RE: Windows 2000 Static arp not static

From: shannong (shannong@texas.net)
Date: 02/16/03

  • Next message: Knud Erik Højgaard: "Re: Unhappy face icon on NT 4 workstation"
    From: "shannong" <shannong@texas.net>
    To: "'Anthony Kim'" <Anthony.Kim@VW.COM>, <focus-ms@securityfocus.com>
    Date: Sun, 16 Feb 2003 10:08:32 -0600
    
    

    The MAC address table mappings on switches have absolutely no effect on
    this. The switch still sees the offending machine as having the correct
    MAC address and the victim as having the correct MAC address. This
    exploit works due to the ARP cache poisoning of the victim as discussed
    in this thread.

    You prevent this from happening like you do other exploits. Use an IDS.
    One that detects these ARP flip-flops.

    -Shannon

    -----Original Message-----
    From: Anthony Kim [mailto:Anthony.Kim@VW.COM]
    Sent: Thursday, February 13, 2003 12:43 PM
    To: focus-ms@securityfocus.com
    Subject: Re: Windows 2000 Static arp not static

    On Thu, Feb 13, 2003, Tim Habex wrote:

    > Dear all,
    >
    > I am quite new to this. I posted this on bugtraq first, but
    > David Ahmad asked to post it in FOCUS-MS and vuln-dev. So here
    > I go :o)
    >
    > This is the setup : 1 Windows 2000 Professional (SP3) 1 Linux
    > Slackware (gateway) 1 Debian Linux 1 switch
    >
    > (The linux distro's doesn't really matter)
    >
    > When using ethercap on the network from de Debian machine, I
    > was able to see and control all trafic. (nothing new right?)
    > Ethercap is doing this by making the network believe everything
    > should be sent to the MAC-address of the ethercap machine which
    > in my case was the Debian machine.

    > To prevent this behaviour, I setup static routes both on the
    > gateway and the Windows machine. Yet I didn't get the result I
    > was expecting. I was still able to see packets on the Debian
    > machine, yet I was no longer able to control the packets.

    Because arp happens before routing, I'm not sure how much static
    routes will get you, unless you meant static arp entries.

    > When I looked at the arp cache of Linux, the static entry was
    > there and working (?), but on the Windows machine, THE VALUE OF
    > THE STATIC ARP WAS CHANGED. When ethercap was disabled, the
    > static arp entry was returned to the original value.

    I wouldn't be surprised if this was still true. (Won't test here
    at work ;-)

    It was a deficiency back in the NT days that caused all sorts of
    problems for software firewalls requiring an arp proxy.
    (Checkpoint anyone?)

    > Meaning Windows 2000 desktops (and servers?) can always be
    > sniffed even when using a switch. On top of that, your network
    > is probably vulnerable to the man-in-the-middle attacks if
    > you're relying on MS-technology only. I don't know if they are
    > still vulnerable to a man-in-the-middle attack if you're using
    > eg. a Linux router with static routes. My "hacking" knowlege is
    > quite limited. But I can imagine there are people who know how
    > to gain from this "feature".

    Most people would lock arp tables on the switch and not on the
    host. If you're relying on MS-technology only, you probably have
    a boatload of other problems to take care of... ;-)



    Relevant Pages

    • RE: gratuitous arp and bad mac
      ... Are you implementing any Layer 2 Switch Fault Tolerance? ... public network only but also NOT recommened in a cluster. ... > I looked at the arp table and found that the mac address for ... > sql-a was now matching the mac for node2. ...
      (microsoft.public.windows.server.clustering)
    • mac address issue
      ... I'm seeing two different mac addresses ... it's arp cache, but the switch plugged into that port doesn't. ...
      (Security-Basics)
    • Re: gratuitous arp and bad mac
      ... Teamed NICs are to seperate switches, but the team is in a failover ... so the 2nd switch should be out of the picture. ... >> I looked at the arp table and found that the mac address for ... >> sql-a was now matching the mac for node2. ...
      (microsoft.public.windows.server.clustering)
    • Re: Sending "magic packets" from OBSD router seems to fail
      ... > through the switch, not the OBSD box. ... > wakeonlan seems to hit the internal ethernet device (attached to the ... the ARP table does not have any entry for this IP. ... its own MAC address. ...
      (comp.unix.bsd.openbsd.misc)
    • RE: [fw-wiz] Link level security with static arp tables
      ... Would it be easier to solve this at a switch? ... So a VLAN could isolate the traffic. ... I could disable arp on eht0 and use static arp tables in the gw, ... will not be able to find it's MAC. ...
      (Firewall-Wizards)