802.11b DoS exploit

From: Mark Osborne (mark@loud-fat-bloke.co.uk)
Date: Tue Mar 11 2003 - 14:26:32 PST

  • Next message: nesumin: "[Opera 7/6] Long Filename Buffer Overflow Vulnerability in Download"

    
     ('binary' encoding is not supported, stored as-is)
    While working to develop code for WIDZ that is equivalent to a standard 
    Intrusion Detection system’s RESET or SHUN functionality, an effective 
    802.11b disruption of service attack has been discovered.  I haven’t 
    spotted any other postings so here we go….
    
    FATA-jack - a modified version of the Wlan-jack, Fata-jack sends an 
    Authentication-Failed packets (with a reason code of previous 
    authentication failed) to a Wireless client PC.  The source and 
    destination macs have been spoofed so as to appear to come from the Access-
    point.  The original Wlan-jack code rate of transmission has been 
    significantly reduced to a meagre rate of 1 every 2.5 seconds, so as to 
    avoid any flood effect.
    
    In limited tests on multiple operating systems including Windows98, 
    Windows ME and Linux, FATA-jack effectively tears down any active session 
    and in many cases causing the client driver or client software to fail 
    requiring a reboot.
    
    Apart from being an extremely lethal DoS attack, FATA-jack is significant 
    for a number of reasons:
    
    -As the transmission rate is very low, it is easy to see how a low-spec PC 
    and a standard 802.11 card could  disable a large wireless network.
    
    -As the malevolent packet are sent directly to the client these will not 
    picked-up by logging functionality on the AP (if you have any) – this 
    highlights the need for Wireless IDS.
    
    -As the malevolent packets are spoofed AND sent directly to client MAC 
    protection or WEP protection will not prevent it.  
    
    -Some workmates have suggested that it could be used to cause IVs/WEP keys 
    to be cycled.  This would significantly reduce the time for a WEP cracking 
    exercise. This is yet to be verified.
    



    This archive was generated by hypermail 2b30 : Tue Mar 11 2003 - 15:03:25 PST