Flytrap

Seasoned attackers, and even some amateur cyber-vandals, find sport in trying to scan servers and hijack them at the same time (Figure 1). Firewalls and Intrusion Detection/Prevention systems can help, but if a single tool could truly stop all potential attacks, the Internet intrusion industry wouldn't even exist.

Figure 1: Number of unuathorized login attempts on the author's network in a two-month period (by country or origin).

A professional intrusion attempt is typically preceded by reconnaissance and scanning. Many attackers simply perform a scan, which is easily automated with tools like Nmap. An attacker who discovers a firewall and similar defensive system can often guess which ports and services are worth attacking. However, a tool called Portspoof [1] intervenes to cause complications and confusion for the attacker. Portspoof answers port requests with a wild mix of signatures and payloads. This confusing and unwanted information slows down any attempted port scan, forcing the attacker to manually evaluate the results in a time-consuming process.

Portspoof was developed in 2012 by Piotr Duszynski, who calls his program a "Service Emulator and Frontend Exploitation Framework." The application is available under the GPLv2 and is implemented in C++.

Portspoof presents various service signatures on some or all available ports, making it very difficult to discover which services are actually running on the computer. The application can simulate more than 8,000 signatures and has the ability to throw a couple of exploits back at the scanning computer.

When launched, Portspoof only listens on one port (port 4444 by default). All other ports, where attackers fall for the trap, are redirected using an iptables rule.

Installation

Portspoof is currently available in version ++1.0. Download the tool as a ZIP archive or via Git (Listing 1).

After unzipping or cloning, change to the directory created in the process, portspoof[-master]; then install the application in the three normal steps (Listing 1, line 2). By default, Portspoof ends up in /usr/local/. To define a different target directory, pass the --prefix parameter into ./configure, along with your directory choice.

After the install, quickly run portspoof -h to check whether everything has worked out so far. If everything is working, you can now run Portspoof.

Gluing and Sticking

Before launching Portspoof, you must tell iptables which ports you want to be "sticky" in the future. To let all ports join in with the fun, use the following directive:

# iptables -t nat -A PREROUTING -i eth0 -p tcp -mtcp --dport 1:65535 -j REDIRECT --to-ports 4444

However, if you are running some genuine services on the computer, omit the matching ports using the --dport option with --match multiport --dports. For example, the following command forwards all ports except 22 and 80:

# iptables -t nat -A PREROUTING -i eth0 -p tcp -m tcp -m multiport --dports 1:21,23:79,81:65535 -j REDIRECT --to-ports 4444

The iptables multiport feature can handle a maximum of 15 port ranges. If you have a busy server – like the ones used in small businesses that serve everything from DNS, FTP, SSH, mail, web, and file sharing – you might need a small loop to conveniently set up and manage the IP ranges.

An example is provided in set-spoofports.sh (Listing 2). You need to make the script executable with chmod 755 set-spoofports.sh to execute it conveniently. Then, you can use iptables -L -t nat to check afterwards whether the firewall rules are registered.

Once iptables is forwarding all the ports to the target port, you can launch Portspoof. This small tool offers a useful feature set. For example, you can switch Portspoof to verbose mode with the -v option and listen in live when a "visitor" drops by. Alternatively, you can start the program as a daemon that performs its work in the background using the with -D option.

If you call Portspoof without any additional parameters, it works in what is known as open port mode. As the name suggests, open port mode means that Portspoof only shows all forwarded ports as open. That's a nice tick, but you can do much nastier things, too. For example, why not try the following command?

# portspoof -c </path/to>/portspoof.conf -s </path/to>/portspoof_signatures -D

The Service Emulator now responds to requests with heaps of signatures or even exploits, which greatly complicate the attacker's task of evaluating the results (Figure 2). Both portspoof.conf, as well as portspoof_signatures, reside by default in /usr/local/etc. The portspoof_signatures file contains prepared signatures.

Figure 2: Portspoof relies on 8000 signatures to thoroughly confuse port scanners.

Little Tricks

The portspoof.conf can be populated with signatures and even exploits that Portspoof then delivers when a remote entity queries a specific port or port range. The specification of the signature file typically suffices; you only need portspoof.conf to output precisely defined feedback for selected ports.

In portspoof.conf, you enter (line by line) the port for which the content ("payload") is intended. The payload can be simple ASCII words, but also hexadecimal characters or regular expressions. Even exploits can be returned (caution: see the "Computer Sabotage" box). The default portspoof.conf is well commented by the Portspoof developer, which makes it easy to add your own messages and niceties.

This is by no means the end of the fun for the target of the scan. Piotr Duszynski seems to have developed Portspoof under the motto that nothing is unfriendly enough for attackers, and therefore built in several additional fuzzing functions. For example, you can return several payloads specified in a file to an individual port or even generate completely arbitrary payloads in queries. A call to portspoof -f </path/to/redridinghood>.txt -D treats the scanner to a rendition of the fairy tale "Little Red Riding Hood" – known as "Little Red Cap" in the German version (Figure 3) – and portspoof -1 -D would generate random signatures (Figure 4).

Figure 3: Lulling port scanners with Little Red Riding Hood (in German, à la the Brothers Grimm).

Figure 4: Inundating the attacker with a payload of random characters.

Enable another fuzzer by adding -1 or -f </path/text>.txt with -n /path/signature_name.txt to a signature file. Now Portspoof not only provides entertaining, instructive, or motley payloads, but additionally adds services for signatures that you actually might encounter in the wild (Figure 5).

Figure 5: Combining tricks with Portspoof.

By returning arbitrary or predefined signatures and payloads, Portspoof significantly slows down a network scanner. With the service emulation alone, the time for a scan of the entire port range in the test was just under 10 hours (Figure 6); without Portspoof, it would have been done in just over 90 seconds. If you bring fuzzing into play, an attacker needs even more patience; a scan of all ports can take 30 hours or more.

Figure 6: The port scanner Nmap needed nearly 10 hours to scan the entire port range with Portspoof running in service emulation mode. Typically, Nmap would complete such a scan in one or two minutes.