Teredo Security Updates

Internet Engineering Task Force (IETF)                         D. Thaler
Request for Comments: 5991                                     Microsoft
Updates: 4380                                                S. Krishnan
Category: Standards Track                                       Ericsson
ISSN: 2070-1721                                              J. Hoagland
                                                                Symantec
                                                          September 2010


                        Teredo Security Updates

Abstract

   The Teredo protocol defines a set of flags that are embedded in every
   Teredo IPv6 address.  This document specifies a set of security
   updates that modify the use of this flags field, but are backward
   compatible.

Status of This Memo

   This is an Internet Standards Track document.

   This document is a product of the Internet Engineering Task Force
   (IETF).  It represents the consensus of the IETF community.  It has
   received public review and has been approved for publication by the
   Internet Engineering Steering Group (IESG).  Further information on
   Internet Standards is available in Section 2 of RFC 5741.

   Information about the current status of this document, any errata,
   and how to provide feedback on it may be obtained at
   http://www.rfc-editor.org/info/rfc5991.

Copyright Notice

   Copyright (c) 2010 IETF Trust and the persons identified as the
   document authors.  All rights reserved.

   This document is subject to BCP 78 and the IETF Trust's Legal
   Provisions Relating to IETF Documents
   (http://trustee.ietf.org/license-info) in effect on the date of
   publication of this document.  Please review these documents
   carefully, as they describe your rights and restrictions with respect
   to this document.  Code Components extracted from this document must
   include Simplified BSD License text as described in Section 4.e of
   the Trust Legal Provisions and are provided without warranty as
   described in the Simplified BSD License.

   This document may contain material from IETF Documents or IETF
   Contributions published or made publicly available before November
   10, 2008.  The person(s) controlling the copyright in some of this
   material may not have granted the IETF Trust the right to allow
   modifications of such material outside the IETF Standards Process.
   Without obtaining an adequate license from the person(s) controlling
   the copyright in such materials, this document may not be modified
   outside the IETF Standards Process, and derivative works of it may
   not be created outside the IETF Standards Process, except to format
   it for publication as an RFC or to translate it into languages other
   than English.

Table of Contents

   1. Introduction ....................................................2
   2. Terminology .....................................................3
   3. Specification ...................................................4
      3.1. Random Address Flags .......................................4
      3.2. Deprecation of Cone Bit ....................................6
   4. Security Considerations .........................................7
   5. Acknowledgments .................................................7
   6. References ......................................................8
      6.1. Normative References .......................................8
      6.2. Informative References .....................................8
   Appendix A.  Implementation Status .................................9
   Appendix B.  Resistance to Address Prediction ......................9

1.  Introduction

   Teredo [RFC4380] defines a set of flags that are embedded in every
   Teredo IPv6 address.  This document specifies a set of security
   updates that modify the use of this flags field, but are backwards
   compatible.  This document updates RFC 4380.

   The Flags field in a Teredo IPv6 address has 13 unused bits out of a
   total of 16 bits.  To guard against address-scanning risks [RFC5157]
   from malicious users, this update randomizes 12 of the 13 unused bits
   when configuring the Teredo IPv6 address.  Even if an attacker were
   able to determine the external (mapped) IPv4 address and port
   assigned by a NAT to the Teredo client, the attacker would still need
   to attack a range of 4,096 IPv6 addresses to determine the actual
   Teredo IPv6 address of the client.

   The cone bit in a Teredo IPv6 address indicates whether a peer needs
   to send Teredo control messages before communicating with a Teredo
   IPv6 address.  Unfortunately, it may also have some value in terms of
   profiling to the extent that it reveals the security posture of the
   network.  If the cone bit is set, an attacker may decide it is
   fruitful to port-scan the embedded external IPv4 address and others
   associated with the same organization, looking for open ports.
   Deprecating the cone bit prevents the a priori revelation of the
   security posture of the NAT.

2.  Terminology

   This document uses the following terminology, for consistency with
   [RFC4380].

   Cone NAT: A NAT that maps all requests from the same internal IP
      address and port to the same external IP address and port.
      Furthermore, any external host can send a packet to the internal
      host by sending a packet to the mapped external address and port.

   Indirect Bubble: A Teredo control message that is sent to another
      Teredo client via the destination's Teredo server, as specified in
      [RFC4380], Section 5.2.4.

   Local Address/Port: The IPv4 address and UDP port from which a Teredo
      client sends Teredo packets.  The local port is referred to as the
      Teredo service port in [RFC4380].  The local address of a node may
      or may not be globally routable because the node can be located
      behind one or more NATs.

   Mapped Address/Port: A global IPv4 address and a UDP port that
      results from the translation of a node's own local address/port by
      one or more NATs.  The node learns these values through the Teredo
      protocol specified in [RFC4380].  The mapped address/port can be
      different for every peer with which a node tries to communicate.

   Network Address Translation (NAT): The process of converting between
      IP addresses used within an intranet or other private network and
      Internet IP addresses.

   Peer: A Teredo client with which another Teredo client needs to
      communicate.

   Port-Preserving NAT: A NAT that translates a local address/port to a
      mapped address/port such that the mapped port has the same value
      as the local port, as long as that same mapped address/port has
      not already been used for a different local address/port.

   Public Address: An external global address used by a NAT.

   Restricted NAT: A NAT where all requests from the same internal IP
      address and port are mapped to the same external IP address and
      port.  Unlike the cone NAT, an external host can send packets to
      an internal host (by sending a packet to the external mapped
      address and port) only if the internal host has first sent a
      packet to the external host.

   Teredo Client: A node that implements the client parts of [RFC4380],
      has access to the IPv4 Internet, and wants to gain access to the
      IPv6 Internet.

   Teredo IPv6 Address: An IPv6 address that starts with the prefix
      2001:0000:/32 and is formed as specified in Section 4 of
      [RFC4380].

   Teredo Server: A node that has a globally routable address on the
      IPv4 Internet, and is used as a helper to provide IPv6
      connectivity to Teredo clients.

   The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
   "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
   document are to be interpreted as described in RFC 2119 [RFC2119].

3.  Specification

3.1.  Random Address Flags

   Teredo addresses are structured, and some of the fields contained in
   them are fairly predictable.  This makes the addresses themselves
   easier to predict and opens up a vulnerability.

   Teredo prefix:  This field is 32 bits and has a single IANA-assigned
      value.

   Server:  This field is 32 bits and is set to the server in use.  The
      server to use is generally statically configured on the client.
      This means that overall entropy of the server field will be low,
      i.e., that the server will not be hard to predict.  Attackers
      could confine their guessing to the most popular server IP
      addresses.

   Flags:  The Flags field is 16 bits in length, but [RFC4380] provides
      for only one of these bits (the cone bit) to vary.

   Client port:  This 16-bit field corresponds to the external port
      number assigned to the client's Teredo service port.  Thus, the
      value of this field depends on two factors (the chosen Teredo
      service port and the NAT port assignment behavior), and it
      therefore is harder to predict the entropy this field will have.
      If clients tend to use a predictable port number and NATs are
      often port-preserving, then the port number can be rather
      predictable.

   Client IPv4 address:  This 32-bit field corresponds to the external
      IPv4 address the NAT has assigned for the client port.  In
      principle, this can be any address in the assigned part of the
      IPv4 unicast address space.  However, if an attacker is looking
      for the address of a specific Teredo client, they will have to
      have the external IPv4 address pretty well narrowed down.  Certain
      IPv4 address ranges could also become well known for having a
      higher concentration of Teredo clients, making it easier to find
      an arbitrary Teredo client.  These addresses could correspond to
      large organizations that allow Teredo, such as a university or
      enterprise, or to Internet Service Providers that only provide
      their customers with RFC 1918 addresses.

   Optimizations in scanning can also reduce the number of addresses
   that need to be checked.  For example, for addresses behind a cone
   NAT, it would likely be easy to probe if a specific port number is
   open on an IPv4 address, prior to trying to form a Teredo address for
   that address and port.

   Hence, the Flags field specified in [RFC4380], Section 4 is updated
   as follows:

                           1
       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |C|z|Random1|U|G|    Random2    |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

   C: This flag is specified in [RFC4380], and its use is modified in
      Section 3.2 below.

   z: This flag is reserved.  It MUST be set to zero when the address is
      constructed, as specified in [RFC4380].

   Random1: MUST be set to a random value.

   U: This flag is specified in [RFC4380].

   G: This flag is specified in [RFC4380].

   Random2: MUST be set to a random value.

3.2.  Deprecation of Cone Bit

   The qualification procedure is specified in [RFC4380], Section 5.2.1,
   and is modified as follows.  Teredo clients SHOULD completely skip
   the first phase of the qualification procedure and implement only the
   second phase where it uses the Teredo link-local address with the
   cone bit set to zero.  Consequently, a distinction between cone and
   restricted NATs can no longer be made.  Teredo communication will
   still succeed, but at the expense of forcing peers to skip case 4 of
   the sending details specified in [RFC4380], Section 5.2.4.  This will
   result in the same number of indirect bubbles being sent as if the
   other end were a peer behind a restricted NAT.  Even though the peer
   behind the cone NAT does not need these indirect bubbles, it replies
   to these indirect bubbles just like it would to any other indirect
   bubbles.  Skipping case 4 is already allowed for reliability reasons
   (as also specified in [RFC4380], Section 5.2.4), and hence this does
   not break interoperability, but the result of skipping the first
   phase of qualification is to force that behavior (which is less
   efficient, but potentially more reliable) to be taken by peers.

   In addition, clients and relays SHOULD ignore the cone bit in the
   address of a Teredo peer and treat it as if it were always clear, as
   specified in [RFC4380], Section 5.2.4 (last paragraph).

   Teredo servers MUST NOT ignore the cone bit for the following
   reasons.

   o  The cone bit in the IPv6 source address of a Router Solicitation
      (RS) from a client controls what IPv4 source address the server
      should use when sending a Router Advertisement (RA).  If this
      behavior is not preserved, legacy clients will conclude that they
      are behind a cone NAT even when they are not (because the client
      WILL receive the RA where previously it would not, since a cone
      bit set to 1 requires the server to respond from another IP
      address).  They will then set their cone bit and lose
      connectivity.

   o  When the Teredo server sends RAs (or bubbles if it's also a
      relay), the cone bit in its own Teredo address is set, indicating
      that it doesn't require bubbles to reach it.

4.  Security Considerations

   The basic threat model for Teredo is described in detail in
   [RFC4380], Section 7, but briefly, the goal is that a Teredo client
   should be as secure as if a host were directly attached to an
   untrusted Internet link.  This document specifies updates to
   [RFC4380] that improve the security of the base Teredo mechanism
   regarding specific threats.

   IPv6 address scanning [RFC5157] by off-path attackers: The Teredo
   IPv6 Address format defined in [RFC4380], Section 4 makes it
   relatively easy for a malicious user to conduct an address-scan to
   determine IPv6 addresses by guessing the external (mapped) IPv4
   address and port assigned to the Teredo client.  The random address
   bits guard against address-scanning risks by providing a range of
   4,096 IPv6 addresses per external IPv4 address/port.  As a result,
   even if a malicious user were able to determine the external (mapped)
   IPv4 address and port assigned to the Teredo client, the malicious
   user would still need to attack a range of 4,096 IPv6 addresses to
   determine the actual Teredo IPv6 address of the client.  Appendix B
   compares the address prediction resistance of a Teredo address
   following this specification to that of an address formed using
   standard IPv6 stateless address autoconfiguration [RFC4862].

   In order to prevent adversaries from easily guessing the values of
   the random bits and hence the address, the Random1 and Random2 bits
   in the Teredo Flags field MUST be constructed following the
   recommendations for random number generation as specified in
   [NIST-RANDOM] and [RFC4086].

   Opening a hole in an enterprise firewall [TUNNEL-SEC]: Teredo is NOT
   RECOMMENDED as a solution for networks that wish to implement strict
   controls for what traffic passes to and from the Internet.
   Administrators of such networks may wish to filter all Teredo traffic
   at the boundaries of their networks.

5.  Acknowledgments

   The authors would like to thank Remi Denis-Courmont, Fred Templin,
   Jordi Palet Martinez, James Woodyatt, Christian Huitema, Tom Yu, Jari
   Arkko, David Black, Tim Polk, and Sean Turner for reviewing earlier
   versions of this document and providing comments to make this
   document better.  The authors would also like to thank Alfred Hoenes
   for a careful review of this document.

6.  References

6.1.  Normative References

   [RFC2119]      Bradner, S., "Key words for use in RFCs to Indicate
                  Requirement Levels", BCP 14, RFC 2119, March 1997.

   [RFC4380]      Huitema, C., "Teredo: Tunneling IPv6 over UDP through
                  Network Address Translations (NATs)", RFC 4380,
                  February 2006.

6.2.  Informative References

   [NIST-RANDOM]  "NIST SP 800-90, Recommendation for Random Number
                  Generation Using Deterministic Random Bit Generators",
                  March 2007, <http://csrc.nist.gov/publications/
                  nistpubs/800-90/SP800-90revised_March2007.pdf>.

   [RFC4086]      Eastlake 3rd, D., Schiller, J., and S. Crocker,
                  "Randomness Requirements for Security", BCP 106,
                  RFC 4086, June 2005.

   [RFC4862]      Thomson, S., Narten, T., and T. Jinmei, "IPv6
                  Stateless Address Autoconfiguration", RFC 4862,
                  September 2007.

   [RFC5157]      Chown, T., "IPv6 Implications for Network Scanning",
                  RFC 5157, March 2008.

   [TUNNEL-SEC]   Hoagland, J., Krishnan, S., and D. Thaler, "Security
                  Concerns With IP Tunneling", Work in Progress, March
                  2010.

Appendix A.  Implementation Status

   Deprecation of the cone bit as specified in this document is
   implemented in Windows Vista and Windows Server 2008.

   The random flags specified in this document are implemented in
   Windows Vista SP1 and Windows Server 2008.

   All Windows implementations automatically disable Teredo if they
   detect that they are on a managed network with a domain controller.

Appendix B.  Resistance to Address Prediction

   This section compares the address prediction resistance of a Teredo
   address as compared to an address formed using IPv6 stateless address
   autoconfiguration (SLAAC) [RFC4862].

   Let's assume that the attacker knows a Teredo client's external IPv4
   address and Ethernet card's vendor.  Since the attacker knows the
   client's external IPv4 address, he does not have to search this
   space.  The attacker does not know the external port (16 bits) and
   the value of the random bits (12 bits), and he has to search this
   space.  This gives the attacker a total search space of 28 bits
   (16+12).  This compares very favorably with the 24 bits of search
   space required to find an address configured using SLAAC (when the
   Ethernet card's vendor is known) as described in Section 2.3 of
   [RFC5157].  Without the 12 random bits, the search space is limited
   to only 16 bits, and this is significantly worse than the 24 bits of
   search space provided by SLAAC.

   As the knowledge of the attacker decreases, the number of bits of
   search space in both cases is likely to increase in a relatively
   similar fashion.  The predictability of Teredo addresses will stay
   comparable to that of SLAAC addresses with the added 12 bits of
   search space, but will be significantly worse without the random
   bits.

Authors' Addresses

   Dave Thaler
   Microsoft Corporation
   One Microsoft Way
   Redmond, WA  98052
   USA

   Phone: +1 425 703 8835
   EMail: dthaler@microsoft.com


   Suresh Krishnan
   Ericsson
   8400 Decarie Blvd.
   Town of Mount Royal, QC
   Canada

   Phone: +1 514 345 7900 x42871
   EMail: suresh.krishnan@ericsson.com


   James Hoagland
   Symantec Corporation
   350 Ellis St.
   Mountain View, CA  94043
   USA

   EMail: Jim_Hoagland@symantec.com
   URI:   http://symantec.com/