NAME
ferm - a firewall rule parser for linux
SYNOPSIS
ferm options inputfile
DESCRIPTION
ferm is a frontend for iptables. It reads the rules from a structured
configuration file and calls iptables(8) to insert them into the
running kernel.
ferm’s goal is to make firewall rules easy to write and easy to read.
It tries to reduce the tedious task of writing down rules, thus
enabling the firewall administrator to spend more time on developing
good rules than the proper implementation of the rule.
To achieve this, ferm uses a simple but powerful configuration
language, which allows variables, functions, arrays, blocks. It also
allows you to include other files, allowing you to create libraries of
commonly used structures and functions.
ferm, pronounced "firm", stands for "For Easy Rule Making".
CAUTION
This manual page does not indend to teach you how firewalling works and
how to write good rules. There is already enough documentation on this
topic.
INTRODUCTION
Let’s start with a simple example:
chain INPUT {
proto tcp ACCEPT;
}
This will add a rule to the predefined input chain, matching and
accepting all tcp packets. Ok, let’s make it more complicated:
chain (INPUT OUTPUT) {
proto (udp tcp) ACCEPT;
}
This will insert 4 rules, namely 2 in chain input, and 2 in chain
output, matching and accepting both udp and tcp packets. Normally you
would type this:
iptables -A INPUT -p tcp -j ACCEPT
iptables -A OUTPUT -p tcp -j ACCEPT
iptables -A INPUT -p udp -j ACCEPT
iptables -A OUTPUT -p udp -j ACCEPT
Note how much less typing we need to do? :-)
Basically, this is all there is to it, although you can make it quite
more complex. Something to look at:
chain INPUT {
policy ACCEPT;
daddr 10.0.0.0/8 proto tcp dport ! ftp jump mychain sport :1023 TOS 4 settos 8 mark 2;
daddr 10.0.0.0/8 proto tcp dport ftp REJECT;
}
My point here is, that *you* need to make nice rules, keep them
readable to you and others, and not make it into a mess.
It would aid the reader if the resulting firewall rules were placed
here for reference. Also, you could include the nested version with
better readability.
Try using comments to show what you are doing:
# this line enables transparent http-proxying for the internal network:
proto tcp if eth0 daddr ! 192.168.0.0/255.255.255.0
dport http REDIRECT to-ports 3128;
You will be thankful for it later!
chain INPUT {
policy ACCEPT;
interface (eth0 ppp0) {
# deny access to notorius hackers, return here if no match
# was found to resume normal firewalling
jump badguys;
protocol tcp jump fw_tcp;
protocol udp jump fw_udp;
}
}
The more you nest, the better it looks. Make sure the order you specify
is correct, you would not want to do this:
chain FORWARD {
proto ! udp DROP;
proto tcp dport ftp ACCEPT;
}
because the second rule will never match. Best way is to specify first
everyting that is allowed, and then deny everything else. Look at the
examples for more good snapshots. Most people do something like this:
proto tcp {
dport (
ssh http ftp
) ACCEPT;
dport 1024:65535 ! syn ACCEPT;
DROP;
}
STRUCTURE OF A FIREWALL FILE
The structure of a proper firewall file looks like simplified C-code.
Only a few syntactic characters are used in ferm- configuration files.
Besides these special caracters, ferm uses ’keys’ and ’values’, think
of them as options and parameters, or as variables and values,
whatever.
With these words, you define the characteristics of your firewall.
Every firewall consists of two things: First, look if network traffic
matches certain conditions, and second, what to do with that traffic.
You may specify conditions that are valid for the kernel interface
program you are using, probably iptables(8). For instance, in iptables,
when you are trying to match tcp packets, you would say:
iptables --protocol tcp
In ferm, this will become:
protocol tcp;
Just typing this in ferm doesn’t do anything, you need to tell ferm
(actually, you need to tell iptables(8) and the kernel) what to do with
any traffic that matches this condition:
iptables --protocol tcp -j ACCEPT
Or, translated to ferm:
protocol tcp ACCEPT;
The ; character is at the end of every ferm rule. Ferm ignores line
breaks, meaning the above example is identical to the following:
protocol tcp
ACCEPT;
Here’s a list of the special characters:
; This character finalizes a rule.
Separated by semicolons, you may write multiple rules in one
line, although this decreases readability:
protocol tcp ACCEPT; protocol udp DROP;
{} The nesting symbol defines a ’block’ of rules.
The curly brackets contain any number of nested rules. All
matches before the block are carried forward to these.
The closing curly bracket finalizes the rule set. You should
not write a ’;’ after that, because that would be an empty
rule.
Example:
chain INPUT proto icmp {
icmp-type echo-request ACCEPT;
DROP;
}
This block shows two rules inside a block, which will both be
merged with anything in front of it, so you will get two rules:
iptables -A INPUT -p icmp --icmp-type echo-request -j ACCEPT
iptables -A INPUT -p icmp -j DROP
There can be multiple nesting levels:
chain INPUT {
proto icmp {
icmp-type echo-request ACCEPT;
DROP;
}
daddr 172.16.0.0/12 REJECT;
}
Note that the ’REJECT’ rule is not affected by ’proto icmp’,
although there is no ’;’ after the closing curly brace.
Translated to iptables:
iptables -A INPUT -p icmp --icmp-type echo-request -j ACCEPT
iptables -A INPUT -p icmp -j DROP
iptables -A INPUT -d 172.16.0.0/12 -j REJECT
$ Variable expansion. Replaces ’$FOO’ by the value of the
variable. See the section VARIABLES for details.
& Function call. See the section FUNCTIONS for details.
() The array symbol. Using the parentheses, you can define a
’list’ of values that should be applied for the key to the left
of it.
Example:
protocol ( tcp udp icmp )
this will result in three rules:
... -p tcp ...
... -p udp ...
... -p icmp ...
Only values can be ’listed’, so you cannot do something like
this:
proto tcp ( ACCEPT LOG );
but you can do this:
chain (INPUT OUTPUT FORWARD) proto (icmp udp tcp) DROP;
(which will result in nine rules!)
Values are separated by spaces. The array symbol is both left-
and right-associative, in contrast with the nesting block,
which is left-associative only.
" # " The comment symbol. Anything that follows this symbol up to the
end of line is ignored.
"`command`"
Execute the command in a shell, and insert the process output.
See the section backticks for details.
'string'
Quote a string which may contain whitespaces, the dollar sign
etc.
LOG log-prefix ' hey, this is my log prefix!';
"string"
Quote a string (see above), but variable references with a
dollar sign are evaluated:
DNAT to "$myhost:$myport";
Keywords
In the previous section, we already introduced some basic keywords like
"chain", "protocol" and "ACCEPT". Let’s explore their nature.
There are three kinds of keywords:
· location keywords define where a rule will be created. Example:
"table", "chain".
· match keywords perform a test on all passing packets. The
current rule is without effect if one (or more) of the matches
does not pass. Example: "proto", "daddr".
Most matches are followed by a parameter: "proto tcp", "daddr
172.16.0.0/12".
· target keywords state what to do with a packet. Example:
"ACCEPT", "REJECT", "jump".
Some targets define more keywords to specify details: "REJECT
reject-with icmp-net-unreachable".
Every rule consists of a location and a target, plus any number of
matches:
table filter # location
proto tcp dport (http https) # match
ACCEPT; # target
Strictly speaking, there is a fourth kind: ferm keywords (which control
ferm’s internal behaviour), but they will be explained later.
Parameters
Many keywords take parameters. These can be specified as literals,
variable references or lists (arrays):
proto udp
saddr %TRUSTED_HOSTS;
proto tcp dport (http https ssh);
LOG log-prefix "funky wardriver alert: ";
Some of them can be negated (lists cannot be negated):
proto !esp;
proto udp dport !domain;
Keywords which take no parameters are negated by a prefixed ’!’:
proto tcp !syn;
Read iptables(8) to see where the ! can be used.
BASIC KEYWORDS
Location keywords
domain [ip|ip6]
Set the domain. "ip" is default and means "IPv4" (iptables).
"ip6" is for IPv6 support, using "ip6tables".
table [filter|nat|mangle]
Specifies which netfilter table this rule will be inserted to:
"filter" (default), "nat" or "mangle".
chain [chain-name]
Specifies the netfilter chain (within the current table) this
rule will be inserted to. Common predefined chain names are
"INPUT", "OUTPUT", "FORWARD", "PREROUTING", "POSTROUTING",
depending on the table. See the netfilter documentation for
details.
If you specify a non-existing chain here, ferm will add the
rule to a custom chain with that name.
policy [ACCEPT|DROP|..]
Specifies the default policy for the current chain (built-in
only). Can be one of the built-in targets (ACCEPT, DROP,
REJECT, ...). A packet that matches no rules in a chain will be
treated as specified by the policy.
To avoid ambiguity, always specify the policies of all
predefined chains explicitly.
@subchain ["CHAIN-NAME"] { ... }
Works like the normal block operators (i.e. without the
@subchain keyword), except that ferm moves rules within the
curly braces into a new custom chain. The name for this chain
is chosen automatically by ferm.
In many cases, this is faster than just a block, because the
kernel may skip a huge block of rules when a precondition is
false. Imagine the following example:
table filter chain INPUT {
saddr (1.2.3.4 2.3.4.5 3.4.5.6 4.5.6.7 5.6.7.8) {
proto tcp dport (http https ssh) ACCEPT;
proto udp dport domain ACCEPT;
}
}
This generates 20 rules. When a packet arrives which does not
pass the saddr match, it nonetheless checks all 20 rules. With
@subchain, this check is done once, resulting in faster network
filtering and less CPU load:
table filter chain INPUT {
saddr (1.2.3.4 2.3.4.5 3.4.5.6 4.5.6.7 5.6.7.8) @subchain {
proto tcp dport (http https ssh) ACCEPT;
proto udp dport domain ACCEPT;
}
}
Optionally, you may define the name of the sub chain:
saddr (1.2.3.4 2.3.4.5 3.4.5.6) @subchain "foobar" {
proto tcp dport (http https ssh) ACCEPT;
proto udp dport domain ACCEPT;
}
You can achieve the same by explicitly declaring a custom
chain, but you may feel that using @subchain requires less
typing.
Basic iptables match keywords
interface [interface-name]
Define the interface name, your outside network card, like
eth0, or dialup like ppp1, or whatever device you want to match
for passing packets. It is equivalent to the "-i" switch in
iptables(8).
outerface [interface-name]
Same as interface, only for matching the outgoing interface for
a packet, as in iptables(8).
protocol [protocol-name|protocol-number]
Currently supported by the kernel are tcp, udp and icmp, or
their respective numbers.
saddr|daddr [address-spec]
Matches on packets originating from the specified address
(saddr) or targeted at the address (daddr).
Examples:
saddr 192.168/8 ACCEPT; # (identical to the next one:)
saddr 192.168.0.0/255.255.255.0 ACCEPT;
daddr my.domain.com ACCEPT;
fragment
Specify that only fragmented IP packets should be matched.
When packets are larger that the maximum packet size your
system can handle (called Maximum Transmission Unit or MTU)
they will be chopped into bits and sent one by one as single
packets. See ifconfig(8) if you want to find the MTU for your
system (the default is usually 1500 bytes).
Fragments are frequently used in DOS attacks, because there is
no way of finding out the origin of a fragment packet.
sport|dport [port-spec]
Matches on packets on the specified TCP or UDP port. "sport"
matches the source port, and dport matches the destination
port.
This match can be used only after you specified "protocol tcp"
or "protocol udp", because only these two protocols actually
have ports.
And some examples of valid ports/ranges:
dport 80 ACCEPT;
dport http ACCEPT;
dport ssh:http ACCEPT;
dport 0:1023 ACCEPT; # equivalent to :1023
dport 1023:65535 ACCEPT;
syn Specify that the SYN flag in a tcp package should be matched,
which are used to build new tcp connections. You can identify
incoming connections with this, and decide wether you want to
allow it or not. Packets that do not have this flag are
probably from an already established connection, so it’s
considered reasonably safe to let these through.
module [module-name]
Load an iptables module. Most modules provide more match
keywords. We’ll get to that later.
Basic target keywords
jump [custom-chain-name]
Jumps to a custom chain. If no rule in the custom chain
matched, netfilter returns to the next rule in the previous
chain.
realgoto [custom-chain-name]
Go to a custom chain. Unlike the jump option, RETURN will not
continue processing in this chain but instead in the chain that
called us via jump.
The keyword realgoto was chosen during the transition period,
because goto (already deprecated) used to be an alias for jump.
ACCEPT Accepts matching packets.
DROP Drop matching packets without further notice.
REJECT Rejects matching packets, i.e. send an ICMP packet to the
sender, which is port-unreachable by default. You may specify
another ICMP type.
REJECT; # default to icmp-port-unreachable
REJECT reject-with icmp-net-unreachable;
Type "iptables -j REJECT -h" for details.
RETURN Finish the current chain and return to the calling chain (if
"jump [custom-chain-name]" was used).
NOP No action at all.
ADDITIONAL KEYWORDS
Netfilter is modular. Modules may provide additional targets and match
keywords. The list of netfilter modules is constantly growing, and ferm
tries to keep up with supporting them all. This chapter describes
modules which are currently supported.
iptables match modules
account Account traffic for all hosts in defined network/netmask. This
is one of the match modules which behave like a target, i.e.
you will mostly have to use the NOP target.
mod account aname mynetwork aaddr 192.168.1.0/24 ashort NOP;
addrtype
Check the address type; either source address or destination
address.
mod addrtype src-type BROADCAST;
mod addrtype dst-type LOCAL;
Type "iptables -m addrtype -h" for details.
ah Checks the SPI header in an AH packet.
mod ah ahspi 0x101;
mod ah ahspi ! 0x200:0x2ff;
Additional arguments for IPv6:
mod ah ahlen 32 ACCEPT;
mod ah ahlen !32 ACCEPT;
mod ah ahres ACCEPT;
comment Adds a comment of up to 256 characters to a rule, without an
effect. Note that unlike ferm comments (’#’), this one will
show up in "iptables -L".
mod comment comment "This is my comment." ACCEPT;
condition
Matches if a value in /proc/net/ipt_condition/NAME is 1 (path
is /proc/net/ip6t_condition/NAME for the ip6 domain).
mod condition condition (abc def) ACCEPT;
mod condition condition !foo ACCEPT;
connbytes
Match by how many bytes or packets a connection (or one of the
two flows constituting the connection) have tranferred so far,
or by average bytes per packet.
mod connbytes connbytes 65536: connbytes-dir both connbytes-mode bytes ACCEPT;
mod connbytes connbytes !1024:2048 connbytes-dir reply connbytes-mode packets ACCEPT;
Valid values for connbytes-dir: original, reply, both; for
connbytes-mode: packets, bytes, avgpkt.
connlimit
Allows you to restrict the number of parallel TCP connections
to a server per client IP address (or address block).
mod connlimit connlimit-above 4 REJECT;
mod connlimit connlimit-above !4 ACCEPT;
mod connlimit connlimit-above 4 connlimit-mask 24 REJECT;
connmark
Check the mark field associated with the connection, set by the
CONNMARK target.
mod connmark mark 64;
mod connmark mark 6/7;
conntrack
Check connection tracking information.
mod conntrack ctstate (ESTABLISHED RELATED);
mod conntrack ctproto tcp;
mod conntrack ctorigsrc 192.168.0.2;
mod conntrack ctorigdst 1.2.3.0/24;
mod conntrack ctreplsrc 2.3.4.5;
mod conntrack ctrepldst ! 3.4.5.6;
mod conntrack ctstatus ASSURED;
mod conntrack ctexpire 60;
mod conntrack ctexpire 180:240;
Type "iptables -m conntrack -h" for details.
dccp Check DCCP (Datagram Congestion Control Protocol) specific
attributes. This module is automatically loaded when you use
"protocol dccp".
proto dccp sport 1234 dport 2345 ACCEPT;
proto dccp dccp-types (SYNCACK ACK) ACCEPT;
proto dccp dccp-types !REQUEST DROP;
proto dccp dccp-option 2 ACCEPT;
dscp Match the 6 bit DSCP field within the TOS field.
mod dscp dscp 11;
mod dscp dscp-class AF41;
dst Matches the Destination Options header (ip6).
mod dst dst-len 8 ACCEPT;
mod dst dst-len !8 ACCEPT;
mod dst dst-opts (1:4 2:8) ACCEPT;
ecn Match the ECN bits of an IPv4 TCP header.
mod ecn ecn-tcp-cwr;
mod ecn ecn-tcp-ece;
mod ecn ecn-ip-ect 2;
Type "iptables -m ecn -h" for details.
esp Checks the SPI header in an ESP packet.
mod esp espspi 0x101;
mod esp espspi ! 0x200:0x2ff;
eui64 "This module matches the EUI-64 part of a stateless
autoconfigured IPv6 address. It compares the EUI-64 derived
from the source MAC address in Ehternet frame with the lower 64
bits of the IPv6 source address. But "Universal/Local" bit is
not compared. This module doesn’t match other link layer
frame, and is only valid in the PREROUTING, INPUT and FORWARD
chains."
mod eui64 ACCEPT;
frag Matches the Fragment header (ip6).
mod frag fragid 123:456 ACCEPT;
mod frag fragid !123:456 ACCEPT;
mod frag fragres ACCEPT;
mod frag fragfirst ACCEPT;
mod frag fragmore ACCEPT;
mod frag fraglast ACCEPT;
fuzzy "This module matches a rate limit based on a fuzzy logic
controller [FLC]."
mod fuzzy lower-limit 10 upper-limit 20 ACCEPT;
hbh Matches the Hop-by-Hop Options header (ip6).
mod hbh hbh-len 8 ACCEPT;
mod hbh hbh-len !8 ACCEPT;
mod hbh hbh-opts (1:4 2:8) ACCEPT;
hl Matches the Hop Limit field (ip6).
mod hl hl-eq (8 10) ACCEPT;
mod hl hl-eq !5 ACCEPT;
mod hl hl-gt 15 ACCEPT;
mod hl hl-lt 2 ACCEPT;
helper Checks which conntrack helper module tracks this connection.
The port may be specified with "-portnr".
mod helper helper irc ACCEPT;
mod helper helper ftp-21 ACCEPT;
icmp Check ICMP specific attributes. This module is automatically
loaded when you use "protocol icmp".
proto icmp icmp-type echo-request ACCEPT;
This option can also be used in be ip6 domain, although this is
called icmpv6 in ip6tables.
Use "iptables -p icmp "-h"" to obtain a list of valid ICMP
types.
iprange Match a range of IPv4 addresses.
mod iprange src-range 192.168.2.0-192.168.3.255;
mod iprange dst-range ! 192.168.6.0-192.168.6.255;
ipv4options
Match on IPv4 header options like source routing, record route,
timestamp and router-alert.
mod ipv4options ssrr ACCEPT;
mod ipv4options lsrr ACCEPT;
mod ipv4options no-srr ACCEPT;
mod ipv4options !rr ACCEPT;
mod ipv4options !ts ACCEPT;
mod ipv4options !ra ACCEPT;
mod ipv4options !any-opt ACCEPT;
ipv6header
Matches the IPv6 extension header (ip6).
mod ipv6header header !(hop frag) ACCEPT;
mod ipv6header header (auth dst) ACCEPT;
hashlimit
Similar to ’mod limit’, but adds the ability to add per-
destination or per-port limits managed in a hash table.
mod hashlimit hashlimit 10/minute hashlimit-burst 30/minute
hashlimit-mode dstip hashlimit-name foobar ACCEPT;
Possible values for hashlimit-mode: dstip dstport srcip
srcport.
There are more possible settings, type "iptables -m hashlimit
-h" for documentation.
length Check the package length.
mod length length 128; # exactly 128 bytes
mod length length 512:768; # range
mod length length ! 256; # negated
limit Limits the packet rate.
mod limit limit 1/second;
mod limit limit 15/minute limit-burst 10;
Type "iptables -m limit -h" for details.
mac Match the source MAC address.
mod mac mac-source 01:23:45:67:89;
mark Matches packets based on their netfilter mark field. This may
be a 32 bit integer between 0 and 4294967295.
mod mark mark 42;
mh Matches the mobility header (domain ip6).
proto mh mh-type binding-update ACCEPT;
multiport
Match a set of source or destination ports (UDP and TCP only).
mod multiport source-ports (https ftp);
mod multiport destination-ports (mysql domain);
This rule has a big advantage over "dport" and "sport": it
generates only one rule for up to 15 ports instead of one rule
for every port.
nth Match every ’n’th packet.
mod nth every 3;
mod nth counter 5 every 2;
mod nth start 2 every 3;
mod nth start 5 packet 2 every 6;
Type "iptables -m nth -h" for details.
owner Check information about the packet creator, namely user id,
group id, process id, session id and command name.
mod owner uid-owner 0;
mod owner gid-owner 1000;
mod owner pid-owner 5432;
mod owner sid-owner 6543;
mod owner cmd-owner "sendmail";
("cmd-owner", "pid-owner" and "sid-owner" require special
kernel patches not included in the vanilla Linux kernel)
physdev Matches the physical device on which a packet entered or is
about to leave the machine. This is useful for bridged
interfaces.
mod physdev physdev-in ppp1;
mod physdev physdev-out eth2;
mod physdev physdev-is-in;
mod physdev physdev-is-out;
mod physdev physdev-is-bridged;
pkttype Check the link-layer packet type.
mod pkttype pkt-type unicast;
mod pkttype pkt-type broadcase;
mod pkttype pkt-type multicast;
policy Matches IPsec policy being applied to this packet.
mod policy dir out pol ipsec ACCEPT;
mod policy strict reqid 23 spi 0x10 proto ah ACCEPT;
mod policy mode tunnel tunnel-src 192.168.1.2 ACCEPT;
mod policy mode tunnel tunnel-dst 192.168.2.1 ACCEPT;
mod policy strict next reqid 24 spi 0x11 ACCEPT;
Note that the keyword proto is also used as a shorthand version
of protocol (built-in match module). You can fix this conflict
by always using the long keyword protocol.
psd Detect TCP/UDP port scans.
mod psd psd-weight-threshold 21 psd-delay-threshold 300
psd-lo-ports-weight 3 psd-hi-ports-weight 1 DROP;
quota Implements network quotas by decrementing a byte counter with
each packet.
mod quota quota 65536 ACCEPT;
random Match a random percentage of all packets.
mod random average 70;
realm Match the routing realm. Useful in environments using BGP.
mod realm realm 3;
recent Temporarily mark source IP addresses.
mod recent set;
mod recent rcheck seconds 60;
mod recent set rsource name "badguy";
mod recent set rdest;
mod recent rcheck rsource name "badguy" seconds 60;
mod recent update seconds 120 hitcount 3 rttl;
This netfilter module has a design flaw: although it is
implemented as a match module, it has target-like behaviour
when using the "set" keyword.
<http://snowman.net/projects/ipt_recent/>
rt Match the IPv6 routing header (ip6 only).
mod rt rt-type 2 rt-len 20 ACCEPT;
mod rt rt-type !2 rt-len !20 ACCEPT;
mod rt rt-segsleft 2:3 ACCEPT;
mod rt rt-segsleft !4:5 ACCEPT;
mod rt rt-0-res rt-0-addrs (::1 ::2) rt-0-not-strict ACCEPT;
sctp Check SCTP (Stream Control Transmission Protocol) specific
attributes. This module is automatically loaded when you use
"protocol sctp".
proto sctp sport 1234 dport 2345 ACCEPT;
proto sctp chunk-types only DATA:Be ACCEPT;
proto sctp chunk-types any (INIT INIT_ACK) ACCEPT;
proto sctp chunk-types !all (HEARTBEAT) ACCEPT;
Use "iptables -p sctp "-h"" to obtain a list of valid chunk
types.
set Checks the source or destination IP/Port/MAC against a set.
mod set set badguys src DROP;
See <http://ipset.netfilter.org/> for more information.
state Checks the connection tracking state.
mod state state INVALID DROP;
mod state state (ESTABLISHED RELATED) ACCEPT;
Type "iptables -m state -h" for details.
statistic
Successor of nth and random, currently undocumented in the
iptables(8) man page.
mod statistic mode random probability 0.8 ACCEPT;
mod statistic mode nth every 5 packet 0 DROP;
string Matches a string.
mod string string "foo bar" ACCEPT;
mod string algo kmp from 64 to 128 hex-string "deadbeef" ACCEPT;
tcp Checks TCP specific attributes. This module is automatically
loaded when you use "protocol tcp".
proto tcp sport 1234;
proto tcp dport 2345;
proto tcp tcp-flags (SYN ACK) SYN;
proto tcp tcp-flags ! (SYN ACK) SYN;
proto tcp tcp-flags ALL (RST ACK);
proto tcp syn;
proto tcp tcp-option 2;
proto tcp mss 512;
Type "iptables -p tcp -h" for details.
tcpmss Check the TCP MSS field of a SYN or SYN/ACK packet.
mod tcpmss mss 123 ACCEPT;
mod tcpmss mss 234:567 ACCEPT;
time Check if the time a packet arrives is in given range.
mod time timestart 12:00;
mod time timestop 13:30;
mod time days (Mon Wed Fri);
mod time datestart 2005:01:01;
mod time datestart 2005:01:01:23:59:59;
mod time datestop 2005:04:01;
mod time monthday (30 31);
mod time weekdays (Wed Thu);
mod time timestart 12:00 utc;
mod time timestart 12:00 localtz;
Type "iptables -m time -h" for details.
tos Matches a packet on the specified TOS-value.
mod tos tos Minimize-Cost ACCEPT;
mod tos tos !Normal-Service ACCEPT;
Type "iptables -m tos -h" for details.
ttl Matches the ttl (time to live) field in the IP header.
mod ttl ttl-eq 12; # ttl equals
mod ttl ttl-gt 10; # ttl greater than
mod ttl ttl-lt 16; # ttl less than
u32 Compares raw data from the packet. You can specify more than
one filter in a ferm list; these are not expanded into multiple
rules.
mod u32 u32 '6&0xFF=1' ACCEPT;
mod u32 u32 ('27&0x8f=7' '31=0x527c4833') DROP;
unclean Matches packets which seem malformed or unusual. This match has
no further parameters.
iptables target modules
The following additional targets are available in ferm, provided that
you enabled them in your kernel:
CLASSIFY
Set the CBQ class.
CLASSIFY set-class 3:50;
CLUSTERIP
Configure a simple cluster of nodes that share a certain IP and
MAC address. Connections are statically distributed between
the nodes.
CLUSTERIP new hashmode sourceip clustermac 00:12:34:45:67:89
total-nodes 4 local-node 2 hash-init 12345;
CONNMARK
Sets the netfilter mark value associated with a connection.
CONNMARK set-mark 42;
CONNMARK save-mark;
CONNMARK restore-mark;
CONNMARK save-mark mask 0x7fff;
CONNMARK restore-mark mask 0x8000;
CONNSECMARK
This module copies security markings from packets to
connections (if unlabeled), and from connections back to
packets (also only if unlabeled). Typically used in
conjunction with SECMARK, it is only valid in the mangle table.
CONNSECMARK save;
CONNSECMARK restore;
DNAT to [ip-address|ip-range|ip-port-range]
Change the destination address of the packet.
DNAT to 10.0.0.4;
DNAT to 10.0.0.4:80;
DNAT to 10.0.0.4:1024-2048;
DNAT to 10.0.1.1-10.0.1.20;
ECN This target allows to selectively work around known ECN
blackholes. It can only be used in the mangle table.
ECN ecn-tcp-remove;
HL Modify the IPv6 Hop Limit field (ip6/mangle only).
HL hl-set 5;
HL hl-dec 2;
HL hl-inc 1;
IPV4OPTSSTRIP
Strip all the IP options from a packet. This module does not
take any options.
IPV4OPTSSTRIP;
LOG Log all packets that match this rule in the kernel log. Be
carefull with log flooding. Note that this is a "non-
terminating target", i.e. rule traversal continues at the next
rule.
LOG log-level warning log-prefix "Look at this: ";
LOG log-tcp-sequence log-tcp-options;
LOG log-ip-options;
MARK Sets the netfilter mark field for the packet (a 32 bit integer
between 0 and 4294967295):
MARK set-mark 42;
MARK set-xmark 7/3;
MARK and-mark 31;
MARK or-mark 1;
MARK xor-mark 12;
MASQUERADE
Masquerades matching packets. Optionally followed by a port or
port-range for iptables. Specify as "123", "123-456" or
"123:456". The port range parameter specifies what local ports
masqueraded connections should originate from.
MASQUERADE;
MASQUERADE to-ports 1234:2345;
MASQUERADE to-ports 1234:2345 random;
MIRROR Experimental demonstration target which inverts the source and
destination fields in the IP header.
MIRROR;
NETMAP Map a whole network onto another network in the nat table.
NETMAP to 192.168.2.0/24;
NOTRACK Disable connection tracking for all packets matching that rule.
proto tcp dport (135:139 445) NOTRACK;
NFLOG Log packets over netlink; this is the successor of ULOG.
NFLOG nflog-group 5 nflog-prefix "Look at this: ";
NFLOG nflog-range 256;
NFLOG nflog-threshold 10;
NFQUEUE Userspace queueing, requires nfnetlink_queue kernel support.
proto tcp dport ftp NFQUEUE queue-num 20;
QUEUE Userspace queueing, the predecessor to NFQUEUE. All packets go
to queue 0.
proto tcp dport ftp QUEUE;
REDIRECT to-ports [ports]
Transparent proxying: alter the destination IP of the packet to
the machine itself.
proto tcp dport http REDIRECT to-ports 3128;
proto tcp dport http REDIRECT to-ports 3128 random;
SAME Similar to SNAT, but a client is mapped to the same source IP
for all its connections.
SAME to 1.2.3.4-1.2.3.7;
SAME to 1.2.3.8-1.2.3.15 nodst;
SAME to 1.2.3.16-1.2.3.31 random;
SECMARK This is used to set the security mark value associated with the
packet for use by security subsystems such as SELinux. It is
only valid in the mangle table.
SECMARK selctx "system_u:object_r:httpd_packet_t:s0";
SET [add-set|del-set] [setname] [flag(s)]
Add the IP to the specified set. See
<http://ipset.netfilter.org/>
proto icmp icmp-type echo-request SET add-set badguys src;
SNAT to [ip-address|ip-range|ip-port-range]
Change the source address of the packet.
SNAT to 1.2.3.4;
SNAT to 1.2.3.4:20000-30000;
SNAT to 1.2.3.4 random;
TCPMSS Alter the MSS value of TCP SYN packets.
TCPMSS set-mss 1400;
TCPMSS clamp-mss-to-pmtu;
TOS set-tos [value]
Set the tcp package Type Of Service bit to this value. This
will be used by whatever traffic scheduler is willing to,
mostly your own linux-machine, but maybe more. The original
tos-bits are blanked and overwritten by this value.
TOS set-tos Maximize-Throughput;
TOS and-tos 7;
TOS or-tos 1;
TOS xor-tos 4;
Type "iptables -j TOS -h" for details.
TTL Modify the TTL header field.
TTL ttl-set 16;
TTL ttl-dec 1; # decrease by 1
TTL ttl-inc 4; # increase by 4
ULOG Log packets to a userspace program.
ULOG ulog-nlgroup 5 ulog-prefix "Look at this: ";
ULOG ulog-cprange 256;
ULOG ulog-qthreshold 10;
OTHER DOMAINS
Since version 2.0, ferm supports not only ip and ip6, but also arp (ARP
tables) and eb (ethernet bridging tables). The concepts are similar to
iptables.
arptables keywords
source-ip, destination-ip
Matches the source or destination IPv4 address. Same as saddr
and daddr in the ip domain.
source-mac, destination-mac
Matches the source or destination MAC address.
interface, outerface
Input and output interface.
h-length
Hardware length of the packet.
chain INPUT h-length 64 ACCEPT;
opcode Operation code, for details see the iptables(8).
opcode 9 ACCEPT;
h-type Hardware type.
h-type 1 ACCEPT;
proto-type
Protocol type.
proto-type 0x800 ACCEPT;
Mangling
The keywords mangle-ip-s, mangle-ip-d, mangle-mac-s, mangle-
mac-d, mangle-target may be used for ARP mangling. See
iptables(8) for details.
ebtables keywords
proto Matches the protocol which created the frame, e.g. IPv4 or PPP.
For a list, see /etc/ethertypes.
interface, outerface
Physical input and output interface.
logical-in, logical-out
The logical bridge interface.
saddr, daddr
Matches source or destination MAC address.
Match modules
The following match modules are supported: 802.3, arp, ip,
mark_m, pkttype, stp, vlan, log.
Target extensions
The following target extensions are supported: arpreply, dnat,
mark, redirect, snat.
Please note that there is a conflict between --mark from the
mark_m match module and -j mark. Since both would be
implemented with the ferm keyword mark, we decided to solve
this by writing the target’s name in uppercase, like in the
other domains. The following example rewrites mark 1 to 2:
mark 1 MARK 2;
ADVANCED FEATURES
Variables
In complex firewall files, it is helpful to use variables, e.g. to give
a network interface a meaningful name.
To set variables, write:
@def $DEV_INTERNET = eth0;
@def $PORTS = (http ftp);
@def $MORE_PORTS = ($PORTS 8080);
In the real ferm code, variables are used like any other keyword
parameter:
chain INPUT interface $DEV_INTERNET proto tcp dport $MORE_PORTS ACCEPT;
Note that variables can only be used in keyword parameters
("192.168.1.1", "http"); they cannot contain ferm keywords like "proto"
or "interface".
Variables are only valid in the current block:
@def $DEV_INTERNET = eth1;
chain INPUT {
proto tcp {
@def $DEV_INTERNET = ppp0;
interface $DEV_INTERNET dport http ACCEPT;
}
interface $DEV_INTERNET DROP;
}
will be expanded to:
chain INPUT {
proto tcp {
interface ppp0 dport http ACCEPT;
}
interface eth1 DROP;
}
The "def $DEV_INTERNET = ppp0" is only valid in the "proto tcp" block;
the parent block still knows "set $DEV_INTERNET = eth1".
Include files are special - variables declared in an included file are
still available in the calling block. This is useful when you include a
file which only declares variables.
Automatic variables
Some variables are set internally by ferm. Ferm scripts can use them
just like any other variable.
$FILENAME
The name of the configuration file.
$DOMAIN The current domain. One of ip, ip6, arp, eb.
$TABLE The current netfilter table.
$CHAIN The current netfilter chain.
Functions
Functions are similar to variables, except that they may have
parameters, and they provide ferm commands, not values.
@def &FOO() = proto (tcp udp) dport domain;
&FOO() ACCEPT;
@def &TCP_TUNNEL($port, $dest) = {
table filter chain FORWARD interface ppp0 proto tcp dport $port daddr $dest outerface eth0 ACCEPT;
table nat chain PREROUTING interface ppp0 proto tcp dport $port daddr 1.2.3.4 DNAT to $dest;
}
&TCP_TUNNEL(http, 192.168.1.33);
&TCP_TUNNEL(ftp, 192.168.1.30);
&TCP_TUNNEL((ssh smtp), 192.168.1.2);
A function call which contains a block (like ’{...}’) must be the last
command in a ferm rule, i.e. it must be followed by ’;’. The ’&FOO()’
example does not contain a block, thus you may write ’ACCEPT’ after the
call. To circumvent this, you can reorder the keywords:
@def &IPSEC() = { proto (esp ah); proto udp dport 500; }
chain INPUT ACCEPT &IPSEC();
Backticks
With backticks, you may use the output of an external command:
@def $DNSSERVERS = `grep nameserver /etc/resolv.conf | awk '{print $2}'`;
chain INPUT proto tcp saddr $DNSSERVERS ACCEPT;
The command is executed with the shell (/bin/sh), just like backticks
in perl. ferm does not do any variable expansion here.
The output is then tokenized, and saved as a ferm list (array). Lines
beginning with ’#’ are ignored; the other lines may contain any number
of values, separated by whitespace.
Includes
The @include keyword allows you to include external files:
@include 'vars.ferm';
The file name is relative to the calling file, e.g. when including from
/etc/ferm/ferm.conf, the above statement includes /etc/ferm/vars.ferm.
Variables and functions declared in an included file are still
available in the calling file.
include works within a block:
chain INPUT {
@include 'input.ferm';
}
If you specify a directory (with a trailing ’/’), all files in this
directory are included, sorted alphabetically:
@include 'ferm.d/';
With a trailing pipe symbol, ferm executes a program and parses its
output:
@include '/root/generate_ferm_rules.sh $HOSTNAME|'
Conditionals
The keyword @if introduces a conditional expression:
@if $condition DROP;
A value is evaluated true just like in Perl: zero, empty list, empty
string are false, everything else is true. Examples for true values:
(a b); 1; 'foo'; (0 0)
Examples for false values:
(); 0; '0'; ''
There is also @else:
@if $condition DROP; @else REJECT;
Note the semicolon before the @else.
It is possible to use curly braces after either @if or @else:
@if $condition {
MARK set-mark 2;
RETURN;
} @else {
MARK set-mark 3;
}
Since the closing curly brace also finishes the command, there is no
need for semicolon.
There is no @elsif, use @else @if instead.
Example:
@def $have_ipv6 = `test -f /proc/net/ip6_tables_names && echo 1 || echo`;
@if $have_ipv6 {
domain ip6 {
# ....
}
}
Hooks
To run custom commands, you may install hooks:
@hook pre "echo 0 >/proc/sys/net/ipv4/conf/eth0/forwarding";
@hook post "echo 1 >/proc/sys/net/ipv4/conf/eth0/forwarding";
@hook flush "echo 0 >/proc/sys/net/ipv4/conf/eth0/forwarding";
The specified command is executed using the shell. "pre" means run the
command before applying the firewall rules, and "post" means run the
command afterwards. "flush" hooks are run after ferm has flushed the
firewall rules (option --flush). You may install any number of hooks.
BUILT-IN FUNCTIONS
There are several built-in functions which you might find useful.
@resolve((hostname1 hostname2 ...))
Usually, host names are resolved by iptables. To let ferm resolve host
names, use the function @resolve:
saddr @resolve(my.host.foo) proto tcp dport ssh ACCEPT;
saddr @resolve((another.host.foo third.host.foo)) proto tcp dport openvpn ACCEPT;
Note the double parentheses in the second line: the inner pair for
creating a ferm list, and the outer pair as function parameter
delimiters.
This function currently only resolves A records (i.e. IPv4 addresses),
rendering it useless for the ip6 domain.
Be careful with resolved host names in firewall configuration. DNS
requests may block the firewall configuration for a long time, leaving
the machine vulnerable, or they may fail.
RECIPES
The ./examples/ directory contains numerous ferm configuration which
can be used to begin a new firewall. This sections contains more
samples, recipes and tricks.
Easy port forwarding
Ferm function make routine tasks quick and easy:
@def &FORWARD_TCP($proto, $port, $dest) = {
table filter chain FORWARD interface $DEV_WORLD outerface $DEV_DMZ daddr $dest proto $proto dport $port ACCEPT;
table nat chain PREROUTING interface $DEV_WORLD daddr $HOST_STATIC proto $proto dport $port DNAT to $dest;
}
&FORWARD_TCP(tcp, http, 192.168.1.2);
&FORWARD_TCP(tcp, smtp, 192.168.1.3);
&FORWARD_TCP((tcp udp), domain, 192.168.1.4);
Remote ferm
If the target machine is not able to run ferm for some reason (maybe an
embedded device without Perl), you can edit the ferm configuration file
on another computer and let ferm generate a shell script there.
Example for OpenWRT:
ferm --remote --shell mywrt/ferm.conf >mywrt/firewall.user
chmod +x mywrt/firewall.user
scp mywrt/firewall.user mywrt.local.net:/etc/
ssh mywrt.local.net /etc/firewall.user
OPTIONS
--noexec Do not execute the iptables(8) commands, but skip instead.
This way you can parse your data, use --lines to view the
output.
--flush Clears the firewall rules and sets the policy of all chains
to ACCEPT. ferm needs a configuration file for that to
determine which domains and tables are affected.
--lines Show the firewall lines that were generated from the rules.
They will be shown just before they are executed, so if you
get error messages from iptables(8) etc., you can see which
rule caused the error.
--interactive
Apply the firewall rules and ask the user for confirmation.
Reverts to the previous ruleset if there is no valid user
response within 30 seconds. This is useful for remote
firewall administration: you can test the rules without
fearing to lock yourself out.
--help Show a brief list of available commandline options.
--version Shows the version number of the program.
--fast Enable fast mode: ferm generates an iptables-save(8) file,
and installs it with iptables-restore(8). This is much
faster, because ferm calls iptables(8) once for every rule
by default.
Fast mode is enabled by default since ferm 2.0, deprecating
this option.
--slow Disable fast mode, i.e. run iptables(8) for every rule, and
don’t use iptables-restore(8).
--shell Generate a shell script which calls iptables-restore(8) and
prints it. Implies --fast --lines.
--remote Generate rules for a remote machine. Implies --noexec and
--lines. Can be combined with --shell.
--domain {ip|ip6}
Handle only the specified domain. ferm output may be empty
if the domain is not configured in the input file.
--def ’$name=value’
Override a variable defined in the configuration file.
SEE ALSO
iptables(8)
REQUIREMENTS
Operating system
Linux 2.4 or newer, with netfilter support and all netfilter modules
used by your firewall script
Software
iptables and perl 5.6
BUGS
Bugs? What bugs?
If you find a bug, please tell us: ferm@foo-projects.org
COPYRIGHT
Copyright (C) 2001-2009 Max Kellermann <max@foo-projects.org>, Auke Kok
<sofar@foo-projects.org>
This program is free software; you can redistribute it and/or modify it
under the terms of the GNU General Public License as published by the
Free Software Foundation; either version 2 of the License, or (at your
option) any later version.
This program is distributed in the hope that it will be useful, but
WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
General Public License for more details.
You should have received a copy of the GNU General Public License along
with this program; if not, write to the Free Software Foundation, Inc.,
59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
AUTHOR
Max Kellermann <max@foo-projects.org>, Auke Kok
<sofar@foo-projects.org>