[CORE-2003-12-05] DCE RPC Vulnerabilities New Attack Vectors Analysis

From: Core Security Technologies (advisories_at_coresecurity.com)
Date: 12/11/03

  • Next message: Kris Kennaway: "Re: A new TCP/IP blind data injection technique?"
    Date: Thu, 11 Dec 2003 12:34:42 -0300
    
    

                           Core Security Technologies Advisory
                               http://www.coresecurity.com

                   DCE RPC Vulnerabilities New Attack Vectors Analysis

    Date Published: 2003-12-10

    Last Update: 2003-12-10

    Advisory ID: CORE-2003-12-05

    Title: DCE RPC Vulnerabilities New Attack Vectors Analysis

    Remotely Exploitable: Yes

    Locally Exploitable: Yes

    Advisory URL:
      http://www.coresecurity.com/common/showdoc.php?idx=393&idxseccion=10

    Vendors contacted:
      - Microsoft
        . Core notification: 2003-12-09
        . Notification acknowledged by Microsoft: 2003-12-09

    Release Mode: USER RELEASE

    *Vulnerability Description:*

      Core Security Technologies researchers discovered new attack vectors for
      recently published vulnerabilities in Microsoft Windows operating systems.

      These new attack methods were found while researching exploitation
    conditions
      for the Workstation Service vulnerability discovered by eEye Digital
    Security and
      disclosed in Microsoft security bulletin MS03-049 of November 11th, 2003.

      They might also apply to other vulnerabilities such as the DCE RPC DCOM
      and the Messenger service vulnerabilities addressed by bulletins MS03-001,
      MS03-026 and MS03-043.

      We found that by combining three protocol characteristics common to the
      vulnerabilities mentioned, an attacker can devise more severe,
    stealthy and
      low-noise attack vectors than those originally concieved. This creates
    the opportunity
      for malicious software to compromise large numbers of vulnerable
      systems in a massive scale, much like the Blaster and Slammer worms
    that caused
      great damage earlier in 2003.

      Core Security Technologies urges users of Microsoft Windows operating
    systems to
      deploy the available patches for these vulnerabilities as they
    effectively fix
      the problem. Suggested workarounds should be revisited to ensure that
    they address
      all currently known attack vectors properly (including the new ones
    disclosed
      in this advisory).

    *Vulnerable Packages:*

      Microsoft RPC services running on Windows 2000 and Windows XP.

    *Solution/Vendor Information/Workaround:*

      Patches are readily available to fix the vulnerabilities and close all
    known
      attack vectors.

      See Microsoft Security Bulletins [MS03-001], [MS03-026], [MS03-043],
    [MS03-049]:

      http://www.microsoft.com/technet/security/bulletin/MS03-001.asp
      http://www.microsoft.com/technet/security/bulletin/MS03-026.asp
      http://www.microsoft.com/technet/security/bulletin/MS03-043.asp
      http://www.microsoft.com/technet/security/bulletin/MS03-049.asp

    *Credits:*

      This advisory was researched and prepared by Javier Kohen and Juliano
      Rizzo from Core Security Technologies.

      jkohen@coresecurity.com
      juliano@coresecurity.com

      The vulnerabilities mentioned herein where discovered independently by
      David Litchfield of Next Generation Security Software, the Last Stage
      of Delirum (LSD) Research Group, and Yuji Ukay, Barnaby Jack and
      Riley Hasell of eEye Digital Security.

    *Technical Description - Exploit/Concept Code:*

      In recent months, several vulnerabilities in the Microsoft RPC code
      (see [MS03-001], [MS03-026], [MS03-043], [MS03-049]) have been disclosed.

      The RPC vulnerabilities account started back in July when LSD disclosed a
      severe security hole in the DCOM service. Since then, different
    workarounds
      were discussed on several security mailing lists but doubt persisted
    as to which
      RPC protocol sequences were potential attack gates. This is, obviously, an
      important factor in determining which ports should be filtered to
    prevent remote
      attacks and how other workarounds should be deployed.

      We have researched three protocol characteristics which when used together
      provide the attacker with new severe, stealthy and low-noise attack
      vectors.

      We were able to successfully exploit some of the latest DCE RPC
    vulnerabilities
      through less noted ports and even on broadcast addresses.

      The following sections provide more specific details about these
    attack vectors.

      [Some RPC services listen on high ports.]

      This is a little-known feature that has been omitted from the
    published alerts
      to date. The importance of this issue lies in the fact that the most
    common
      filtering rules used on current firewall configurations will allow
    incoming traffic
      on these ports.

      For instance, the latest Workstation Service vulnerability can be
    exploited on a
      high TCP/UDP port. Usually the TCP port is 1025 and the UDP port is
    higher,
      depending on system settings.

      [Some RPC services listen to broadcast traffic.]

      Further tests showed that those attacks conducted with datagram protocols
      (like UDP) could be targeted to broadcast addresses and still succeed.
      Our first tests required performing a two-way handshake with each host
      that responded to the broadcasted query. This situation seems to confuse
      the native Windows 2000 RPC implementation, so we suspected that this
      kind of attack can not be built with the stock implementation.

      [The idempotent bit.]

      RPC has an interesting feature that allows the client to avoid the
      two-way handshake customary to datagram protocols. This can be enabled
      by turning on the idempotent flag in RPCv4 request packets. This not
      only reduces the traffic needed to perform the attack, but it also
    makes it possible
      to spoof the request's source. This handshake involves a 20-byte
      secret number, apparently not easily guessable, that can be avoided by
      setting the idempotent flag.

      We were able to exploit [MS03-026] using 445/TCP 139/TCP 135/TCP 135/UDP
      and 80/TCP. [MS03-049] can be successfully exploited through 445/TCP
      139/TCP and dynamically assigned TCP/UDP ports over 1024. We have not
    seen
      public exploits or worms using those ports, and we are not sure
    whether the
      Windows API can be bent for this purpose. We used our own DCE RPC
      implementation which is part of the publicly available Impacket
      project. Presumably [MS03-043] (Messenger service) can be exploited
      using the same techniques, but we haven't attempted an attack, although
      third party reports describe messenger service attacks using UDP
    broadcasts
      in the wild.

      [Firewall bypassing.]

      Since the attack can be conducted over the UDP
      protocol and that it can be spoofed, it is easy to bypass common
      filtering rules. Some personal firewalls enable the blocking of
    traffic on an
      application basis, but some of the vulnerable services actually run
      inside the same application that does the DNS resolution. This can be
      used to the attacker's advantage to reach the vulnerable targets by
    spoofing
      the attack packets as if they came from a legitimate server sending
      DNS responses back to DNS clients on vulnerable workstations.

      It is common to see filtering rules like the following:

      allow UDP packets from DNSSERVER port 53 to WORKSTATION port above 1024

      The outlined attack vector will pass through the above rule and succeed.
      Even personal firewall rules that specify an application will allow these
      attacks to pass:

      allow UDP packets from DNSSERVER port 53 to WORKSTATION application
      services.exe

      [Conclusions.]

      Patches for the vulnerabilities mentioned have already been made available
      by Microsoft. Installing them will effectively fix the bugs and close all
      attack vectors discovered herein.
      Workarounds should be revisited to ensure they properly cover these
      attack vectors.
      As a general conclusion, we recommend careful inspection of Windows
    Service
      vulnerabilities in order to identify potential avenues of attack
    related to
      these services providing RPC endpoints that listen to UDP and TCP traffic
      on high ports.

      [Appendix.]

      [DCE RPC protocol sequences.]

      A protocol sequence is a "character string that represents a valid
      combination of an RPC protocol (such as ncacn), a transport protocol
      (such as TCP), and a network protocol (such as IP)" (see [MSDN]).

      [What protocol sequences are available?]

      A standard Windows installation has default services accessible
      through many protocol sequences. For example, the Workstation Service
      can be accessed by means of the following:

      ncacn_ip_tcp:[####]
      ncacn_np:[\\PIPE\wkssvc]
      ncadg_ip_udp:[####]

      Notes: #### is a port number above 1024. Datagram based sequences
      (like ncadg_ip_udp) are also accessible through the broadcast address.
      Named pipes (strings like ncacn_np) can be contacted in several ways, via
      TCP ports 139, 445, 593 and 80.

      [How easy is it to build an attack over an alternative transport?]

      Starting from a working attack to an RPC service it is trivial to adapt
      it to work over other protocol sequences. Of course the attacker must
      have a DCE RPC implementation that allows her to use her choice of
      transport, here is where Impacket fits perfectly into the task as
      changing transports requires no additional effort.

    ===IMPACKET EXAMPLE=======

    class ExploitPacket(ImpactPacket.Header):
          OP_NUM = 0x1B # Interface's method number
          def get_header_size(self):
              return 0
          def __init__(self, aBuffer = None):
              ImpactPacket.Header.__init__(self, 0)

    class Attack:
          def do(self,host):
              # The next two lines could be changed to use different protocol
    sequences.
              #port = 445
              #stringbinding = "ncacn_np:%s[\\pipe\\service_np]" % host # SMB
    over IP/TCP on port 445
              port = 135
              stringbinding = "ncacn_ip_tcp:%s[%d]" % (host,port) # IP/TCP
    transport on the specified port
              exploitStub = ImpactPacket.Data()
              exploitStub.set_bytes_from_string(exploitBytes)
              exploitPacket = ExploitPacket()
              exploitPacket.contains(exploitStub)
              rpcTransport = transport.DCERpcTransportFactory(stringbinding)
              rpcTransport.set_dport(port)
              # Uncomment for UDP protocols:
              #rpcConn = impacket.dcerpc.dcerpc_v4.DCERPC_v4(rpcTransport)
              rpcConn = impacket.dcerpc.dcerpc.DCERPC_v5(rpcTransport)
              rpcConn.connect()
              rpcConn.bind(SERVICE_UUID) # 20-byte UUID (including version)
              rpcConn.send(exploitPacket)
              #( ... exploit specific code ...)
              rpcConn.disconnect()

    ==========================

      [Retrieving the list of RPC endpoints.]

      Core Security Technologies provides, as part of its free, open source
      Impacket package (downloadable from http://oss.coresecurity.com/ ),
      a tool that allows remote enumeration of RPC of services listening
      and their assigned port numbers and supported transports.

      This code is platform independent Python. A similar tool (RPCDUMP)
      is available from Microsoft.

    *References:*

      [Core Security Technologies - Impacket tools]
      http://oss.coresecurity.com/

      [Microsoft]

      MS03-001
      http://www.microsoft.com/technet/security/bulletin/MS03-001.asp

      MS03-026
      http://www.microsoft.com/technet/security/bulletin/MS03-026.asp

      MS03-043
      http://www.microsoft.com/technet/security/bulletin/MS03-043.asp

      MS03-049
      http://www.microsoft.com/technet/security/bulletin/MS03-049.asp

      MSDN reference to endpoints and protocol sequences
      http://msdn.microsoft.com/library/en-us/midl/midl/endpoint.asp

      [eEye Digital Security]

      Windows Workstation Service Remote Buffer Overflow
      http://www.eeye.com/html/Research/Advisories/AD20031111.html

      Microsoft RPC Heap Corruption Vulnerability - Part II
      http://www.eeye.com/html/Research/Advisories/AD20030910.html

      [Last Stage of Delirum Research Group]
      http://lsd-pl.net/files/get?WINDOWS/win32_dcom

      [Next Generation SEcurity Software]
      http://www.nextgenss.com/research.html

    *About Core Security Technologies*

      Core Security Technologies develops strategic security solutions for
      Fortune 1000 corporations, government agencies and military
      organizations. The company offers information security software and
      services designed to assess risk and protect and manage information
      assets.
      Headquartered in Boston, MA, Core Security Technologies can be reached
      at 617-399-6980 or on the Web at http://www.coresecurity.com.

      To learn more about CORE IMPACT, the first comprehensive penetration
      testing framework, visit:
      http://www.coresecurity.com/products/coreimpact

    *DISCLAIMER:*

      The contents of this advisory are copyright (c) 2003 CORE Security
      Technologies and may be distributed freely provided that no fee is
      charged for this distribution and proper credit is given.

    $Id: DCE\040RPC\040new\040attack\040vectors.txt,v 1.1 2003/12/10
    20:21:16 carlos Exp $


  • Next message: Kris Kennaway: "Re: A new TCP/IP blind data injection technique?"

    Relevant Pages