Re: Cisco CSS 11000 Series DoS

From: Mike Caudill (mcaudill_at_cisco.com)
Date: 09/08/03

  • Next message: GreyMagic Software: "RE: BAD NEWS: Microsoft Security Bulletin MS03-032"
    Date: Sun, 7 Sep 2003 18:13:08 -0400
    To: S21SEC <vul-serv@s21seccom.s21sec.com>
    
    

    -----BEGIN PGP SIGNED MESSAGE-----
    Hash: SHA1

    Hello S21Sec,

    Apologies for our delay in response, but we have been thoroughly testing
    and investigating this issue to ensure our response is accurate.

    Normally, it is not our policy to comment on customer cases, however, since
    this has been posted as a Security Advisory that contains inaccurate
    information, we must respond with the correct details.

    It appears that the original customer case was possibly misdiagnosed,
    leading to the incorrect information in your advisory. Hence the details
    on the mechanics of this attack are incorrect. We will only correct the
    defect identifier, upgrade information, affected platform information and
    workaround details here.

    The upgrade listed in the solution section of the advisory will not provide
    a solution to the problem stated in your advisory. We have been able to
    reproduce a reload given the instructions in your advisory ONLY on the
    11800 platform with a heavy storm of TCP SYN packets sent to the circuit
    address of the CSS. This problem has been documented in CSCec01994. We
    are working on delivering a fix for the specific problem (CSCec01994) into
    the next 5.0 and 6.10 maintenance releases which will be available shortly,
    possibly by the end of the month.

    This problem is seen on the 5.0.2.03 and 6.10 Build 4 versions and is
    specific to the 11800 platform. It does *not* affect the 11150 and 11050
    platforms.

    Using ACLs on an upstream router to protect the circuit address is
    recommended as a prevention measure, or workaround. For example, the command

    access-list 116 deny tcp any <circuit address of CSS>

    can be used on an upstream router in combination with applying the
    access-group to an outgoing interface to deny TCP to circuit addresses on
    the CSS.

    Thanks much for posting this information, although working with the Cisco
    PSIRT in the future on advisories will eliminate this type of confusion and
    inaccurate information.

    We do greatly appreciate the opportunity to work with researchers on
    security vulnerabilities, and welcome the opportunity to review and assist
    with Product Security Advisories. Our ultimate goal is to ensure that
    customers have accurate information on which to base upgrade and workaround
    decisions and we welcome partnership with researchers towards that goal.

    Thanks,

    - -Mike-

    -----BEGIN PGP SIGNATURE-----
    Version: PGP 6.5.2

    iQA/AwUBP1ujXopjyUnrvVJxEQIQtgCgww98VayXf99hahKND7Cwa4GNVdAAn09Z
    BzpgOjLY1Lh5dLTuNiLy5Bsj
    =N+45
    -----END PGP SIGNATURE-----

    > S21SEC <vul-serv@s21seccom.s21sec.com> [2003-08-07 18:30] wrote:
    > ###############################################################
    > ID: S21SEC-025-en
    > Title: Cisco CSS 11000 Series DoS
    > Date: 04/07/2003
    > Status: Solution available
    > Scope: Interruption of service, high CPU load.
    > Platforms: All/Chassis CS800.
    > Author: ecruz, egarcia, jandre
    > Location: http://www.s21sec.com/en/avisos/s21sec-025-en.txt
    > Release: External
    > ###############################################################
    >
    > S 2 1 S E C
    >
    > http://www.s21sec.com
    >
    > Cisco CSS 11000 Series Denial of service
    >
    >
    >
    > Description of vulnerability
    > ----------------------------
    >
    > A heavy storm of TCP SYN packets directed to the circuit address of the
    > CSS
    > can cause DoS on it, high cpu load or even sudden reboots.
    >
    > The issue is known by cisco as the ONDM Ping failure (CSCdz00787). On the
    > CS800 chassis the
    > system controller module (SCM) sends ONDM (online diagnostics monitor)
    > pings to each SFP card
    > in order to see if they are alive, if the SCM doesn't get a response in
    > about 30 seconds the
    > SCM will reboot the CS800 and there will be no core.
    >
    > By attacking the circuit IP address of the CSS with SYN packets the
    > traffic is sent up to the SCM
    > over the internal MADLAN ethernet interface. If this internal interface
    > becomes overloaded
    > the ONDM ping request and response traffic can be dropped leading this to
    > an internal DoS
    > since no internal comunications are available.
    >
    > Any attacker could do this externally with a few sessions of NMAP and a
    > cable/ADSL internet
    > connection.
    >
    >
    > Affected Versions and platforms
    > -------------------------------
    >
    > This vulnerability affects the models 11800, 11150 and 11050 with chassis
    > CS800.
    >
    >
    > Solution
    > --------
    >
    > Upgrade to software release WebNS 5.00.110s or above.
    > http://www.cisco.com/en/US/products/hw/contnetw/ps789/prod_release_note0918
    > 6a008014ee04.html
    >
    > AcL's to protect the circuit address are recomended.
    >
    >
    > Additional information
    > ----------------------
    >
    > These vulnerabilities have been found and researched by:
    >
    > Eduardo Cruz ecruz@s21sec.com
    > Emilin Garcia egarcia@s21sec.com
    > Jordi Andre jandre@s21sec.com
    >
    > You can find the last version of this warning in:
    >
    > http://www.s21sec.com/en/avisos/s21sec-025-en.txt
    >
    > And other S21SEC warnings in http://www.s21sec.com/en/avisos/
    >
    > [ ----- End of Included Message ----- ]

    -- 
    ----------------------------------------------------------------------------
    |      ||        ||       | Mike Caudill           | mcaudill@cisco.com    |
    |      ||        ||       | PSIRT Incident Manager | +1.919.392.2855       |
    |     ||||      ||||      | DSS PGP: 0xEBBD5271    | +1.919.522.4931 (cell)|
    | ..:||||||:..:||||||:..  | RSA PGP: 0xF482F607    ------------------------|
    | C i s c o S y s t e m s | http://www.cisco.com/go/psirt                  |
    ----------------------------------------------------------------------------
    

  • Next message: GreyMagic Software: "RE: BAD NEWS: Microsoft Security Bulletin MS03-032"

    Relevant Pages

    • mercilessly co-ordinate their depressed spokesman
      ... respectively advisory, so I drop you. ... Do not deserve the qualifications ... Are you established, I mean, constituting in response to comprehensive ...
      (sci.crypt)
    • Re: 12v relay on 3v ??
      ... >> circuit examples and cannot find 14-pin circuit examples. ... allowed to keep it ON until it falls back to 146 Hz (4400 RPM.) ... a faster response and lower ripple/hysteresis both suggest ... The ripple calculation suggests ...
      (sci.electronics.basics)
    • Re: Resolve sub-millivolts with a PIC?
      ... I don't want to build a circuit board with ... good ground plane and complementary I/O to null out ground current ... Is the step response with 100mW of power. ... your PIC breadboard must be much better than mine. ...
      (sci.electronics.design)
    • Re: Convolving the Displacement
      ... Thanks for your response, and thanks also to Greg and Andy. ... a cautionary note that the fundamental approximation in circuit ... > electrons everywhere in the circuit, which, with some crude hack of ... photons on the standard assumptions of causality. ...
      (sci.physics)
    • Re: Horowitz-Hill: Serious scholarly query
      ... > matter by email but saw no response to date. ... > particular remarkable circuit design, for example Figure 9.73 of the ... > striking about it was how closely it resembles the circuit published, ... > by Michael Workman of IBM. ...
      (sci.electronics.design)