Re: ARP hole in Windows NT/2000

From: Keith Simonsen (bangelat_private)
Date: Sat Nov 24 2001 - 07:39:18 PST

  • Next message: Nelson Brito: "Re: ARP hole in Windows NT/2000"

    Hello,
    
    
    I came across this problem awhile back too... the main point I think is he
    used "arp -s" which should create a permanent entry in the arp table.
    
    ettercap probably floods the lan with gratuitous arp requests, so it can steal
    the gateway's ip address. If NT had a functional arp cache, the entry set
    with arp -s should not be changed.
    
    {root@blak 10:07am} ~# arp -S 192.168.1.1 0:60:8:af:8c:e4                       
    {root@blak 10:07am} ~# arp -an
    ? (192.168.1.1) at 0:60:8:af:8c:e4 permanent [ethernet]
    
    I ran arpspoof (from dsniff pkg) for awhile against that host
    {bangel@nomad 11:17am} /home/bangel# arpspoof -i ep0 -t 192.168.1.2 nomad       
    from another machine, the arp entry never changed, which is how it should
    be.
    
    I walked up stairs to a windows95 machine, did this:
    C:\>arp -s 192.168.1.1 00-60-08-af-8c-e4
    C:\>arp -a
    Interface: 192.168.1.3 on Interface 2
      Internet Address      Physical Address      Type
      192.168.1.1           00-60-08-af-8c-e4     static
    
    I then ran arpspoof against the 95 machine. The supposedly static entry
    changed immediately to the machine i was trying to spoof.
    
    C:\>arp -a
    Interface: 192.168.1.3 on Interface 2
      Internet Address      Physical Address      Type
      192.168.1.1           00-a0-c9-89-16-4a     static
    
    I repeated the same thing on win2000 SP1 I had on a laptop here... Same
    results.
    
    Awhile back, a friend and I tested many platforms against this bug, using
    both spoofed arp replies and spoofed gratuitious arp requests. Unfortunately
    I can't find our results, but I do remember that all versions of Windows
    we tested were vulnerable to changing static arp entries w/ spoofed arp 
    replies.
    
    Thanks
    Keith
    
    On 23-Nov-2001, Tomas Nybrand IT wrote:
    > Hi
    > 
    > Well ARP poisoning canīt be considered as something new, and I would
    > prefer to call it a vulnerability in the ARP protocol rather than a
    > windows vulnerability.
    > 
    > ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
    >  Tomas Nybrand - UNIX Administrator
    > ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
    >     --   Bene qui latuit, bene vixit.   --
    > 
    > Grzegorz.Flakat_private writes:
    > >Hi,
    > >
    > >I am not sure, if it is something new, but I think I found serious 
    > >vulnerability in ARP implementation in WindowsNT/2000 (I checked it on 
    > >NT4 SP6 and Win2000 SP1). The problem is when somebody whant to use "man 
    > >in the middle" technik to evesdrop your traffic. This example was done 
    > >with ettercap.
    > >To fill protect I use 'arp -s' to specify correct MAC for default 
    > >geteway. So I had :
    > >  10.10.1.4             00-b0-64-49-1e-01     static
    > >
    > >then I use ettercap to capture my traffic to the gateway. Ofcourse I 
    > >could see my POP3 pass ;) Then I checked arp table once again:
    > >
    > >  10.10.1.4             00-01-02-23-85-e1     static
    > >
    > >The MAC is different (this is MAC of my linux box). I checked the same 
    > >on Solaris 2.7 and Linux 2.4.8 and they look unvulnerable.
    > >Is this already known vulnerabilty (I found indication of similar 
    > >weakness, but that was on Windows 9x).
    > >
    > >Any suggestions how to get rid off that.
    > >
    > >Reagards
    > 
    



    This archive was generated by hypermail 2b30 : Sat Nov 24 2001 - 15:21:27 PST