 label "$dstaddr:$dstport"
expands to:
pass in inet proto tcp from any to 1.2.3.4 \ port > 1023 label "1.2.3.4:>1023" pass in inet proto tcp from any to 1.2.3.5 \ port > 1023 label "1.2.3.5:>1023"
The macro expansion for the label directive occurs only when the configuration file is parsed, not during runtime.
pass in proto tcp to port 25 queue mail pass in proto tcp to port 22 queue(ssh_bulk, ssh_prio)
Tags are “sticky”, meaning that the packet is tagged even if the rule isn't the last matching rule. Further matching rules can replace the tag with a new one, but don't remove a previously applied tag. A packet is only ever assigned one tag at a time. Any pass rules that use the tag keyword must also use keep state, modulate state or synproxy state.
You can tag packets during nat, rdr, or binat rules in addition to filter rules. Tags take the same macros as labels (see above).
block in proto icmp probability 20%
If a packet matches a rule with a route option set, the packet filter routes the packet according to the type of route option. When such a rule creates state, the route option is also applied to all packets matching the same connection.
For nat and rdr rules, (as well as for the route-to, reply-to and dup-to rule options) for which there is a single redirection address that has a subnet mask smaller than 32 for IPv4 or 128 for IPv6 (more than one IP address), you can use a variety of different methods for assigning this address:
When more than one redirection address is specified, round-robin is the only permitted pool type.
Additionally, you can specify the sticky-address option to help ensure that multiple connections from the same source are mapped to the same redirection address. You can use this option with the random and round-robin pool options. Note that by default, these associations are destroyed as soon as there are no longer states that refer to them; in order to make the mappings last beyond the lifetime of the states, increase the global options with set timeout source-track. See “Stateful tracking options” for more ways to control the source tracking.
The pf packet filter is stateful, which means it can track the state of a connection. Instead of passing all traffic to port 25, for instance, it's possible to pass only the initial packet, and then begin to keep state. Subsequent traffic will flow because the filter is aware of the connection.
If a packet matches a pass ... keep state rule, the filter creates a state for this connection and automatically lets pass all subsequent packets of that connection.
Before any rules are evaluated, the filter checks whether the packet matches any state. If it does, the packet is passed without evaluation of any rules.
States are removed after the connection is closed or has timed out.
This has several advantages:
For instance:
block all pass out proto tcp from any to any flags S/SA keep state pass in proto tcp from any to any port 25 flags S/SA keep state
This rule set blocks everything by default. Only outgoing connections and incoming connections to port 25 are allowed. The initial packet of each connection has the SYN flag set, will be passed and creates state. All further packets of these connections are passed if they match a state.
By default, packets coming in and out of any interface can match a state, but it's also possible to change that behavior by assigning states to a single interface or a group of interfaces.
The default policy is specified by the state-policy global option, but you can adjust this on a per-rule basis by adding one of the if-bound, group-bound, or floating keywords to the keep state option. For example, if a rule is defined as:
pass out on ppp from any to 10.12/16 keep state (group-bound)
a state created on ppp0 would match packets an all PPP interfaces, but not packets flowing through fxp0 or any other interface.
Keeping rules floating is the more flexible option when the firewall is in a dynamic routing environment. However, this has some security implications, since a state created by one trusted network could allow potentially hostile packets coming in from other interfaces.
Specifying flags S/SA restricts state creation to the initial SYN packet of the TCP handshake. You can also be less restrictive, and allow state creation from intermediate (non-SYN) packets. This causes pf to synchronize to existing connections, for instance if you flush the state table.
For UDP, which is stateless by nature, keep state creates state as well. UDP packets are matched to states using only host addresses and ports.
ICMP messages fall into two categories: ICMP error messages, which always refer to a TCP or UDP packet, are matched against the referred to connection. If you keep state on a TCP connection, and an ICMP source quench message referring to this TCP connection arrives, it will be matched to the right state and get passed.
For ICMP queries, keep state creates an ICMP state, and pf knows how to match ICMP replies to states. For example:
pass out inet proto icmp all icmp-type echoreq keep state
allows echo requests (such as those created by ping) out, creates state, and matches incoming echo replies correctly to states.
The nat, binat, and rdr rules implicitly create state for connections. |
Much of the security derived from TCP is attributable to how well the initial sequence numbers (ISNs) are chosen. Some popular stack implementations choose very poor ISNs, and thus are normally susceptible to ISN prediction exploits. By applying a modulate state rule to a TCP connection, pf creates a high-quality random sequence number for each connection endpoint.
The modulate state directive implicitly keeps state on the rule and is applicable only to TCP connections. For instance:
block all pass out proto tcp from any to any modulate state pass in proto tcp from any to any port 25 flags S/SA modulate state
There are some caveats associated with state modulation:
By default, pf passes packets that are part of a TCP handshake between the endpoints. You can use the synproxy state option to cause pf itself to complete the handshake with the active endpoint, perform a handshake with the passive endpoint, and then forward packets between the endpoints.
No packets are sent to the passive endpoint before the active endpoint has completed the handshake, hence so-called SYN floods with spoofed source addresses will not reach the passive endpoint, as the sender can't complete the handshake.
The proxy is transparent to both endpoints, they each see a single connection from/to the other endpoint. The pf packet filter chooses random initial sequence numbers for both handshakes. Once the handshakes are completed, the sequence number modulators (see previous section) are used to translate further packets of the connection. Hence, synproxy state includes modulate state and keep state.
Rules with synproxy won't work if pf operates on a bridge.
Example:
pass in proto tcp from any to any port www flags S/SA synproxy state
All of keep state, modulate state and synproxy state support the following options:
You can specify multiple options, separated by commas:
pass in proto tcp from any to any \ port www flags S/SA keep state \ (max 100, source-track rule, max-src-nodes 75, \ max-src-states 3, tcp.established 60, tcp.closing 5)
If you specify the source-track keyword, the number of states per source IP is tracked:
You can set the following limits:
For stateful TCP connections, limits on established connections (those that have completed the TCP 3-way handshake) can also be enforced per source IP.
Because the 3-way handshake ensures that the source address isn't being spoofed, more aggressive action can be taken based on these limits. With the overload table state option, source IP addresses that hit either of the limits on established connections will be added to the named table. You can use this table in the rule set to block further activity from the offending host, redirect it to a tarpit process, or restrict its bandwidth.
The optional flush keyword kills all states created by the matching rule that originate from the host that exceeds these limits. The global modifier to the flush command kills all states originating from the offending host, regardless of which rule created the state.
For example, the following rules protect the web server against hosts making more than 100 connections in 10 seconds. Any host that connects faster than this rate will have its address added to the bad_hosts table and have all states originating from it flushed. Any new packets arriving from this host will be dropped unconditionally by the block rule.
block quick from bad_hosts pass in on $ext_if proto tcp to $webserver port www flags S/SA keep state \ (max-src-conn-rate 100/10, overload bad_hosts flush global)
Passive OS Fingerprinting is a mechanism to inspect nuances of a TCP con nection's initial SYN packet and guess at the host's operating system. Unfortunately these nuances are easily spoofed by an attacker so the fingerprint isn't useful in making security decisions. But the fingerprint is typically accurate enough to make policy decisions upon.
The fingerprints may be specified by operating system class, by version, or by subtype/patchlevel. The class of an operating system is typically the vendor or genre and would be OpenBSD for the pf firewall itself. The version of the oldest available OpenBSD release on the main ftp site would be 2.6 and the fingerprint would be written
"OpenBSD 2.6"
The subtype of an operating system is typically used to describe the patchlevel if that patch led to changes in the TCP stack behavior. In the case of OpenBSD, the only subtype is for a fingerprint that was normalized by the no-df scrub option and would be specified as:
"OpenBSD 3.3 no-df"
Fingerprints for most popular operating systems are provided by /etc/pf.os. Once pf is running, you can get a complete list of known operating system finger listed by running:
# pfctl -so
Filter rules can enforce policy at any level of operating system specification assuming a fingerprint is present. Policy could limit traffic to approved operating systems or even ban traffic from hosts that aren't at the latest service pack.
You can also use the unknown class as the fingerprint that matches packets for which no operating system fingerprint is known.
Examples:
pass out proto tcp from any os OpenBSD keep state block out proto tcp from any os Doors block out proto tcp from any os "Doors PT" block out proto tcp from any os "Doors PT SP3" block out from any os "unknown" pass on lo0 proto tcp from any os "OpenBSD 3.3 lo0" keep state
Operating system fingerprinting is limited only to the TCP SYN packet. This means that it doesn't work on other protocols and doesn't match a currently established connection.
Operating system fingerprints are occasionally wrong.
There are several problems:
|
Spoofing is the faking of IP addresses, typically for malicious purposes. The antispoof directive expands to a set of filter rules that block all traffic with a source IP from the network(s) directly connected to the specified interface(s) from entering the system through any other interface. For example, the line:
antispoof for lo0
expands to:
block drop in on ! lo0 inet from 127.0.0.1/8 to any block drop in on ! lo0 inet6 from ::1 to any
For non-loopback interfaces, there are additional rules to block incoming packets with a source IP address identical to the interface's IP(s). For example, assuming the interface wi0 has an IP address of 10.0.0.1 and a netmask of 255.255.255.0, the line:
antispoof for wi0 inet
expands to:
block drop in on ! wi0 inet from 10.0.0.0/24 to any block drop in inet from 10.0.0.1 to any
Rules created by the antispoof directive interfere with packets sent over loopback interfaces to local addresses. You should pass these explicitly. |
The size of IP datagrams (packets) can be significantly larger than the maximum transmission unit (MTU) of the network. In cases when it is necessary or more efficient to send such large packets, the large packet will be fragmented into many smaller packets that will each fit onto the wire. Unfortunately for a firewalling device, only the first logical fragment will contain the necessary header information for the subprotocol that allows pf to filter on things such as TCP ports or to perform NAT.
Besides the use of scrub rules as described in “Traffic normalization” above, there are three options for handling fragments in the packet filter.
One alternative is to filter individual fragments with filter rules. If no scrub rule applies to a fragment, it's passed to the filter. Filter rules with matching IP header parameters decide whether the fragment is passed or blocked, in the same way as complete packets are filtered. Without reassembly, fragments can be filtered based only on IP header fields (source/destination address, protocol), since subprotocol header fields aren't available (TCP/UDP port numbers, ICMP code/type). You can use the fragment option to restrict filter rules to apply only to fragments, but not complete packets. Filter rules without the fragment option still apply to fragments, if they only specify IP header fields. For instance, the rule:
pass in proto tcp from any to any port 80
never applies to a fragment, even if the fragment is part of a TCP packet with destination port 80, because without reassembly this information isn't available for each fragment. This also means that fragments can't create new or match existing state table entries, which makes stateful filtering and address translation (NAT, redirection) for fragments impossible.
It's also possible to reassemble only certain fragments by specifying source or destination addresses or protocols as parameters in scrub rules.
In most cases, the benefits of reassembly outweigh the additional memory cost, and it's recommended you use scrub rules to reassemble all fragments via the fragment reassemble modifier.
You can limit the memory allocated for fragment caching by using pfctl. Once this limit is reached, fragments that would have to be cached are dropped until other entries time out. You can also adjust the timeout value.
Currently, only IPv4 fragments are supported, and IPv6 fragments are blocked unconditionally.
Besides the main rule set, pfctl can load rule sets into anchor attachment points. An anchor is a container that can hold rules, address tables, and other anchors.
An anchor has a name that specifies the path where you can use pfctl to access the anchor to perform operations on it, such as attaching child anchors to it or loading rules into it. Anchors may be nested, with components separated by slashes (/), similar to how file system hierarchies are laid out. The main rule set is actually the default anchor, so filter and translation rules, for example, may also be contained in any anchor.
An anchor can reference another anchor attachment point using the following kinds of rules:
When evaluation of the main rule set reaches an anchor rule, pf evaluates all rules specified in that anchor.
Matching filter and translation rules in anchors with the quick option are final and abort the evaluation of the rules in other anchors and the main rule set.
Anchor rules are evaluated relative to the anchor in which they are contained. For example, all anchor rules specified in the main rule set reference anchor attachment points underneath the main rule set, and anchor rules specified in a file loaded from a load anchor rule are attached under that anchor point.
Rules may be contained in anchor attachment points that don't contain any rules when the main rule set is loaded, and later such anchors can be manipulated through pfctl without reloading the main rule set or other anchors. For example:
ext_if = "kue0" block on $ext_if all anchor spam pass out on $ext_if all keep state pass in on $ext_if proto tcp from any \ to $ext_if port smtp keep state
blocks all packets on the external interface by default, then evaluates all rules in the anchor named spam, and finally passes all outgoing connections and incoming connections to port 25.
The following command loads a single rule into the anchor, which blocks all packets from a specific address:
# echo "block in quick from 1.2.3.4 to any" | \ pfctl -a spam -f -
The anchor can also be populated by adding a load anchor rule after the anchor rule:
anchor spam load anchor spam from "/etc/pf-spam.conf"
When pfctl loads pf.conf, it also loads all the rules from the file /etc/pf-spam.conf into the anchor.
Optionally, anchor rules can specify the parameter's direction, interface, address family, protocol and source/destination address/port using the same syntax as filter rules. When parameters are used, the anchor rule is evaluated only for matching packets. This allows conditional evaluation of anchors, like:
block on $ext_if all anchor spam proto tcp from any to any port smtp pass out on $ext_if all keep state pass in on $ext_if proto tcp from any to $ext_if port smtp keep state
The rules inside the anchor spam are evaluated only for TCP packets with destination port 25. Hence:
# echo "block in quick from 1.2.3.4 to any" | \ pfctl -a spam -f -
blocks connections only from 1.2.3.4 to port 25.
Anchors may end with the asterisk (*) character, which signifies that all anchors attached at that point should be evaluated in the alphabetical ordering of their anchor name. For example:
anchor "spam/*"
evaluates each rule in each anchor attached to the spam anchor. Note that it evaluates only anchors that are directly attached to the spam anchor, and doesn't descend to evaluate anchors recursively.
Since anchors are evaluated relative to the anchor in which they are contained, there's a mechanism for accessing the parent and ancestor anchors of a given anchor. Similar to file system path name resolution, if the sequence .. appears as an anchor path component, the parent anchor of the current anchor in the path evaluation at that point becomes the new current anchor. As an example, consider the following:
# echo ' anchor "spam/allowed" ' | pfctl -f - # echo -e ' anchor "../banned" \n pass' | \ pfctl -a spam/allowed -f -
Evaluation of the main rule set leads into the spam/allowed anchor, which evaluates the rules in the spam/banned anchor, if any, before finally evaluating the pass rule.
Since the parser specification for anchor names is a string, any reference to an anchor name containing slash (/) characters requires double quote (") characters around the anchor name.
This example maps incoming requests on port 80 to port 8080, on which a daemon is running (because, for example, it isn't run as root, and therefore lacks permission to bind to port 80):
# use a macro for the interface name, so it can be changed easily ext_if = "ne3" # map daemon on 8080 to appear to be on 80 rdr on $ext_if proto tcp from any to any port 80 -> 127.0.0.1 port 8080
If the pass modifier is given, packets matching the translation rule are passed without inspecting the filter rules:
rdr pass on $ext_if proto tcp from any to any port 80 -> 127.0.0.1 \ port 8080
In the example below, vlan12 is configured as 192.168.168.1; the machine translates all packets coming from 192.168.168.0/24 to 204.92.77.111 when they're going out any interface except vlan12. This has the net effect of making traffic from the 192.168.168.0/24 network appear as though it is the Internet routable address 204.92.77.111 to nodes behind any interface on the router except for the nodes on vlan12. (Thus, 192.168.168.1 can talk to the 192.168.168.0/24 nodes.)
nat on ! vlan12 from 192.168.168.0/24 to any -> 204.92.77.111
In the example below, the machine sits between a fake internal 144.19.74.* network, and a routable external IP of 204.92.77.100. The no nat rule excludes protocol AH from being translated:
# NO NAT no nat on $ext_if proto ah from 144.19.74.0/24 to any nat on $ext_if from 144.19.74.0/24 to any -> 204.92.77.100
In the example below, packets bound for one specific server, as well as those generated by the system administrators aren't proxied; all other connections are:
# NO RDR no rdr on $int_if proto { tcp, udp } from any to $server port 80 no rdr on $int_if proto { tcp, udp } from $sysadmins to any port 80 rdr on $int_if proto { tcp, udp } from any to any port 80 -> 127.0.0.1 \ port 80
This longer example uses both a NAT and a redirection. The external interface has the address 157.161.48.183. On the internal interface, we are running ftp-proxy, listening for outbound ftp sessions captured to port 8021:
# NAT # Translate outgoing packets' source addresses (any protocol). # In this case, any address but the gateway's external address is mapped. nat on $ext_if inet from ! ($ext_if) to any -> ($ext_if) # NAT PROXYING # Map outgoing packets' source port to an assigned proxy port instead of # an arbitrary port. # In this case, proxy outgoing isakmp with port 500 on the gateway. nat on $ext_if inet proto udp from any port = isakmp to any -> ($ext_if) \ port 500 # BINAT # Translate outgoing packets' source address (any protocol). # Translate incoming packets' destination address to an internal machine # (bidirectional). binat on $ext_if from 10.1.2.150 to any -> $ext_if # RDR # Translate incoming packets' destination addresses. # As an example, redirect a TCP and UDP port to an internal machine. rdr on $ext_if inet proto tcp from any to ($ext_if) port 8080 \ -> 10.1.2.151 port 22 rdr on $ext_if inet proto udp from any to ($ext_if) port 8080 \ -> 10.1.2.151 port 53 # RDR # Translate outgoing ftp control connections to send them to localhost # for proxying with ftp-proxy(8) running on port 8021. rdr on $int_if proto tcp from any to any port 21 -> 127.0.0.1 port 8021
In this example, a NAT gateway is set up to translate internal addresses using a pool of public addresses (192.0.2.16/28) and to redirect incoming web server connections to a group of web servers on the internal network:
# NAT LOAD BALANCE # Translate outgoing packets' source addresses using an address pool. # A given source address is always translated to the same pool address by # using the source-hash keyword. nat on $ext_if inet from any to any -> 192.0.2.16/28 source-hash # RDR ROUND ROBIN # Translate incoming web server connections to a group of web servers on # the internal network. rdr on $ext_if proto tcp from any to any port 80 \ -> { 10.1.2.155, 10.1.2.160, 10.1.2.161 } round-robin
# The external interface is kue0 # (157.161.48.183, the only routable address) # and the private network is 10.0.0.0/8, for which we are doing NAT. # use a macro for the interface name, so it can be changed easily ext_if = "kue0" # normalize all incoming traffic scrub in on $ext_if all fragment reassemble # block and log everything by default block return log on $ext_if all # block anything coming from source we have no back routes for block in from no-route to any # block and log outgoing packets that don't have our address as source, # they are either spoofed or something is misconfigured (NAT disabled, # for instance), we want to be nice and don't send out garbage. block out log quick on $ext_if from ! 157.161.48.183 to any # silently drop broadcasts (cable modem noise) block in quick on $ext_if from any to 255.255.255.255 # block and log incoming packets from reserved address space and invalid # addresses, they are either spoofed or misconfigured, we can't reply to # them anyway (hence, no return-rst). block in log quick on $ext_if from { 10.0.0.0/8, 172.16.0.0/12, \ 192.168.0.0/16, 255.255.255.255/32 } to any # ICMP # pass out/in certain ICMP queries and keep state (ping) # state matching is done on host addresses and ICMP ID (not type/code), # so replies (like 0/0 for 8/0) will match queries # ICMP error messages (which always refer to a TCP/UDP packet) are # handled by the TCP/UDP states pass on $ext_if inet proto icmp all icmp-type 8 code 0 keep state # UDP # pass out all UDP connections and keep state pass out on $ext_if proto udp all keep state # pass in certain UDP connections and keep state (DNS) pass in on $ext_if proto udp from any to any port domain keep state # TCP # pass out all TCP connections and modulate state pass out on $ext_if proto tcp all modulate state # pass in certain TCP connections and keep state (SSH, SMTP, DNS, IDENT) pass in on $ext_if proto tcp from any to any port { ssh, smtp, domain, \ auth } flags S/SA keep state # pass in data mode connections for ftp-proxy running on this host. # (see ftp-proxy(8) for details) pass in on $ext_if proto tcp from any to 157.161.48.183 port >= 49152 \ flags S/SA keep state # Don't allow Windows 9x SMTP connections since they are typically # a viral worm. Alternately we could limit these OSes to 1 connection each. block in on $ext_if proto tcp from any os {"Windows 95", "Windows 98"} \ to any port smtp # Packet Tagging # three interfaces: $int_if, $ext_if, and $wifi_if (wireless). NAT is # being done on $ext_if for all outgoing packets. tag packets in on # $int_if and pass those tagged packets out on $ext_if. all other # outgoing packets (i.e., packets from the wireless network) are only # permitted to access port 80. pass in on $int_if from any to any tag INTNET keep state pass in on $wifi_if from any to any keep state block out on $ext_if from any to any pass out quick on $ext_if tagged INTNET keep state pass out on $ext_if proto tcp from any to any port 80 keep state
The syntax for pf.conf in BNF is as follows:
line = ( option | pf-rule | nat-rule | binat-rule | rdr-rule | antispoof-rule | altq-rule | queue-rule | anchor-rule | trans-anchors | load-anchors | table-rule ) option = "set" ( [ "timeout" ( timeout | "{" timeout-list "}" ) ] | [ "optimization" [ "default" | "normal" | "high-latency" | "satellite" | "aggressive" | "conservative" ] ] [ "limit" ( limit-item | "{" limit-list "}" ) ] | [ "loginterface" ( interface-name | "none" ) ] | [ "block-policy" ( "drop" | "return" ) ] | [ "state-policy" ( "if-bound" | "group-bound" | "floating" ) ] [ "require-order" ( "yes" | "no" ) ] [ "fingerprints" filename ] | [ "debug" ( "none" | "urgent" | "misc" | "loud" ) ] ) pf-rule = action [ ( "in" | "out" ) ] [ "log" | "log-all" ] [ "quick" ] [ "on" ifspec ] [ route ] [ af ] [ protospec ] hosts [ filteropt-list ] filteropt-list = filteropt-list filteropt | filteropt filteropt = user | flags | icmp-type | icmp6-type | tos | ( "keep" | "modulate" | "synproxy" ) "state" [ "(" state-opts ")" ] | "fragment" | "no-df" | "min-ttl" number | "max-mss" number | "random-id" | "reassemble tcp" | fragmentation | "allow-opts" | "label" string | "tag" string | [ ! ] "tagged" string "queue" ( string | "(" string [ [ "," ] string ] ")" ) | "probability" number"%" nat-rule = [ "no" ] "nat" [ "pass" ] [ "on" ifspec ] [ af ] [ protospec ] hosts [ "tag" string ] [ "tagged" string ] [ "->" ( redirhost | "{" redirhost-list "}" ) [ portspec ] [ pooltype ] [ "static-port" ] ] binat-rule = [ "no" ] "binat" [ "pass" ] [ "on" interface-name ] [ af ] [ "proto" ( proto-name | proto-number ) ] "from" address [ "/" mask-bits ] "to" ipspec [ "tag" string ] [ "tagged" string ] [ "->" address [ "/" mask-bits ] ] rdr-rule = [ "no" ] "rdr" [ "pass" ] [ "on" ifspec ] [ af ] [ protospec ] hosts [ "tag" string ] [ "tagged" string ] [ "->" ( redirhost | "{" redirhost-list "}" ) [ portspec ] [ pooltype ] ] antispoof-rule = "antispoof" [ "log" ] [ "quick" ] "for" ( interface-name | "{" interface-list "}" ) [ af ] [ "label" string ] table-rule = "table" "<" string ">" [ tableopts-list ] tableopts-list = tableopts-list tableopts | tableopts tableopts = "persist" | "const" | "file" string | "{" [ tableaddr-list ] "}" tableaddr-list = tableaddr-list [ "," ] tableaddr-spec | tableaddr-spec tableaddr-spec = [ "!" ] tableaddr [ "/" mask-bits ] tableaddr = hostname | ipv4-dotted-quad | ipv6-coloned-hex | interface-name | "self" altq-rule = "altq on" interface-name queueopts-list "queue" subqueue queue-rule = "queue" string [ "on" interface-name ] queueopts-list subqueue anchor-rule = "anchor" string [ ( "in" | "out" ) ] [ "on" ifspec ] [ af ] [ "proto" ] [ protospec ] [ hosts ] trans-anchors = ( "nat-anchor" | "rdr-anchor" | "binat-anchor" ) string [ "on" ifspec ] [ af ] [ "proto" ] [ protospec ] [ hosts ] load-anchor = "load anchor" string "from" filename queueopts-list = queueopts-list queueopts | queueopts queueopts = [ "bandwidth" bandwidth-spec ] | [ "qlimit" number ] | [ "tbrsize" number ] | [ "priority" number ] | [ schedulers ] schedulers = ( cbq-def | priq-def | hfsc-def ) bandwidth-spec = "number" ( "b" | "Kb" | "Mb" | "Gb" | "%" ) action = "pass" | "block" [ return ] | [ "no" ] "scrub" return = "drop" | "return" | "return-rst" [ "( ttl" number ")" ] | "return-icmp" [ "(" icmpcode [ [ "," ] icmp6code ] ")" ] | "return-icmp6" [ "(" icmp6code ")" ] icmpcode = ( icmp-code-name | icmp-code-number ) icmp6code = ( icmp6-code-name | icmp6-code-number ) ifspec = ( [ "!" ] interface-name ) | "{" interface-list "}" interface-list = [ "!" ] interface-name [ [ "," ] interface-list ] route = "fastroute" | ( "route-to" | "reply-to" | "dup-to" ) ( routehost | "{" routehost-list "}" ) [ pooltype ] af = "inet" | "inet6" protospec = "proto" ( proto-name | proto-number | "{" proto-list "}" ) proto-list = ( proto-name | proto-number ) [ [ "," ] proto-list ] hosts = "all" | "from" ( "any" | "no-route" | "self" | host | "{" host-list "}" | "route" string ) [ port ] [ os ] "to" ( "any" | "no-route" | "self" | host | "{" host-list "}" | "route" string ) [ port ] ipspec = "any" | host | "{" host-list "}" host = [ "!" ] ( address [ "/" mask-bits ] | "<" string ">" ) redirhost = address [ "/" mask-bits ] routehost = ( interface-name [ address [ "/" mask-bits ] ] ) address = ( interface-name | "(" interface-name ")" | hostname | ipv4-dotted-quad | ipv6-coloned-hex ) host-list = host [ [ "," ] host-list ] redirhost-list = redirhost [ [ "," ] redirhost-list ] routehost-list = routehost [ [ "," ] routehost-list ] port = "port" ( unary-op | binary-op | "{" op-list "}" ) portspec = "port" ( number | name ) [ ":" ( "*" | number | name ) ] os = "os" ( os-name | "{" os-list "}" ) user = "user" ( unary-op | binary-op | "{" op-list "}" ) unary-op = [ "=" | "!=" | "<" | "<=" | ">" | ">=" ] ( name | number ) binary-op = number ( "<>" | "><" | ":" ) number op-list = ( unary-op | binary-op ) [ [ "," ] op-list ] os-name = operating-system-name os-list = os-name [ [ "," ] os-list ] flags = "flags" [ flag-set ] "/" flag-set flag-set = [ "F" ] [ "S" ] [ "R" ] [ "P" ] [ "A" ] [ "U" ] [ "E" ] [ "W" ] icmp-type = "icmp-type" ( icmp-type-code | "{" icmp-list "}" ) icmp6-type = "icmp6-type" ( icmp-type-code | "{" icmp-list "}" ) icmp-type-code = ( icmp-type-name | icmp-type-number ) [ "code" ( icmp-code-name | icmp-code-number ) ] icmp-list = icmp-type-code [ [ "," ] icmp-list ] tos = "tos" ( "lowdelay" | "throughput" | "reliability" | [ "0x" ] number ) state-opts = state-opt [ [ "," ] state-opts ] state-opt = ( "max" number | timeout | "source-track" [ ( "rule" | "global" ) ] | "max-src-nodes" number | "max-src-states" number | "max-src-conn" number | "max-src-conn-rate" number "/" number | "overload" "<" string ">" [ "flush" ] | "if-bound" | "group-bound" | "floating" ) fragmentation = [ "fragment reassemble" | "fragment crop" | "fragment drop-ovl" ] timeout-list = timeout [ [ "," ] timeout-list ] timeout = ( "tcp.first" | "tcp.opening" | "tcp.established" | "tcp.closing" | "tcp.finwait" | "tcp.closed" | "udp.first" | "udp.single" | "udp.multiple" | "icmp.first" | "icmp.error" | "other.first" | "other.single" | "other.multiple" | "frag" | "interval" | "src.track" | "adaptive.start" | "adaptive.end" ) number limit-list = limit-item [ [ "," ] limit-list ] limit-item = ( "states" | "frags" | "src-nodes" ) number pooltype = ( "bitmask" | "random" | "source-hash" [ ( hex-key | string-key ) ] | "round-robin" ) [ sticky-address ] subqueue = string | "{" queue-list "}" queue-list = string [ [ "," ] string ] cbq-def = "cbq" [ "(" cbq-opt [ [ "," ] cbq-opt ] ")" ] priq-def = "priq" [ "(" priq-opt [ [ "," ] priq-opt ] ")" ] hfsc-def = "hfsc" [ "(" hfsc-opt [ [ "," ] hfsc-opt ] ")" ] cbq-opt = ( "default" | "borrow" | "red" | "ecn" | "rio" ) priq-opt = ( "default" | "red" | "ecn" | "rio" ) hfsc-opt = ( "default" | "red" | "ecn" | "rio" | linkshare-sc | realtime-sc | upperlimit-sc ) linkshare-sc = "linkshare" sc-spec realtime-sc = "realtime" sc-spec upperlimit-sc = "upperlimit" sc-spec sc-spec = ( bandwidth-spec | "(" bandwidth-spec number bandwidth-spec ")" )
/etc/hosts, pf, /etc/pf.os, pfctl, /etc/protocols, route, /etc/services
ICMP, ICMP6, IP, IP6, ROUTE, TCP, UDP in the Neutrino Library Reference
bridge, ftp-proxy, pcap, pflog, pflogd in the NetBSD documentation at http://www.netbsd.org/docs/