- The Easiest Way to Get in
- A Short Fence to Climb: Bypassing Closed ESSIDs, MAC, and Protocols Filtering
- Picking a Trivial Lock: Various Means of Cracking WEP
- Picking the Trivial Lock in a Less Trivial Way: Injecting Traffic to Accelerate WEP Cracking
- Field Observations in WEP Cracking
- Cracking TKIP: The New Menace
- The Frame of Deception: Wireless Man-in-the-Middle Attacks and Rogue Access Points Deployment
- Breaking the Secure Safe
- The Last Resort: Wireless DoS Attacks
A Short Fence to Climb: Bypassing Closed ESSIDs, MAC, and Protocols Filtering
Let us explore slightly more protected WLANs. How about so-called closed networks? ESSID makes a bad shared secret. The reason is that it is not removed from all management frames. For example, reauthenticate and reassociate frames will contain the ESSID value. Thus, a network with roaming hosts will not benefit from the closed ESSIDs at all and sending a deauthenticate frame to one or more hosts on the closed WLAN is easy:
arhontus:~# ./essid_jack -h Essid Jack: Proof of concept so people will stop calling an ssid a password. Usage: ./essid_jack -b <bssid> [ -d <destination mac> ] [ -c <channel number> ] [ -i <interface name> ] -b: bssid, the mac address of the access point (e.g. 00:de:ad:be:ef:00) -d: destination mac address, defaults to broadcast address. -c: channel number (1-14) that the access point is on, defaults to current. -i: the name of the AirJack interface to use (defaults to aj0). arhontus:~# essid_jack -b 00:02:2d:ab:cd: -c 11 Got it, the essid is (escape characters are c style): "ArhOnt-X"
On a BSD platform, use the dinject-deauth utility from Wnet and sniff the passing traffic while using it.
Of course, such methodology will only work against a network with several reachable associated hosts present. In the rare case of a lonely access point, your best bet would be to guess the closed ESSID. It is surprising, but many users enable closed ESSID but do not change the actual ESSID value from the default (perhaps counting on the fact that it is not broadcasted anyway). Use the OUI, which is the first 3 bytes of the MAC address, to find out the access point manufacturer (see RFC 1700) and check the default ESSID values for the access points produced by this particular vendor and supporting closed ESSIDs. You can find these values and many other interesting facts in Appendix H.
MAC filtering is also trivial to bypass, even though we have seen some wi-fi inexperienced security consultants claiming it to be a good protection shame on you guys. Sniff the network traffic to determine which MAC addresses are present. When the host quits the network, assume it's MAC and associate. You can also change your MAC and IP address to the same values as those on the victim's host and coexist peacefully on the same (shared) network (piggybacking). Surely you would need to disable ARPs on your interface and go to Defcon 1 with your firewall. You would also have to be careful about what traffic you send out to the network to prevent the victim host from sending too many TCP resets and ICMP port unreachables, so their rare and megaexpensive knowledge-based IDS does not get triggered. You should try to restrict your communications to ICMP when communicating with the outside world. You can use any Loki-style ICMP-based backdoor (e.g., encapsulate data in echo replies or any other ICMP types that do not illicit responses). If you want to enjoy full network interoperability, you don't have to wait for the host to leave and can simply kick it out. Such action might lead to user complaints and an IDS alarm, in particular if WIDS is in place, but who cares, especially since you urgently need to check the latest posts at http://www.wi-foo.com. Therefore, try to use your common sense and pick a host that does not seem to generate any current traffic and send it a deassociate frame spoofing your MAC address as an access point. At the same time, have a second client card plugged in and configured with the MAC of a target host and other WLAN parameters to associate. It is a race condition that you are going to win, because no one can stop you from flooding the spoofed host with deassociate frames continuously. To flood the host with deassociate frames from Linux you can use wlan_jack:
arhontus:~# ./wlan_jack -h Wlan Jack: 802.11b DOS attack. Usage: ./wlan_jack -b <bssid> [ -v <victum address> ] [ -c <channel number> ] [ -i <interface name> ] -b: bssid, the mac address of the access point (e.g. 00:de:ad:be:ef:00) -v: victim mac address, defaults to broadcast address. -c: channel number (1-14) that the access point is on, defaults to current. -i: the name of the AirJack interface to use (defaults to aj0). arhontus:~# ./wlan_jack -b 00:02:2d:ab:cd: -v 00:05:5D:F9:ab:cd -c 11 Wlan Jack: 802.11 DOS utility. Jacking Wlan...
Alternatively, you can employ File2air. If running HostAP drivers, you can launch Void11 or craft your own frames with Libwlan. Another way of flooding the host with deassociate frames is using Mike Schiffman's omerta utility under HostAP and employing the Libradiate library. In this book we do not describe Libradiate, because it ceased to be supported more than a year ago and at the moment omerta is probably the only tool worth mentioning here that employs Libradiate. On the OpenBSD platform you can employ the dinject-disas utility, perhaps run from a simple looping shell script. Finally, a different way of launching very efficient DoS attacks with AirJack is using fata_jack. Please consult the wireless DoS attacks section at the end of this chapter to learn more about it.
Just to remind you how to change a MAC address when you need it:
# ifconfig wlan0 hw ether DE:AD:BE:EF:CO:DE (Linux ifconfig) # ip link set dev wlan0 address DE:AD:BE:EF:CO:DE (Linux iproute) # ifconfig wi0 ether DEADBEEFCODE (FreeBSD) # sea -v wi0 DE:AD:BE:EF:CO:DE (OpenBSD)
Sea is a separate utility that does not come with OpenBSD but can be found at http://www.openbsd.org.
Protocol filtering is harder to bypass. Unfortunately for system administrators and fortunately for attackers, very few access points on the market implement proper protocol filtering and they tend to be high-end, expensive devices. Also, protocol filtering applies only to a few specific situations in which user activity is limited to a narrow set of actions, for example, browsing a corporate site through HTTPS or sending e-mails via Secure Multipurpose Internet Mail Extensiosn (S/MIME) from PDAs given to employees for these aims specifically. SSH port forwarding might help, but you have to be sure that both sides support SSHv2.
The main attacks against networks protected by protocol filtering are attacks against the allowed secure protocol (which might not be as secure as it seems). Good examples of such insecurity are well-known attacks against SSHv1 implemented in Dug Song's Dsniff by the sshow and sshmitm utilities. Whereas sshow can help an attacker disclose some useful information about the bypassing SSH traffic (e.g., the authentication attempts or length of transmitted passwords or commands with both SSHv1 and SSHv2 traffic), sshmitm is a powerful man-in-the-middle for SSHv1 utility that allows SSHv1 password login capture and connection hijacking attacks. Unfortunately, although the majority of complete networked operational systems currently support SSHv2, SSHv1 is often the only choice available to log in to routers, some firewalls, and other networking devices and this is still preferable to telnet or rlogin. On wired networks, traffic redirection via DNS spoofing is necessary for sshmitm to work. However, Layer 2 monkey_jack-style man-in-the-middle attacks can successfully replace DNS spoofing on 802.11 links, leaving fewer traces in the network IDS logs unless a proper wireless IDS is implemented (which is rarely the case).
The creator of Dsniff did not leave HTTPS without attention as well. webmitm can transparently proxy and sniff HTTPS traffic to capture most of the "secure" SSL-encrypted Web mail logins and Web site form submissions. Again, dnsspoof traffic redirection for webmitm can be substituted by a wireless-specific man-in-the-middle attack, raising fewer system administrators' eyebrows. Another remarkable man-in-the-middle tool specifically designed for attacking various SSL connections (HTTPS, IMAPS, etc.) is Omen. Just like webmitm, more information on using Omen follows in the next chapter.
If network designers and management decided to rely on SSH, HTTPS, and so on as their main line of defense and did not implement lower-layer encryption and proper mutual authentication (e.g., 802.1x/EAP-TLS or better), you might not even have to attack Layer 6 security protocols. Nothing would stop a cracker from associating with the target network, running a quick nmap scan, and launching an attack against the discovered sshd (e.g., using sshnuke to exploit the CRC32 vulnerability, if you want to be as 1337 as Trinity). Of course, the real-life CRC32 bug was patched eons ago, but new sshd vulnerabilities tend to appear on a regular basis. As for HTTPS security, the latest CGI vulnerability scanners support HTTPS (e.g., Nikto with the -ssl option) and in the majority of cases the difference in exploitation of the discovered CGI holes over the HTTPS protocol is limited to changing the target port to 443 from 80 or piping data through stunnel.
Finally, a desperate cracker can always resort to brute force. There are a variety of utilities and scripts for SSH brute forcing: guess-who, ssh-crack, ssh-brute.sh, 55hb_v1.sh, and so on. With SSL-protected Web logins you can try the php-ssl-brute script. Although brute forcing leaves telltale multiple login signs in the logs, wireless attackers might be unconcerned, as it is more difficult to locate and prosecute a cracker on a WLAN anyway. Although brute force is both time and battery power consuming for a mobile wireless attacker, if it is the only choice available, someone will eventually give it a try and perhaps succeed.