RFC 7359

Internet Engineering Task Force (IETF)                           F. Gont
Request for Comments: 7359                           Huawei Technologies
Category: Informational                                      August 2014
ISSN: 2070-1721

     Layer 3 Virtual Private Network (VPN) Tunnel Traffic Leakages
                      in Dual-Stack Hosts/Networks


   The subtle way in which the IPv6 and IPv4 protocols coexist in
   typical networks, together with the lack of proper IPv6 support in
   popular Virtual Private Network (VPN) tunnel products, may
   inadvertently result in VPN tunnel traffic leakages.  That is,
   traffic meant to be transferred over an encrypted and integrity-
   protected VPN tunnel may leak out of such a tunnel and be sent in the
   clear on the local network towards the final destination.  This
   document discusses some scenarios in which such VPN tunnel traffic
   leakages may occur as a result of employing IPv6-unaware VPN
   software.  Additionally, this document offers possible mitigations
   for this issue.

Status of This Memo

   This document is not an Internet Standards Track specification; it is
   published for informational purposes.

   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).  Not all documents
   approved by the IESG are a candidate for any level of Internet
   Standard; see 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

Gont                          Informational                     [Page 1]

RFC 7359                  VPN Traffic Leakages               August 2014


   This document describes a problem of information leakage in VPN
   software and attributes that problem to the software's inability to
   deal with IPv6.  We do not think this is an appropriate
   characterization of the problem.  It is true that when a device
   supports more than one address family, the inability to apply policy
   to more than one address family on that device is a defect.  Despite
   that, inadvertent or maliciously induced information leakage may also
   occur due to the existence of any unencrypted interface allowed on
   the system, including the configuration of split tunnels in the VPN
   software itself.  While there are some attacks that are unique to an
   IPv6 interface, the sort of information leakage described by this
   document is a general problem that is not unique to the situation of
   IPv6-unaware VPN software.  We do not think this document
   sufficiently describes the general issue.

Copyright Notice

   Copyright (c) 2014 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.

Gont                          Informational                     [Page 2]

RFC 7359                  VPN Traffic Leakages               August 2014

Table of Contents

   1.  Introduction  . . . . . . . . . . . . . . . . . . . . . . . .   3
   2.  Terminology . . . . . . . . . . . . . . . . . . . . . . . . .   4
   3.  IPv4 and IPv6 Coexistence . . . . . . . . . . . . . . . . . .   5
   4.  Virtual Private Networks in IPv4/IPv6 Dual-Stack
       Hosts/Networks  . . . . . . . . . . . . . . . . . . . . . . .   6
   5.  Inadvertent VPN Tunnel Traffic Leakages in Legitimate
       Scenarios . . . . . . . . . . . . . . . . . . . . . . . . . .   7
   6.  VPN Tunnel Traffic Leakage Attacks  . . . . . . . . . . . . .   7
   7.  Mitigations to VPN Tunnel Traffic Leakage Vulnerabilities . .   8
     7.1.  Fixing VPN Client Software  . . . . . . . . . . . . . . .   8
     7.2.  Operational Mitigations . . . . . . . . . . . . . . . . .  10
   8.  Security Considerations . . . . . . . . . . . . . . . . . . .  10
   9.  Acknowledgements  . . . . . . . . . . . . . . . . . . . . . .  10
   10. References  . . . . . . . . . . . . . . . . . . . . . . . . .  10
     10.1.  Normative References . . . . . . . . . . . . . . . . . .  10
     10.2.  Informative References . . . . . . . . . . . . . . . . .  11

1.  Introduction

   It is a very common practice for users to employ VPN software when
   employing a public (and possibly rogue) local network.  This is
   typically done not only to gain access to remote resources that may
   not otherwise be accessible from the public Internet, but also to
   secure the host's traffic against attackers that might be connected
   to the same local network as the victim host.  The latter case
   constitutes the problem space of this document.  Indeed, it is
   sometimes assumed that employing a VPN tunnel makes the use of
   insecure protocols (e.g., that transfer sensitive information in the
   clear) acceptable, as a VPN tunnel provides security services (such
   as data integrity and/or confidentiality) for all communications made
   over it.  However, this document illustrates that under certain
   circumstances, some traffic might not be mapped onto the VPN tunnel
   and thus might be sent in the clear on the local network.

   Many VPN products that are typically employed for the aforementioned
   VPN tunnels only support the IPv4 protocol: that is, they perform the
   necessary actions such that IPv4 traffic is sent over the VPN tunnel,
   but they do nothing to secure IPv6 traffic originated from (or being
   received at) the host employing the VPN client.  However, the hosts
   themselves are typically dual-stacked: they support (and enable by
   default) both IPv4 and IPv6 (even if such IPv6 connectivity is simply
   "dormant" when they connect to IPv4-only networks).  When the IPv6
   connectivity of such hosts is enabled, they may end up employing an
   IPv6-unaware VPN client in a dual-stack network.  This may have
   "unexpected" consequences, as explained below.

Gont                          Informational                     [Page 3]

RFC 7359                  VPN Traffic Leakages               August 2014

   The subtle way in which the IPv4 and IPv6 protocols interact and
   coexist in dual-stacked networks might, either inadvertently or as a
   result of a deliberate attack, result in VPN tunnel traffic leakages
   -- that is, traffic meant to be transferred over a VPN tunnel could
   leak out of the VPN tunnel and be transmitted in the clear from the
   local network to the final destination, without employing the VPN
   services at all.

   Since this issue is specific to VPN solutions that route Layer 3
   traffic, it is applicable to the following types of VPN technologies:

   o  IPsec-based VPN tunnels

   o  SSL/TLS VPN tunnels

      NOTE: see Section 2 for a definition and description of these

   Section 2 clarifies the terminology employed throughout this
   document.  Section 3 provides some background about IPv6 and IPv4
   coexistence, summarizing how IPv6 and IPv4 interact on a typical
   dual-stacked network.  Section 4 describes the underlying problem
   that leads to the aforementioned VPN traffic leakages.  Section 5
   describes legitimate scenarios in which such traffic leakages might
   occur, while Section 6 describes how VPN traffic leakages can be
   triggered by deliberate attacks.  Finally, Section 7 discusses
   possible mitigations for the aforementioned issue.

2.  Terminology

   When employing the term "Virtual Private Network tunnel" (or "VPN
   tunnel"), this document refers to VPN tunnels based on IPsec or SSL/
   TLS, where Layer 3 packets are encapsulated and sent from a client to
   a middlebox, to access multiple network services (possibly employing
   different transport and/or application protocols).

   IPsec-based VPN tunnels simply employ IPsec in tunnel mode to
   encapsulate and transfer Layer 3 packets over the VPN tunnel.  On the
   other hand, the term "SSL/TLS-based VPN tunnels" warrants a
   clarification, since two different technologies are usually referred
   to as "SSL/TLS VPN":

   SSL/TLS VPN tunnel:
      A technology that encapsulates traffic from a client to a
      middlebox.  In essence, an SSL/TLS VPN tunnel acts just like an
      IPsec-based tunnel, but instead employs SSL/TLS for encryption and
      some proprietary/unspecified mechanism for encapsulation and

Gont                          Informational                     [Page 4]

RFC 7359                  VPN Traffic Leakages               August 2014

   SSL/TLS VPN portal:
      A front-end provided by the middlebox to add security to a
      normally unsecured site.  An SSL/TLS VPN portal is typically
      application specific, wrapping the specific protocol, such as
      HTTP, to provide access to specific services on a network.  In
      such a case, the SSL/TLS VPN portal would be accessed just like
      any HTTPS URL.  SSL/TLS VPN portals are used when one wants to
      restrict access and only provide remote users to very specific
      services on the network.  SSL/TLS VPN portals do not require an
      agent, and the policy is typically more liberal from organizations
      allowing access from anywhere via this access method.  All other
      traffic on the system may be routed directly to the destination,
      whether it is IPv4, IPv6, or even other service level (HTTP)
      destination addresses.

   Our document focuses on Layer 3 VPNs that employ IPsec-based or SSL/
   TLS-based tunnels, and excludes the so-called SSL/TLS VPN portals.
   Both IPsec-based and SSL/TLS-based VPN tunnels are designed to route
   Layer 3 traffic via the VPN tunnel through to the VPN tunnel server.
   Typically, these solutions are agent based, meaning that software is
   required on the client endpoint to establish the VPN tunnel and
   manage access control or routing rules.  This provides an opportunity
   for controls to be managed through that application as well as on the
   client endpoint.

      NOTE: Further discussion of SSL-based VPNs can be found in

   We note that, in addition to the general case of "send all traffic
   through the VPN", this document considers the so-called "split-
   tunnel" case, where some subset of the traffic is sent through the
   VPN, while other traffic is sent to its intended destination with a
   direct routing path (i.e., without employing the VPN tunnel).  We
   note that many organizations will prevent split-tunneling in their
   VPN configurations if they would like to make sure the users data
   goes through a gateway with protections (malware detection, URL
   filtering, etc.), but others are more interested in performance of
   the user's access or the ability for researchers to have options to
   access sites they may not be able to through the gateway.

3.  IPv4 and IPv6 Coexistence

   The coexistence of the IPv4 and IPv6 protocols has a number of
   interesting and subtle aspects that may have "surprising"
   consequences.  While IPv6 is not backwards-compatible with IPv4, the
   two protocols are "glued" together by the Domain Name System (DNS).

Gont                          Informational                     [Page 5]

RFC 7359                  VPN Traffic Leakages               August 2014

   For example, consider a site (say, www.example.com) that has both
   IPv4 and IPv6 support.  The corresponding domain name
   (www.example.com, in our case) will contain both A and AAAA DNS
   resource records (RRs).  Each A record will contain one IPv4 address,
   while each AAAA record will contain one IPv6 address -- and there
   might be more than one instance of each of these record types.  Thus,
   when a dual-stacked client application means to communicate with
   www.example.com, it can request both A and AAAA records and use any
   of the available addresses.  The preferred address family (IPv4 or
   IPv6) and the specific address that will be used (assuming more than
   one address of each family is available) varies from one protocol
   implementation to another, with many host implementations preferring
   IPv6 addresses over IPv4 addresses.

      NOTE: [RFC6724] specifies an algorithm for selecting a destination
      address from a list of IPv6 and IPv4 addresses.  [RFC6555]
      discusses the challenge of selecting the most appropriate
      destination address, along with a proposed implementation approach
      that mitigates connection-establishment delays.

   As a result of this "coexistence" between IPv6 and IPv4, when a dual-
   stacked client means to communicate with some other system, the
   availability of A and AAAA DNS resource records will typically affect
   which protocol is employed to communicate with that system.

4.  Virtual Private Networks in IPv4/IPv6 Dual-Stack Hosts/Networks

   Many VPN tunnel implementations do not support the IPv6 protocol --
   or, what is worse, they completely ignore IPv6.  This typically means
   that, once a VPN tunnel has been established, the VPN software takes
   care of the IPv4 connectivity by, e.g., inserting an IPv4 default
   route that causes all IPv4 traffic to be sent over the VPN tunnel (as
   opposed to sending the traffic in the clear, employing the local
   router).  However, if IPv6 is not supported (or completely ignored),
   any packets destined to an IPv6 address will be sent in the clear
   using the local IPv6 router.  That is, the VPN software will do
   nothing about the IPv6 traffic.

   The underlying reason for which these VPN leakages may occur is that,
   for dual-stacked systems, it is not possible to secure the
   communication with another system without securing both protocols
   (IPv6 and IPv4).  Therefore, as long as the traffic for one of these
   protocols is not secured, there is the potential for VPN traffic

Gont                          Informational                     [Page 6]

RFC 7359                  VPN Traffic Leakages               August 2014

5.  Inadvertent VPN Tunnel Traffic Leakages in Legitimate Scenarios

   Consider a dual-stacked host that employs IPv4-only VPN software to
   establish a VPN tunnel with a VPN server, and that said host now
   connects to a dual-stacked network (that provides both IPv6 and IPv4
   connectivity).  If some application on the client means to
   communicate with a dual-stacked destination, the client will
   typically query both A and AAAA DNS resource records.  Since the host
   will have both IPv4 and IPv6 connectivity, and the intended
   destination will have both A and AAAA DNS resource records, one of
   the possible outcomes is that the host will employ IPv6 to
   communicate with the intended destination.  Since the VPN software
   does not support IPv6, the IPv6 traffic will not employ the VPN
   tunnel; hence, it will have neither integrity nor confidentiality
   protection from the source host to the final destination.

   This could inadvertently expose sensitive traffic that was assumed to
   be secured by the VPN software.  In this particular scenario, the
   resulting VPN tunnel traffic leakage is a side effect of employing
   IPv6-unaware VPN software in a dual-stacked host/network.

6.  VPN Tunnel Traffic Leakage Attacks

   A local attacker could deliberately trigger IPv6 connectivity on the
   victim host by sending forged ICMPv6 Router Advertisement messages
   [RFC4861].  Such packets could be sent by employing standard software
   such as rtadvd [RTADVD], or by employing packet-crafting tools such
   as SI6 Network's IPv6 Toolkit [SI6-Toolkit] or THC's IPv6 Attack
   Toolkit [THC-IPv6].  Once IPv6 connectivity has been enabled,
   communications with dual-stacked systems could result in VPN tunnel
   traffic leakages, as previously described.

   While this attack may be useful enough (due to the increasing number
   of IPv6-enabled sites), it will only lead to traffic leakages when
   the destination system is dual-stacked.  However, it is usually
   trivial for an attacker to trigger such VPN tunnel traffic leakages
   for any destination system: an attacker could simply advertise
   himself as the local recursive DNS server by sending forged Router
   Advertisement messages [RFC4861] that include the corresponding
   Recursive DNS Server (RDNSS) option [RFC6106], and then perform a DNS
   spoofing attack such that he can become a "man in the middle" and
   intercept the corresponding traffic.  As with the previous attack
   scenario, packet-crafting tools such as [SI6-Toolkit] and [THC-IPv6]
   can readily perform this attack.

Gont                          Informational                     [Page 7]

RFC 7359                  VPN Traffic Leakages               August 2014

      NOTE: Some systems are known to prefer IPv6-based recursive DNS
      servers over IPv4-based ones; hence, the "malicious" recursive DNS
      servers would be preferred over the legitimate ones advertised by
      the VPN server.

7.  Mitigations to VPN Tunnel Traffic Leakage Vulnerabilities

   At the time of this writing, a large number of VPN implementations
   have not yet addressed the issue of VPN tunnel traffic leakages.
   Most of these implementations simply ignore IPv6; hence, IPv6 traffic
   leaks out of the VPN tunnel.  Some VPN tunnel implementations handle
   IPv6, but not properly.  For example, they may be able to prevent
   inadvertent VPN tunnel traffic leakages arising in legitimate dual-
   stack networks, but they may fail to properly handle the myriad of
   vectors available to an attacker for injecting "more specific
   routes", such as ICMPv6 Router Advertisement messages with Prefix
   Information Options and/or Route Information Options, and ICMPv6
   Redirect messages.

   Clearly, the issue of VPN tunnel traffic leakages warrants proper
   IPv6 support in VPN tunnel implementations.

7.1.  Fixing VPN Client Software

   There are a number of possible mitigations for the VPN tunnel traffic
   leakage vulnerability discussed in this document.

   If the VPN client is configured by administrative decision to
   redirect all IPv4 traffic to the VPN, it should:

   1.  If IPv6 is not supported in the VPN software, disable IPv6
       support in all network interfaces.

          NOTE: For IPv6-unaware VPN clients, the most simple mitigation
          would be to disable IPv6 support in all network interface
          cards when a VPN tunnel is meant to be employed.  Thus,
          applications on the host running the VPN client software will
          have no other option than to employ IPv4; hence, they will
          simply not even try to send/process IPv6 traffic.  We note
          that this should only be regarded as a temporary workaround,
          since the proper mitigation would be to correctly handle IPv6

Gont                          Informational                     [Page 8]

RFC 7359                  VPN Traffic Leakages               August 2014

   2.  If IPv6 is supported in the VPN software, ensure that all IPv6
       traffic is also sent via the VPN.

          NOTE: This would imply, among other things, properly handling
          any vectors that might be employed by an attacker to install
          IPv6 routes at the victim system (such as ICMPv6 Router
          Advertisement messages with Prefix Information Options or
          Route information Options [RFC4191], ICMPv6 Redirect messages,
          etc.).  We note that properly handling all the aforementioned
          vectors may prove to be nontrivial.

   If the VPN client is configured to only send a subset of IPv4 traffic
   to the VPN tunnel (split-tunnel mode), then:

   1.  If the VPN client does not support IPv6, it should disable IPv6
       support in all network interfaces.

          NOTE: As noted above, this should only be regarded as a
          temporary workaround, since the proper mitigation would be to
          correctly handle IPv6 traffic.

   2.  If the VPN client supports IPv6, it is the administrators
       responsibility to ensure that the correct corresponding sets of
       IPv4 and IPv6 networks get routed into the VPN tunnel.

          NOTE: As noted above, this would imply, among other things,
          properly handling any vectors that might be employed by an
          attacker to install IPv6 routes at the victim system.  This
          may prove to be a nontrivial task.

   A network may prevent local attackers from successfully performing
   the aforementioned attacks against other local hosts by implementing
   First-Hop Security solutions such as Router Advertisement Guard
   (RA-Guard) [RFC6105] and DHCPv6-Shield [DHCPv6-SHIELD].  However, for
   obvious reasons, a host cannot and should not rely on this type of
   mitigations when connecting to an open network (cybercafe, etc.).

      NOTE: Besides, popular implementations of RA-Guard are known to be
      vulnerable to evasion attacks [RFC7113].

   Finally, we note that if (eventually) IPv6-only VPN implementations
   become available, similar issues to the ones discussed in this
   document could arise if these IPv6-only VPN implementations do
   nothing about the IPv4 traffic.

Gont                          Informational                     [Page 9]

RFC 7359                  VPN Traffic Leakages               August 2014

7.2.  Operational Mitigations

   While the desired mitigation for the issues discussed in this
   document is for VPN clients to be IPv6 aware, we note that in
   scenarios where this would be unfeasible, an administrator may want
   to disable IPv6 connectivity on all network interfaces of the node
   employing the IPv6-unaware VPN client.

8.  Security Considerations

   This document discusses how traffic meant to be transferred over a
   VPN tunnel can leak out of such a tunnel and, hence, appear in the
   clear on the local network.  This is the result of employing
   IPv6-unaware VPN client software on dual-stacked hosts.

   The proper mitigation of this issue is to correctly handle IPv6
   traffic in the VPN client software.  However, in scenarios in which
   such a mitigation is unfeasible, an administrator may choose to
   disable IPv6 connectivity on all network interfaces of the host
   employing the VPN client.

9.  Acknowledgements

   The author would like to thank Kathleen Moriarty and Paul Hoffman who
   contributed text that was readily incorporated into Section 2 of this

   The author of this document would like to thank (in alphabetical
   order) Cameron Byrne, Spencer Dawkins, Gert Doering, Stephen Farrell,
   Seth Hall, Paul Hoffman, Tor Houghton, Russ Housley, Joel Jaeggli,
   Alastair Johnson, Merike Kaeo, Panos Kampanakis, Stephen Kent, Henrik
   Lund Kramshoj, Warren Kumari, Barry Leiba, Kathleen Moriarty, Thomas
   Osterried, Jim Small, Martin Vigoureux, and Andrew Yourtchenko for
   providing valuable comments on earlier draft versions of this

   The author wishes to express deep and heartfelt gratitude to Enrique
   Garcia and Vicenta Tejedo, for their precious love and support.

10.  References

10.1.  Normative References

   [RFC4191]  Draves, R. and D. Thaler, "Default Router Preferences and
              More-Specific Routes", RFC 4191, November 2005.

Gont                          Informational                    [Page 10]

RFC 7359                  VPN Traffic Leakages               August 2014

   [RFC4861]  Narten, T., Nordmark, E., Simpson, W., and H. Soliman,
              "Neighbor Discovery for IP version 6 (IPv6)", RFC 4861,
              September 2007.

   [RFC6106]  Jeong, J., Park, S., Beloeil, L., and S. Madanapalli,
              "IPv6 Router Advertisement Options for DNS Configuration",
              RFC 6106, November 2010.

   [RFC6555]  Wing, D. and A. Yourtchenko, "Happy Eyeballs: Success with
              Dual-Stack Hosts", RFC 6555, April 2012.

   [RFC6724]  Thaler, D., Draves, R., Matsumoto, A., and T. Chown,
              "Default Address Selection for Internet Protocol Version 6
              (IPv6)", RFC 6724, September 2012.

10.2.  Informative References

              Gont, F., Liu, W., and G. Van de Velde, "DHCPv6-Shield:
              Protecting Against Rogue DHCPv6 Servers", Work in
              Progress, July 2014.

   [RFC6105]  Levy-Abegnoli, E., Van de Velde, G., Popoviciu, C., and J.
              Mohacsi, "IPv6 Router Advertisement Guard", RFC 6105,
              February 2011.

   [RFC7113]  Gont, F., "Implementation Advice for IPv6 Router
              Advertisement Guard (RA-Guard)", RFC 7113, February 2014.

   [RTADVD]   "rtadvd(8) manual page", <http://www.freebsd.org/cgi/

              SI6 Networks, "SI6 Networks' IPv6 Toolkit",

   [SSL-VPNs] Hoffman, P., "SSL VPNs: An IETF Perspective", IETF 72,
              SAAG Meeting, 2008,

   [THC-IPv6] The Hacker's Choice, "THC-IPV6 - attacking the IPV6
              protocol suite", <http://www.thc.org/thc-ipv6/>.

Gont                          Informational                    [Page 11]

RFC 7359                  VPN Traffic Leakages               August 2014

Author's Address

   Fernando Gont
   Huawei Technologies
   Evaristo Carriego 2644
   Haedo, Provincia de Buenos Aires  1706

   Phone: +54 11 4650 8472
   EMail: fgont@si6networks.com
   URI:   http://www.si6networks.com

Gont                          Informational                    [Page 12]