Iptables is a powerful administration tool for IPv4 packet filtering and NAT. It is used to set up, maintain, and inspect the tables of IP packet filter rules in the Linux kernel.
Iptables commands can be entered by command line interface, and/or saved as a Firewall script in the dd-wrt Administration panel. I tend to recommend testing and confirming your rules at the command line first. This way, if you happen to make a big mistake (like blocking access to the router), simply rebooting the router should repair it rather than having to do a hard reset. To get your rules to survive a reboot of the router, save them in a Firewall script as mentioned earlier.
I think we should have something about Firewall Builder on this page, since they're kind of related...
[edit]Basic Usage
iptables -[AD] chain rule-specification [options] iptables -[RI] chain rulenum rule-specification [options] iptables -D chain rulenum [options] iptables -[LFZ] [chain] [options] iptables -[NX] chain iptables -E old-chain-name new-chain-name iptables -P chain target [options] iptables -h (print this help information)
[edit]Commands
--append -A chain Append to chain
--delete -D chain Delete matching rule from chain
--delete -D chain rulenum
Delete rule rulenum (1 = first) from chain
--insert -I chain [rulenum]
Insert in chain as rulenum (default 1=first)
--replace -R chain rulenum
Replace rule rulenum (1 = first) in chain
--list -L [chain] List the rules in a chain or all chains
--flush -F [chain] Delete all rules in chain or all chains
--zero -Z [chain] Zero counters in chain or all chains
--new -N chain Create a new user-defined chain
--delete-chain
-X [chain] Delete a user-defined chain
--policy -P chain target
Change policy on chain to target
--rename-chain
-E old-chain new-chain
Change chain name, (moving any references)[edit]Options
--proto -p [!] proto
protocol: by number or name, eg. `tcp'
--source -s [!] address[/mask]
source specification. The mask can be either a network mask or a plain number,
specifying the number of 1's at the left side of the network mask. Thus, a mask of 24 is
equivalent to 255.255.255.0. A "!" argument before the address specification inverts the
sense of the address.
--destination -d [!] address[/mask]
destination specification
--sport [!] port[:endport]
source port (use `:' when specifying range)
--dport [!] port[:endport]
destination port
--in-interface -i [!] input name[+]
network interface name ([+] for wildcard)
--jump -j target
target for rule (may load target extension)
--match -m match
extended match (may load extension)
--state state
connection states to match:
INVALID NEW ESTABLISHED RELATED
--tcp-flags [!] mask
match when the TCP flags are as specified:
SYN ACK FIN RST URG PSH ALL NONE
--numeric -n numeric output of addresses and ports
--out-interface -o [!] output name[+]
network interface name ([+] for wildcard)
--table -t table table to manipulate (default: `filter')
--verbose -v verbose mode
--line-numbers print line numbers when listing
--exact -x expand numbers (display exact values)
--fragment -f match second or further fragments only
--modprobe=<command> try to insert modules using this command
--set-counters PKTS BYTES set the counter during insert/append
--version -V print package version
MAC v1.3.7 options:
--mac-source [!] XX:XX:XX:XX:XX:XX
Match source MAC address[edit]Interfaces
When using the -i or -o to define the physical interfaces, remember that by default:
vlan0 is the 4 LAN ports
vlan1 is the WAN port (ppp0 is the WAN interface when PPPoE is used)
eth1 is the WIFI
br0 is a bridge connecting the 4 LAN and the WIFI together
vlan0 is the 4 LAN ports
vlan1 is the WAN port (ppp0 is the WAN interface when PPPoE is used)
eth1 is the WIFI
br0 is a bridge connecting the 4 LAN and the WIFI together
Note: ppp0 is the WAN interface when PPPoE is used. This information is from IPv6 page and quoted here: "The detailled configuration steps are targeted toward users with a basic DHCP connection for the WAN part. So, if using PPPoE will require replacing vlan1 with ppp0 in each instance. Other connection types will vary."
Tip: To list the network interfaces on the router use 'ifconfig' on the command line.
[edit]Tables, Chains, and Targets
[edit]Tables
The main tables we are concerned with are the "filter" table and the "nat" table. To list the contents of either table, do
iptables -t filter -L iptables -t nat -L
The filter table is default and this includes chains like INPUT, OUTPUT, and FORWARD. The nat table is for Network Address Translation and it includes the PREROUTING and POSTROUTING chains.
[edit]Chains
INPUT is for packets destined to or entering the router's local sockets.
OUTPUT is for packets sourced from or leaving the router's local sockets.
FORWARD is for packets being forwarded through the router (e.g. packets not necessarily destined for local sockets).
PREROUTING is for manipulating packets before they are routed.
POSTROUTING is for manipulating packets after they are routed.
[edit]Targets
ACCEPT - packets are accepted/allowed
DROP - packets are dropped/denied
REJECT - packets are rejected/denied
logaccept - packets are accepted and logged to /tmp/var/log/messages
logdrop - packets are dropped and logged to /tmp/var/log/messages
logreject - packets are rejected and logged to /tmp/var/log/messages
DNAT is for altering packet's destination address.
SNAT is for altering packet's source address.
TRIGGER - dynamically redirect input ports based on output traffic (aka port triggering)
[edit]TRIGGER Target Options
The trigger target has additional options which must appear immediately after it on the command line
--trigger-type [dnat|in|out] --trigger-proto [tcp|udp|all] (if this option is not specificed the default is all) --trigger-match [port[-port]] (a port or a range of ports which the outbound connection uses) --trigger-relate [port[-port]] (a port or range of ports to open on the inbound side)
[edit]Examples
I think examples are the best way to demonstrate the use of iptables. (Take note, chains are to be typed in caps as shown!)
[edit]Listing the rules in a chain
First I want to view the rules on my INPUT chain, this is the first chain traffic coming into my router will hit.
iptables -L INPUT
You will find that it is really slow to list all many rules after you enter the above iptables command since it is doing reverse DNS lookups to convert IP addresses to host names. You can add -n option to only see numerical addresses. Note: '0.0.0.0/0' = 'anywhere' (any IP address), and '0' prot = 'any' protocol.
iptables -nL INPUT
To get a more detailed list with actual IP numbers and packet counts for each rule do this.
iptables -vnL INPUT
Please always use -vnL when troubleshooting, especially if you're asking for help on the forums. Anything less hides valuable information and are only explained on this page for reference.
Suppose I might want to add a rule so that I can ssh into my router from a specific host/address outside. Then I might type the following:
iptables -I INPUT -p tcp -s 150.100.whatever.something --dport 22 -j logaccept
So I am saying: Append to the INPUT chain a rule allowing protocol tcp, with a source address of <my external IP that i want access from> traffic destined for port 22 on my router, jump to logaccept. I could have used -j ACCEPT which simply jumps to ACCEPT, but in this case I want to log it just to keep track so I use logaccept, which is a chain we have set up for this purpose.
Note: Simply adding a rule to the INPUT chain may be enough to allow remote SSH access from the WAN. However, if your router is still in NAT/Gateway mode and you wish to remap the SSH port to something less traditional on the WAN side (say port 2222), you may Insert a PREROUTING rule instead. This is actually how the GUI does it when you enable remote WAN SSH management.
iptables -I INPUT -p tcp -m tcp -d `nvram get lan_ipaddr` --dport 22 -j logaccept iptables -t nat -I PREROUTING -p tcp -m tcp -d `nvram get wan_ipaddr` --dport 2222 -j DNAT --to-destination `nvram get lan_ipaddr`:22
Now if I type
iptables -vnL INPUT
I see my shiny new rule appended to the INPUT chain. However, this is no good because in my case I have a rule blocking this traffic which occurs BEFORE the rule allowing it.
How do I change it? Simple.
First let's delete the rule we just made
iptables -vnL INPUT --line-numbers
will list the rules with their rule numbers. Let's say our rule is number 11
iptables -D INPUT 11
Clearly this Deletes rule number 11 from the input chain.
Now instead of Appending I am going to Insert my rule into the number 1 (by default) position.
iptables -I INPUT -p tcp -s 150.100.whatever.something --dport 22 -j logaccept
So now rule number 1 is my new rule and the other rules have all shifted down a position.
If I wanted to change the IP address or any other aspect of my ssh rule I could use the -R (Replace) option for a specific rule number, and simply type in the new rule, i.e.
iptables -R INPUT 1 -p tcp -s 100.100.200.100 --dport 22 -j ACCEPT
This would replace rule number 1 on the INPUT chain with the new rule which has a new source IP address and jumps to ACCEPT instead of logaccept.
One more example: I want to run a mini web server on my router. Let's assume that it is already running on port 8000 and I can access it from the LAN side, but not from the WAN side. With
iptables -I INPUT -p tcp -d 192.168.1.1 --dport 8000 -j logaccept
the port 8000 will be opened. But I also have to setup NAT PREROUTING, so that the kernel forwards all packets on port 8000 from the outside to itself, 192.168.1.1:
iptables -t nat -I PREROUTING -p tcp -d $(nvram get wan_ipaddr) --dport 8000 -j DNAT --to 192.168.1.1:8000
[edit]Port Forwarding to a specific LAN IP
Port Forwarding can be accomplished from within the web interface here. However, the very same thing can be done a bit differently (tested and working), via command line. --u3gyxap: Example with port 443 and IP 192.168.1.2
iptables -t nat -I PREROUTING -p tcp -d $(nvram get wan_ipaddr) --dport 443 -j DNAT --to 192.168.1.2:443 iptables -I FORWARD -p tcp -d 192.168.1.2 --dport 443 -j ACCEPT
If you want to restrict the source IP (a question that is asked a lot on the forums), add -s 123.45.67.89 to one of your rules (replacing the IP address with the real one of course).
iptables -t nat -I PREROUTING -p tcp -s 123.45.67.89 -d $(nvram get wan_ipaddr) --dport 443 -j DNAT --to 192.168.1.2:443 iptables -I FORWARD -p tcp -d 192.168.1.2 --dport 443 -j ACCEPT
This should make it so only one IP address is able to access your forwarded port from the Internet.
In order for me to get this to work (v.24) I needed to put the "-s 123.45.67.89" in the "iptables -I FORWARD" command also - When it was in the PREROUTING command only I was still able to access the internal resource from any IP address!
[edit]Deny access to a specific IP address
iptables -I FORWARD -d 123.123.123.123 -j DROP
Which would DROP all packets destined to the given IP. Useful to block access to whatnot. If you want to log the entry when the IP is blocked you would set the jump location to logdrop, instead of DROP.
[edit]Deny access to a specific Subnet
iptables -I FORWARD -s 192.168.2.0/255.255.255.0 -j DROP
[edit]Deny access to a specific IP address range with Logging
iptables -I FORWARD -m iprange --src-range 192.168.1.10-192.168.1.13 -j logdrop
Many builds do not have the iprange match but you can use clever subnet masks to accomplish something similar as well, if the range aligns well on subnet boundaries. You may also be able to download a version of iptables that includes the iprange match via Optware.
[edit]Deny access to a specific Outbound IP address with logging
iptables -I OUTPUT -d 239.255.255.250 -j logdrop
This becomes useful if there is a program that wants to gain an outbound connection to a specific address, but you don't want to allow the connection. In this specific example Windows uses this IP incorrectly as a broadcast address (search Google for more info). While viewing your router logs you will see Windows broadcast to this IP several times per minute. By default the router passes the broadcast and announces to everyone outside of your router that your PC exists. This rule will block traffic to this specific outbound IP and add an entry into the router log.
edit: There is nothing incorrect about this. This is the service announcement/discovery multicast address used by SSDP. This is required to discover uPnP based devices in your network. If you drop these, your DLNA media servers, ushare, minidlna, PS3s, Xbox's etc will not see each other if they are across subnets. These packets have a TTL of 4, so won't get too far out of your network. 239.x.x.x is private IPv4 multicast range, so ISPs would drop this at their ingress points.
[edit]Block SMTP traffic except to specified hosts
Simple Mail Transfer Protocol operates on tcp port 25.
/usr/sbin/iptables -I FORWARD 1 -p tcp -d safe.server1.com --dport 25 -j logaccept /usr/sbin/iptables -I FORWARD 2 -p tcp -d safe.server2.com --dport 25 -j logaccept /usr/sbin/iptables -I FORWARD 3 -p tcp --dport 25 -j logdrop
Which would accept and log all smtp traffic to safe.server1.com and safe.server2.com, while blocking and dropping all other outgoing smtp traffic.
[edit]Block outgoing SMTP traffic except from specified hosts
Simple Mail Transfer Protocol operates on tcp port 25.
iptables -I FORWARD 1 -p tcp -s 192.168.1.2 --dport 25 -j ACCEPT iptables -I FORWARD 2 -p tcp -s 192.168.1.1/24 --dport 25 -j REJECT
Which would accept outgoing SMTP traffic from your internal SMTP server (192.168.1.2) but reject outgoing SMTP traffic from all other hosts on your LAN (192.168.1.1/24). Useful to enforce all your LAN clients to use your internal SMTP server, as well as to block any viruses and spam-generating trojans from sending mail to remote servers on their own.
Change "REJECT" to "logdrop" or "ACCEPT" to "logaccept" to add logging.
Caution! This will also block internal users from using your external IP as their SMTP server.
[edit]Allow HTTP traffic only to specific domain(s)
Similarly, we can use the above method to filter other ports and protocols as well, such as standard web traffic operating on tcp port 80.
iptables -I FORWARD 1 -p tcp -d dd-wrt.com --dport 80 -j ACCEPT iptables -I FORWARD 2 -p tcp --dport 80 -j DROP
Which would accept all http traffic to dd-wrt.com, while blocking outgoing http traffic to anywhere else. If you wish to allow multiple sites, insert additional rules before the DROP (making sure to order and number them correctly).
[edit]Block all traffic except HTTP HTTPS and FTP
This example blocks everything except our normal web traffic, encrypted (ssl), and the file transfer protocol.
iptables -I FORWARD 1 -p tcp -m multiport --dports 21,80,443 -j ACCEPT iptables -I FORWARD 2 -m state --state ESTABLISHED,RELATED -j ACCEPT iptables -I FORWARD 3 -j DROP
Caution! Users are still able to get through the firewall if they are sly enough to use these permitted port numbers for their P2P or other application. In that case, you should consider using Access Restrictions to mitigate the possibility of that happening.
[edit]Reject clients from accessing the router's configuration
This should prevent clients on the LAN (interface br0) from accessing the configuration interface of the router through any of the following ports (telnet/23, ssh/22, http/80, https/443)
iptables -I INPUT -i br0 -p tcp --dport telnet -j REJECT --reject-with tcp-reset iptables -I INPUT -i br0 -p tcp --dport ssh -j REJECT --reject-with tcp-reset iptables -I INPUT -i br0 -p tcp --dport www -j REJECT --reject-with tcp-reset iptables -I INPUT -i br0 -p tcp --dport https -j REJECT --reject-with tcp-reset
Tip: If you disable management from the LAN, be sure to enable remote management on the WAN (or vice versa) or you will probably lock yourself out of the router.
[edit]Restrict access by MAC address
In this example, we will demonstrate how to restrict access to the router's web interface by MAC address. In other words, only the computer having the specified MAC address should be able to access the web interface from the LAN.
First, if there are no Access Restrictions policies enabled and filtering by MAC addresses, you may need to insert the iptables mac module manually:
insmod xt_mac #k2.6 module name insmod ipt_mac #k2.4 module name iptables -I INPUT -p tcp --dport 80 -m mac ! --mac-source 00:12:34:56:78:9A -j REJECT --reject-with tcp-reset
Notice the ! (bang) which is another new concept introduced here. It means "NOT". So, by inspecting the rule closely, we see that it will REJECT packets destined to port 80 of the router so long as they do NOT originate from our computer with the desired MAC address.
Caution! As usual when dealing with MAC addresses, be aware that it is possible for malicious user(s) to spoof their MAC address with that of a trusted machine. You can help combat this by use of static ARP entries, VLANs, etc.
[edit]Modifying the TTL
The Time To Live is the maximum number of routers a packet will travel through before it is discarded. In certain situations, it may prove useful to increase it (typically) in order to make your network more reliable.
- Example 1: Set the incoming TTL to 10, before the router routes it into the LAN
iptables -t mangle -I PREROUTING -i `get_wanface` -j TTL --ttl-set 10
- Example 2: Set the outgoing TTL to 128, just as if a Windows machine was connected directly to the modem.
iptables -t mangle -I POSTROUTING -o `get_wanface` -j TTL --ttl-set 128
- Example 3: Try to hide the fact that an outgoing packet was routed, by incrementing the TTL by one.
iptables -t mangle -I POSTROUTING -o `get_wanface` -j TTL --ttl-inc 1
[edit]Firewall Forwarded Ports
If you have enabled SPI firewall feature on DD-WRT, your router is pretty much protected. However one practical use of iptables is protecting certain ports forwarded to internal IP addresses. The simplest way to do this is:
- Create port forwarding to internal IP addresses using the DD-WRT "Port Forwarding" web interface
- Supplement those rules with custom iptables on the Firewall script found under Administration - Commands interface to restrict which hosts can access the ports involved
[edit]Port Forward Example
The current port forwarded setup via the web GUI will be used as the basis to illustrate some examples:
- Application: ssh Port: 4022 Protocol: TCP forward to IP address: 192.168.1.5 Port: 22
- Application: ftp Port: 21 Protocol: TCP forward to IP address: 192.168.1.6 Port: 21
The example here port forwards external IP on port 4022 to internal server 192.168.1.5:22 for ssh and external port 21 to internal server 192.168.1.6:21 for ftp.
[edit]Firewall Rule Examples
You may first want to limit your ssh port to script kiddies, and prevent brute force attack. Thus you can limit the number of NEW ssh connections to about 3 attempts per minute. Any further attempts to crack the ssh port will be dropped:
iptables -I FORWARD -p tcp -d 192.168.1.5 --dport 22 -j DROP iptables -I FORWARD -p tcp --dport 22 -m state --state NEW -m limit --limit 3/min -j ACCEPT iptables -I FORWARD -p tcp --dport 22 -m state --state RELATED,ESTABLISHED -j ACCEPT
FTP access can also be limited to a certain network or network range in the following manner:
iptables -I FORWARD -p tcp -d 192.168.1.6 --dport 21 -j DROP iptables -I FORWARD -p tcp -s 198.133.219.0/24 --dport 21 -j ACCEPT iptables -I FORWARD -p tcp -s 69.147.64.0/18 --dport 21 -j ACCEPT iptables -I FORWARD -p tcp --dport 21 -m state --state RELATED,ESTABLISHED -j ACCEPT
You can of course combine both, rate limit and IP addresses limiting. This following example limits ssh connection from 207.171.160.0/19 with the same rate limit applied, along with the FTP rules all on the same script:
# limit SSH connection from script kiddies in Amazon's network iptables -I FORWARD -p tcp -d 192.168.1.5 --dport 22 -j DROP iptables -I FORWARD -p tcp -s 207.171.160.0/19 --dport 22 -m state --state NEW -m limit --limit 3/min -j ACCEPT iptables -I FORWARD -p tcp --dport 22 -m state --state RELATED,ESTABLISHED -j ACCEPT # limit FTP connection to two subnets iptables -I FORWARD -p tcp -d 192.168.1.6 --dport 21 -j DROP iptables -I FORWARD -p tcp -s 198.133.219.0/24 --dport 21 -j ACCEPT iptables -I FORWARD -p tcp -s 69.147.64.0/18 --dport 21 -j ACCEPT iptables -I FORWARD -p tcp --dport 21 -m state --state RELATED,ESTABLISHED -j ACCEPT
For multiport INPUT (or FORWARD if you choose) rate limiting, the following syntax rules can be esstablshed:
wanf=`get_wanface` iptables -I INPUT 2 -i $wanf -p tcp -m multiport --dports 21,22,80,82 -j logdrop iptables -I INPUT 2 -i $wanf -p tcp -m state --state NEW -m multiport --dports 21,22,80,82 -m limit --limit 3/minute --limit-burst 2 -j logaccept
To verify that the rules are working, open a terminal session and type iptables -vnL | more
If you're adding a lot of rules, it helps to separate them with comments using the # prefix. You can of course use the basics of iptables explained in this article to make your rules more complex to suit your needs. However, instead of ordering the rules, the examples here merely insert these new ones on top of the FORWARD chain. This ensures that the firewall rules that limit traffic appear on top of the chain and gets applied first.
You may examine these rules on the router at anytime by accessing the router's command prompt and running the command "iptables -vnL"
[edit]Logging
You can consider turning on logging temporarily for any of your rules. This is useful if you're testing new setup to confirm that the rules are doing what you intend to block or allow. First enable logging via the web UI at Security - Firewall tab. Then substitute the jump target or "-j" to a logging target for each of your iptables rule:
- DROP with logdrop
- REJECT with logreject
- ACCEPT with logaccept
Example if you wanted to check and confirm if traffic forwarded to port 21 is correctly dropped you would substitute:
iptables -I FORWARD -d 192.168.1.6 -p tcp --dport 21 -j DROP
with
iptables -I FORWARD -d 192.168.1.6 -p tcp --dport 21 -j logdrop
Logged data can be viewed on the web UI on the same page or on the command prompt in the file "/var/log/messages"
[edit]Firewall blocks DHCP renewal responses
iptables -I INPUT -p udp --dport 68 -j ACCEPT
The default configuration of the firewall blocks DHCP renewal responses which causes the router's DHCP client to request a new IP and for current connections to be dropped whether the address changes or not. ~phuzi0n Use this command to fix it. Replace ACCEPT with logaccept to verify it is functioning.
[edit]Caution
Adding iptables commands to your startup routine risks locking yourself out of the box with no option but to start over. If you are experimenting with new commands, you can insure yourself against this scenario by inserting a sleep command before the iptables command(s). This way you can grant yourself, say five minutes (sleep 300), before your commands take effect. If your commands do backfire and you are unable to log in to your box, simply restart it by switching it off and on again and you will have five minutes to get in.
[edit]See Also
[edit]External Resources
- blocking all traffic except HTTP HTTPS and MSN (Windows Live Messenger) With iptables
- linux_journal_video_tut
- http://linux.die.net/man/8/iptables
- http://www.iptables.org/documentation/HOWTO//packet-filtering-HOWTO-7.html
- http://www.iptables.org/documentation/HOWTO//netfilter-hacking-HOWTO.html
- iptables command line examples
http://www.dd-wrt.com/wiki/index.php/Iptables#Modifying_the_TTL
