RFC 3931






Network Working Group                                        J. Lau, Ed.
Request for Comments: 3931                              M. Townsley, Ed.
Category: Standards Track                                  Cisco Systems
                                                          I. Goyret, Ed.
                                                     Lucent Technologies
                                                              March 2005


           Layer Two Tunneling Protocol - Version 3 (L2TPv3)

Status of this Memo



   This document specifies an Internet standards track protocol for the
   Internet community, and requests discussion and suggestions for
   improvements.  Please refer to the current edition of the "Internet
   Official Protocol Standards" (STD 1) for the standardization state
   and status of this protocol.  Distribution of this memo is unlimited.

Copyright Notice



   Copyright (C) The Internet Society (2005).

Abstract



   This document describes "version 3" of the Layer Two Tunneling
   Protocol (L2TPv3).  L2TPv3 defines the base control protocol and
   encapsulation for tunneling multiple Layer 2 connections between two
   IP nodes.  Additional documents detail the specifics for each data
   link type being emulated.

Table of Contents



   1.  Introduction . . . . . . . . . . . . . . . . . . . . . . . . .  3
       1.1.  Changes from RFC 2661. . . . . . . . . . . . . . . . . .  4
       1.2.  Specification of Requirements. . . . . . . . . . . . . .  4
       1.3.  Terminology. . . . . . . . . . . . . . . . . . . . . . .  5
   2.  Topology . . . . . . . . . . . . . . . . . . . . . . . . . . .  8
   3.  Protocol Overview. . . . . . . . . . . . . . . . . . . . . . .  9
       3.1.  Control Message Types. . . . . . . . . . . . . . . . . . 10
       3.2.  L2TP Header Formats. . . . . . . . . . . . . . . . . . . 11
             3.2.1.  L2TP Control Message Header. . . . . . . . . . . 11
             3.2.2.  L2TP Data Message. . . . . . . . . . . . . . . . 12
       3.3.  Control Connection Management. . . . . . . . . . . . . . 13
             3.3.1.  Control Connection Establishment . . . . . . . . 14
             3.3.2.  Control Connection Teardown. . . . . . . . . . . 14
       3.4.  Session Management . . . . . . . . . . . . . . . . . . . 15
             3.4.1.  Session Establishment for an Incoming Call . . . 15
             3.4.2.  Session Establishment for an Outgoing Call . . . 15



Lau, et al.                 Standards Track                     [Page 1]

RFC 3931                         L2TPv3                       March 2005


             3.4.3.  Session Teardown . . . . . . . . . . . . . . . . 16
   4.  Protocol Operation . . . . . . . . . . . . . . . . . . . . . . 16
       4.1.  L2TP Over Specific Packet-Switched Networks (PSNs) . . . 16
             4.1.1.  L2TPv3 over IP . . . . . . . . . . . . . . . . . 17
             4.1.2.  L2TP over UDP. . . . . . . . . . . . . . . . . . 18
             4.1.3.  L2TP and IPsec . . . . . . . . . . . . . . . . . 20
             4.1.4.  IP Fragmentation Issues. . . . . . . . . . . . . 21
       4.2.  Reliable Delivery of Control Messages. . . . . . . . . . 23
       4.3.  Control Message Authentication . . . . . . . . . . . . . 25
       4.4.  Keepalive (Hello). . . . . . . . . . . . . . . . . . . . 26
       4.5.  Forwarding Session Data Frames . . . . . . . . . . . . . 26
       4.6.  Default L2-Specific Sublayer . . . . . . . . . . . . . . 27
             4.6.1.  Sequencing Data Packets. . . . . . . . . . . . . 28
       4.7.  L2TPv2/v3 Interoperability and Migration . . . . . . . . 28
             4.7.1.  L2TPv3 over IP . . . . . . . . . . . . . . . . . 29
             4.7.2.  L2TPv3 over UDP. . . . . . . . . . . . . . . . . 29
             4.7.3.  Automatic L2TPv2 Fallback. . . . . . . . . . . . 29
   5.  Control Message Attribute Value Pairs. . . . . . . . . . . . . 30
       5.1.  AVP Format . . . . . . . . . . . . . . . . . . . . . . . 30
       5.2.  Mandatory AVPs and Setting the M Bit . . . . . . . . . . 32
       5.3.  Hiding of AVP Attribute Values . . . . . . . . . . . . . 33
       5.4.  AVP Summary. . . . . . . . . . . . . . . . . . . . . . . 36
             5.4.1.  General Control Message AVPs . . . . . . . . . . 36
             5.4.2.  Result and Error Codes . . . . . . . . . . . . . 40
             5.4.3.  Control Connection Management AVPs . . . . . . . 43
             5.4.4.  Session Management AVPs. . . . . . . . . . . . . 48
             5.4.5.  Circuit Status AVPs. . . . . . . . . . . . . . . 57
   6.  Control Connection Protocol Specification. . . . . . . . . . . 59
       6.1.  Start-Control-Connection-Request (SCCRQ) . . . . . . . . 60
       6.2.  Start-Control-Connection-Reply (SCCRP) . . . . . . . . . 60
       6.3.  Start-Control-Connection-Connected (SCCCN) . . . . . . . 61
       6.4.  Stop-Control-Connection-Notification (StopCCN) . . . . . 61
       6.5.  Hello (HELLO). . . . . . . . . . . . . . . . . . . . . . 61
       6.6.  Incoming-Call-Request (ICRQ) . . . . . . . . . . . . . . 62
       6.7.  Incoming-Call-Reply (ICRP) . . . . . . . . . . . . . . . 63
       6.8.  Incoming-Call-Connected (ICCN) . . . . . . . . . . . . . 63
       6.9.  Outgoing-Call-Request (OCRQ) . . . . . . . . . . . . . . 64
       6.10. Outgoing-Call-Reply (OCRP) . . . . . . . . . . . . . . . 65
       6.11. Outgoing-Call-Connected (OCCN) . . . . . . . . . . . . . 65
       6.12. Call-Disconnect-Notify (CDN) . . . . . . . . . . . . . . 66
       6.13. WAN-Error-Notify (WEN) . . . . . . . . . . . . . . . . . 66
       6.14. Set-Link-Info (SLI). . . . . . . . . . . . . . . . . . . 67
       6.15. Explicit-Acknowledgement (ACK) . . . . . . . . . . . . . 67
   7.  Control Connection State Machines. . . . . . . . . . . . . . . 68
       7.1.  Malformed AVPs and Control Messages. . . . . . . . . . . 68
       7.2.  Control Connection States. . . . . . . . . . . . . . . . 69
       7.3.  Incoming Calls . . . . . . . . . . . . . . . . . . . . . 71
             7.3.1.  ICRQ Sender States . . . . . . . . . . . . . . . 72



Lau, et al.                 Standards Track                     [Page 2]

RFC 3931                         L2TPv3                       March 2005


             7.3.2.  ICRQ Recipient States. . . . . . . . . . . . . . 73
       7.4.  Outgoing Calls . . . . . . . . . . . . . . . . . . . . . 74
             7.4.1.  OCRQ Sender States . . . . . . . . . . . . . . . 75
             7.4.2.  OCRQ Recipient (LAC) States. . . . . . . . . . . 76
       7.5.  Termination of a Control Connection. . . . . . . . . . . 77
   8.  Security Considerations. . . . . . . . . . . . . . . . . . . . 78
       8.1.  Control Connection Endpoint and Message Security . . . . 78
       8.2.  Data Packet Spoofing . . . . . . . . . . . . . . . . . . 78
   9.  Internationalization Considerations. . . . . . . . . . . . . . 79
   10. IANA Considerations. . . . . . . . . . . . . . . . . . . . . . 80
       10.1. Control Message Attribute Value Pairs (AVPs) . . . . . . 80
       10.2. Message Type AVP Values. . . . . . . . . . . . . . . . . 81
       10.3. Result Code AVP Values . . . . . . . . . . . . . . . . . 81
       10.4. AVP Header Bits. . . . . . . . . . . . . . . . . . . . . 82
       10.5. L2TP Control Message Header Bits . . . . . . . . . . . . 82
       10.6. Pseudowire Types . . . . . . . . . . . . . . . . . . . . 83
       10.7. Circuit Status Bits. . . . . . . . . . . . . . . . . . . 83
       10.8. Default L2-Specific Sublayer bits. . . . . . . . . . . . 84
       10.9. L2-Specific Sublayer Type. . . . . . . . . . . . . . . . 84
       10.10 Data Sequencing Level. . . . . . . . . . . . . . . . . . 84
   11. References . . . . . . . . . . . . . . . . . . . . . . . . . . 85
       11.1. Normative References . . . . . . . . . . . . . . . . . . 85
       11.2. Informative References . . . . . . . . . . . . . . . . . 85
   12. Acknowledgments. . . . . . . . . . . . . . . . . . . . . . . . 87
   Appendix A: Control Slow Start and Congestion Avoidance. . . . . . 89
   Appendix B: Control Message Examples . . . . . . . . . . . . . . . 90
   Appendix C: Processing Sequence Numbers. . . . . . . . . . . . . . 91
   Editors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 93
   Full Copyright Statement . . . . . . . . . . . . . . . . . . . . . 94

1.  Introduction



   The Layer Two Tunneling Protocol (L2TP) provides a dynamic mechanism
   for tunneling Layer 2 (L2) "circuits" across a packet-oriented data
   network (e.g., over IP).  L2TP, as originally defined in RFC 2661, is
   a standard method for tunneling Point-to-Point Protocol (PPP)
   [RFC1661] sessions.  L2TP has since been adopted for tunneling a
   number of other L2 protocols.  In order to provide greater
   modularity, this document describes the base L2TP protocol,
   independent of the L2 payload that is being tunneled.

   The base L2TP protocol defined in this document consists of (1) the
   control protocol for dynamic creation, maintenance, and teardown of
   L2TP sessions, and (2) the L2TP data encapsulation to multiplex and
   demultiplex L2 data streams between two L2TP nodes across an IP
   network.  Additional documents are expected to be published for each
   L2 data link emulation type (a.k.a. pseudowire-type) supported by
   L2TP (i.e., PPP, Ethernet, Frame Relay, etc.).  These documents will



Lau, et al.                 Standards Track                     [Page 3]

RFC 3931                         L2TPv3                       March 2005


   contain any pseudowire-type specific details that are outside the
   scope of this base specification.

   When the designation between L2TPv2 and L2TPv3 is necessary, L2TP as
   defined in RFC 2661 will be referred to as "L2TPv2", corresponding to
   the value in the Version field of an L2TP header.  (Layer 2
   Forwarding, L2F, [RFC2341] was defined as "version 1".)  At times,
   L2TP as defined in this document will be referred to as "L2TPv3".
   Otherwise, the acronym "L2TP" will refer to L2TPv3 or L2TP in
   general.

1.1.  Changes from RFC 2661



   Many of the protocol constructs described in this document are
   carried over from RFC 2661.  Changes include clarifications based on
   years of interoperability and deployment experience as well as
   modifications to either improve protocol operation or provide a
   clearer separation from PPP.  The intent of these modifications is to
   achieve a healthy balance between code reuse, interoperability
   experience, and a directed evolution of L2TP as it is applied to new
   tasks.

   Notable differences between L2TPv2 and L2TPv3 include the following:

      Separation of all PPP-related AVPs, references, etc., including a
      portion of the L2TP data header that was specific to the needs of
      PPP.  The PPP-specific constructs are described in a companion
      document.

      Transition from a 16-bit Session ID and Tunnel ID to a 32-bit
      Session ID and Control Connection ID, respectively.

      Extension of the Tunnel Authentication mechanism to cover the
      entire control message rather than just a portion of certain
      messages.

   Details of these changes and a recommendation for transitioning to
   L2TPv3 are discussed in Section 4.7.

1.2.  Specification of Requirements



   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 [RFC2119].







Lau, et al.                 Standards Track                     [Page 4]

RFC 3931                         L2TPv3                       March 2005


1.3.  Terminology



   Attribute Value Pair (AVP)

      The variable-length concatenation of a unique Attribute
      (represented by an integer), a length field, and a Value
      containing the actual value identified by the attribute.  Zero or
      more AVPs make up the body of control messages, which are used in
      the establishment, maintenance, and teardown of control
      connections.  This basic construct is sometimes referred to as a
      Type-Length-Value (TLV) in some specifications.  (See also:
      Control Connection, Control Message.)

   Call (Circuit Up)

      The action of transitioning a circuit on an L2TP Access
      Concentrator (LAC) to an "up" or "active" state.  A call may be
      dynamically established through signaling properties (e.g., an
      incoming or outgoing call through the Public Switched Telephone
      Network (PSTN)) or statically configured (e.g., provisioning a
      Virtual Circuit on an interface).  A call is defined by its
      properties (e.g., type of call, called number, etc.) and its data
      traffic.  (See also: Circuit, Session, Incoming Call, Outgoing
      Call, Outgoing Call Request.)

   Circuit

      A general term identifying any one of a wide range of L2
      connections.  A circuit may be virtual in nature (e.g., an ATM
      PVC, an IEEE 802 VLAN, or an L2TP session), or it may have direct
      correlation to a physical layer (e.g., an RS-232 serial line).
      Circuits may be statically configured with a relatively long-lived
      uptime, or dynamically established with signaling to govern the
      establishment, maintenance, and teardown of the circuit.  For the
      purposes of this document, a statically configured circuit is
      considered to be essentially the same as a very simple, long-
      lived, dynamic circuit.  (See also: Call, Remote System.)

   Client

      (See Remote System.)

   Control Connection

      An L2TP control connection is a reliable control channel that is
      used to establish, maintain, and release individual L2TP sessions
      as well as the control connection itself.  (See also: Control
      Message, Data Channel.)



Lau, et al.                 Standards Track                     [Page 5]

RFC 3931                         L2TPv3                       March 2005


   Control Message

      An L2TP message used by the control connection.  (See also:
      Control Connection.)

   Data Message

      Message used by the data channel.  (a.k.a. Data Packet, See also:
      Data Channel.)

   Data Channel

      The channel for L2TP-encapsulated data traffic that passes between
      two LCCEs over a Packet-Switched Network (i.e., IP).  (See also:
      Control Connection, Data Message.)

   Incoming Call

      The action of receiving a call (circuit up event) on an LAC.  The
      call may have been placed by a remote system (e.g., a phone call
      over a PSTN), or it may have been triggered by a local event
      (e.g., interesting traffic routed to a virtual interface).  An
      incoming call that needs to be tunneled (as determined by the LAC)
      results in the generation of an L2TP ICRQ message.  (See also:
      Call, Outgoing Call, Outgoing Call Request.)

   L2TP Access Concentrator (LAC)

      If an L2TP Control Connection Endpoint (LCCE) is being used to
      cross-connect an L2TP session directly to a data link, we refer to
      it as an L2TP Access Concentrator (LAC).  An LCCE may act as both
      an L2TP Network Server (LNS) for some sessions and an LAC for
      others, so these terms must only be used within the context of a
      given set of sessions unless the LCCE is in fact single purpose
      for a given topology.  (See also: LCCE, LNS.)

   L2TP Control Connection Endpoint (LCCE)

      An L2TP node that exists at either end of an L2TP control
      connection.  May also be referred to as an LAC or LNS, depending
      on whether tunneled frames are processed at the data link (LAC) or
      network layer (LNS).  (See also: LAC, LNS.)

   L2TP Network Server (LNS)

      If a given L2TP session is terminated at the L2TP node and the
      encapsulated network layer (L3) packet processed on a virtual
      interface, we refer to this L2TP node as an L2TP Network Server



Lau, et al.                 Standards Track                     [Page 6]

RFC 3931                         L2TPv3                       March 2005


      (LNS).  A given LCCE may act as both an LNS for some sessions and
      an LAC for others, so these terms must only be used within the
      context of a given set of sessions unless the LCCE is in fact
      single purpose for a given topology.  (See also: LCCE, LAC.)

   Outgoing Call

      The action of placing a call by an LAC, typically in response to
      policy directed by the peer in an Outgoing Call Request.  (See
      also: Call, Incoming Call, Outgoing Call Request.)

   Outgoing Call Request

      A request sent to an LAC to place an outgoing call.  The request
      contains specific information not known a priori by the LAC (e.g.,
      a number to dial).  (See also: Call, Incoming Call, Outgoing
      Call.)

   Packet-Switched Network (PSN)

      A network that uses packet switching technology for data delivery.
      For L2TPv3, this layer is principally IP.  Other examples include
      MPLS, Frame Relay, and ATM.

   Peer

      When used in context with L2TP, Peer refers to the far end of an
      L2TP control connection (i.e., the remote LCCE).  An LAC's peer
      may be either an LNS or another LAC.  Similarly, an LNS's peer may
      be either an LAC or another LNS.  (See also: LAC, LCCE, LNS.)

   Pseudowire (PW)

      An emulated circuit as it traverses a PSN.  There is one
      Pseudowire per L2TP Session.  (See also: Packet-Switched Network,
      Session.)

   Pseudowire Type

      The payload type being carried within an L2TP session.  Examples
      include PPP, Ethernet, and Frame Relay.  (See also: Session.)

   Remote System

      An end system or router connected by a circuit to an LAC.






Lau, et al.                 Standards Track                     [Page 7]

RFC 3931                         L2TPv3                       March 2005


   Session

      An L2TP session is the entity that is created between two LCCEs in
      order to exchange parameters for and maintain an emulated L2
      connection.  Multiple sessions may be associated with a single
      Control Connection.

   Zero-Length Body (ZLB) Message

      A control message with only an L2TP header.  ZLB messages are used
      only to acknowledge messages on the L2TP reliable control
      connection.  (See also: Control Message.)

2.  Topology



   L2TP operates between two L2TP Control Connection Endpoints (LCCEs),
   tunneling traffic across a packet network.  There are three
   predominant tunneling models in which L2TP operates: LAC-LNS (or vice
   versa), LAC-LAC, and LNS-LNS.  These models are diagrammed below.
   (Dotted lines designate network connections.  Solid lines designate
   circuit connections.)

                     Figure 2.0: L2TP Reference Models

   (a) LAC-LNS Reference Model: On one side, the LAC receives traffic
   from an L2 circuit, which it forwards via L2TP across an IP or other
   packet-based network.  On the other side, an LNS logically terminates
   the L2 circuit locally and routes network traffic to the home
   network.  The action of session establishment is driven by the LAC
   (as an incoming call) or the LNS (as an outgoing call).

    +-----+  L2  +-----+                        +-----+
    |     |------| LAC |.........[ IP ].........| LNS |...[home network]
    +-----+      +-----+                        +-----+
    remote
    system
                       |<-- emulated service -->|
          |<----------- L2 service ------------>|

   (b) LAC-LAC Reference Model: In this model, both LCCEs are LACs.
   Each LAC forwards circuit traffic from the remote system to the peer
   LAC using L2TP, and vice versa.  In its simplest form, an LAC acts as
   a simple cross-connect between a circuit to a remote system and an
   L2TP session.  This model typically involves symmetric establishment;
   that is, either side of the connection may initiate a session at any
   time (or simultaneously, in which a tie breaking mechanism is
   utilized).




Lau, et al.                 Standards Track                     [Page 8]

RFC 3931                         L2TPv3                       March 2005


   +-----+  L2  +-----+                      +-----+  L2  +-----+
   |     |------| LAC |........[ IP ]........| LAC |------|     |
   +-----+      +-----+                      +-----+      +-----+
   remote                                                 remote
   system                                                 system
                      |<- emulated service ->|
         |<----------------- L2 service ----------------->|

   (c) LNS-LNS Reference Model: This model has two LNSs as the LCCEs.  A
   user-level, traffic-generated, or signaled event typically drives
   session establishment from one side of the tunnel.  For example, a
   tunnel generated from a PC by a user, or automatically by customer
   premises equipment.

                   +-----+                      +-----+
  [home network]...| LNS |........[ IP ]........| LNS |...[home network]
                   +-----+                      +-----+
                         |<- emulated service ->|
                         |<---- L2 service ---->|

   Note: In L2TPv2, user-driven tunneling of this type is often referred
   to as "voluntary tunneling" [RFC2809].  Further, an LNS acting as
   part of a software package on a host is sometimes referred to as an
   "LAC Client" [RFC2661].

3.  Protocol Overview



   L2TP is comprised of two types of messages, control messages and data
   messages (sometimes referred to as "control packets" and "data
   packets", respectively).  Control messages are used in the
   establishment, maintenance, and clearing of control connections and
   sessions.  These messages utilize a reliable control channel within
   L2TP to guarantee delivery (see Section 4.2 for details).  Data
   messages are used to encapsulate the L2 traffic being carried over
   the L2TP session.  Unlike control messages, data messages are not
   retransmitted when packet loss occurs.

   The L2TPv3 control message format defined in this document borrows
   largely from L2TPv2.  These control messages are used in conjunction
   with the associated protocol state machines that govern the dynamic
   setup, maintenance, and teardown for L2TP sessions.  The data message
   format for tunneling data packets may be utilized with or without the
   L2TP control channel, either via manual configuration or via other
   signaling methods to pre-configure or distribute L2TP session
   information.  Utilization of the L2TP data message format with other
   signaling methods is outside the scope of this document.





Lau, et al.                 Standards Track                     [Page 9]

RFC 3931                         L2TPv3                       March 2005


                       Figure 3.0: L2TPv3 Structure

             +-------------------+    +-----------------------+
             | Tunneled Frame    |    | L2TP Control Message  |
             +-------------------+    +-----------------------+
             | L2TP Data Header  |    | L2TP Control Header   |
             +-------------------+    +-----------------------+
             | L2TP Data Channel |    | L2TP Control Channel  |
             | (unreliable)      |    | (reliable)            |
             +-------------------+----+-----------------------+
             | Packet-Switched Network (IP, FR, MPLS, etc.)   |
             +------------------------------------------------+

   Figure 3.0 depicts the relationship of control messages and data
   messages over the L2TP control and data channels, respectively.  Data
   messages are passed over an unreliable data channel, encapsulated by
   an L2TP header, and sent over a Packet-Switched Network (PSN) such as
   IP, UDP, Frame Relay, ATM, MPLS, etc.  Control messages are sent over
   a reliable L2TP control channel, which operates over the same PSN.

   The necessary setup for tunneling a session with L2TP consists of two
   steps: (1) Establishing the control connection, and (2) establishing
   a session as triggered by an incoming call or outgoing call.  An L2TP
   session MUST be established before L2TP can begin to forward session
   frames.  Multiple sessions may be bound to a single control
   connection, and multiple control connections may exist between the
   same two LCCEs.

3.1.  Control Message Types



   The Message Type AVP (see Section 5.4.1) defines the specific type of
   control message being sent.

   This document defines the following control message types (see
   Sections 6.1 through 6.15 for details on the construction and use of
   each message):

   Control Connection Management

       0  (reserved)
       1  (SCCRQ)    Start-Control-Connection-Request
       2  (SCCRP)    Start-Control-Connection-Reply
       3  (SCCCN)    Start-Control-Connection-Connected
       4  (StopCCN)  Stop-Control-Connection-Notification
       5  (reserved)
       6  (HELLO)    Hello
      20  (ACK)      Explicit Acknowledgement




Lau, et al.                 Standards Track                    [Page 10]

RFC 3931                         L2TPv3                       March 2005


   Call Management

       7  (OCRQ)     Outgoing-Call-Request
       8  (OCRP)     Outgoing-Call-Reply
       9  (OCCN)     Outgoing-Call-Connected
      10  (ICRQ)     Incoming-Call-Request
      11  (ICRP)     Incoming-Call-Reply
      12  (ICCN)     Incoming-Call-Connected
      13  (reserved)
      14  (CDN)      Call-Disconnect-Notify

   Error Reporting

      15  (WEN)      WAN-Error-Notify



   Link Status Change Reporting

      16  (SLI)      Set-Link-Info



3.2.  L2TP Header Formats



   This section defines header formats for L2TP control messages and
   L2TP data messages.  All values are placed into their respective
   fields and sent in network order (high-order octets first).

3.2.1.  L2TP Control Message Header



   The L2TP control message header provides information for the reliable
   transport of messages that govern the establishment, maintenance, and
   teardown of L2TP sessions.  By default, control messages are sent
   over the underlying media in-band with L2TP data messages.

   The L2TP control message header is formatted as follows:

                 Figure 3.2.1: L2TP Control Message Header

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |T|L|x|x|S|x|x|x|x|x|x|x|  Ver  |             Length            |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                     Control Connection ID                     |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |               Ns              |               Nr              |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

   The T bit MUST be set to 1, indicating that this is a control
   message.



Lau, et al.                 Standards Track                    [Page 11]

RFC 3931                         L2TPv3                       March 2005


   The L and S bits MUST be set to 1, indicating that the Length field
   and sequence numbers are present.

   The x bits are reserved for future extensions.  All reserved bits
   MUST be set to 0 on outgoing messages and ignored on incoming
   messages.

   The Ver field indicates the version of the L2TP control message
   header described in this document.  On sending, this field MUST be
   set to 3 for all messages (unless operating in an environment that
   includes L2TPv2 [RFC2661] and/or L2F [RFC2341] as well, see Section
   4.1 for details).

   The Length field indicates the total length of the message in octets,
   always calculated from the start of the control message header itself
   (beginning with the T bit).

   The Control Connection ID field contains the identifier for the
   control connection.  L2TP control connections are named by
   identifiers that have local significance only.  That is, the same
   control connection will be given unique Control Connection IDs by
   each LCCE from within each endpoint's own Control Connection ID
   number space.  As such, the Control Connection ID in each message is
   that of the intended recipient, not the sender.  Non-zero Control
   Connection IDs are selected and exchanged as Assigned Control
   Connection ID AVPs during the creation of a control connection.

   Ns indicates the sequence number for this control message, beginning
   at zero and incrementing by one (modulo 2**16) for each message sent.
   See Section 4.2 for more information on using this field.

   Nr indicates the sequence number expected in the next control message
   to be received.  Thus, Nr is set to the Ns of the last in-order
   message received plus one (modulo 2**16).  See Section 4.2 for more
   information on using this field.

3.2.2.  L2TP Data Message



   In general, an L2TP data message consists of a (1) Session Header,
   (2) an optional L2-Specific Sublayer, and (3) the Tunnel Payload, as
   depicted below.










Lau, et al.                 Standards Track                    [Page 12]

RFC 3931                         L2TPv3                       March 2005


                  Figure 3.2.2: L2TP Data Message Header

   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                      L2TP Session Header                      |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                      L2-Specific Sublayer                     |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                        Tunnel Payload                      ...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

   The L2TP Session Header is specific to the encapsulating PSN over
   which the L2TP traffic is delivered.  The Session Header MUST provide
   (1) a method of distinguishing traffic among multiple L2TP data
   sessions and (2) a method of distinguishing data messages from
   control messages.

   Each type of encapsulating PSN MUST define its own session header,
   clearly identifying the format of the header and parameters necessary
   to setup the session.  Section 4.1 defines two session headers, one
   for transport over UDP and one for transport over IP.

   The L2-Specific Sublayer is an intermediary layer between the L2TP
   session header and the start of the tunneled frame.  It contains
   control fields that are used to facilitate the tunneling of each
   frame (e.g., sequence numbers or flags).  The Default L2-Specific
   Sublayer for L2TPv3 is defined in Section 4.6.

   The Data Message Header is followed by the Tunnel Payload, including
   any necessary L2 framing as defined in the payload-specific companion
   documents.

3.3.  Control Connection Management



   The L2TP control connection handles dynamic establishment, teardown,
   and maintenance of the L2TP sessions and of the control connection
   itself.  The reliable delivery of control messages is described in
   Section 4.2.

   This section describes typical control connection establishment and
   teardown exchanges.  It is important to note that, in the diagrams
   that follow, the reliable control message delivery mechanism exists
   independently of the L2TP state machine.  For instance, Explicit
   Acknowledgement (ACK) messages may be sent after any of the control
   messages indicated in the exchanges below if an acknowledgment is not
   piggybacked on a later control message.






Lau, et al.                 Standards Track                    [Page 13]

RFC 3931                         L2TPv3                       March 2005


   LCCEs are identified during control connection establishment either
   by the Host Name AVP, the Router ID AVP, or a combination of the two
   (see Section 5.4.3).  The identity of a peer LCCE is central to
   selecting proper configuration parameters (i.e., Hello interval,
   window size, etc.) for a control connection, as well as for
   determining how to set up associated sessions within the control
   connection, password lookup for control connection authentication,
   control connection level tie breaking, etc.

3.3.1.  Control Connection Establishment



   Establishment of the control connection involves an exchange of AVPs
   that identifies the peer and its capabilities.

   A three-message exchange is used to establish the control connection.
   The following is a typical message exchange:

      LCCE A      LCCE B
      ------      ------
      SCCRQ ->
                  <- SCCRP
      SCCCN ->

3.3.2.  Control Connection Teardown



   Control connection teardown may be initiated by either LCCE and is
   accomplished by sending a single StopCCN control message.  As part of
   the reliable control message delivery mechanism, the recipient of a
   StopCCN MUST send an ACK message to acknowledge receipt of the
   message and maintain enough control connection state to properly
   accept StopCCN retransmissions over at least a full retransmission
   cycle (in case the ACK message is lost).  The recommended time for a
   full retransmission cycle is at least 31 seconds (see Section 4.2).
   The following is an example of a typical control message exchange:

      LCCE A      LCCE B
      ------      ------
      StopCCN ->
      (Clean up)

                  (Wait)
                  (Clean up)

   An implementation may shut down an entire control connection and all
   sessions associated with the control connection by sending the
   StopCCN.  Thus, it is not necessary to clear each session
   individually when tearing down the whole control connection.




Lau, et al.                 Standards Track                    [Page 14]

RFC 3931                         L2TPv3                       March 2005


3.4.  Session Management



   After successful control connection establishment, individual
   sessions may be created.  Each session corresponds to a single data
   stream between the two LCCEs.  This section describes the typical
   call establishment and teardown exchanges.

3.4.1.  Session Establishment for an Incoming Call



   A three-message exchange is used to establish the session.  The
   following is a typical sequence of events:

      LCCE A      LCCE B
      ------      ------
      (Call
       Detected)

      ICRQ ->
                 <- ICRP
      (Call
       Accepted)

      ICCN ->

3.4.2.  Session Establishment for an Outgoing Call



   A three-message exchange is used to set up the session.  The
   following is a typical sequence of events:

      LCCE A      LCCE B
      ------      ------
                 <- OCRQ
      OCRP ->

      (Perform
       Call
       Operation)

      OCCN ->

      (Call Operation
       Completed
       Successfully)








Lau, et al.                 Standards Track                    [Page 15]

RFC 3931                         L2TPv3                       March 2005


3.4.3.  Session Teardown



   Session teardown may be initiated by either the LAC or LNS and is
   accomplished by sending a CDN control message.  After the last
   session is cleared, the control connection MAY be torn down as well
   (and typically is).  The following is an example of a typical control
   message exchange:

      LCCE A      LCCE B
      ------      ------
      CDN ->
      (Clean up)

                  (Clean up)

4.  Protocol Operation



4.1.  L2TP Over Specific Packet-Switched Networks (PSNs)



   L2TP may operate over a variety of PSNs.  There are two modes
   described for operation over IP, L2TP directly over IP (see Section
   4.1.1) and L2TP over UDP (see Section 4.1.2).  L2TPv3 implementations
   MUST support L2TP over IP and SHOULD support L2TP over UDP for better
   NAT and firewall traversal, and for easier migration from L2TPv2.

   L2TP over other PSNs may be defined, but the specifics are outside
   the scope of this document.  Examples of L2TPv2 over other PSNs
   include [RFC3070] and [RFC3355].

   The following field definitions are defined for use in all L2TP
   Session Header encapsulations.

   Session ID

      A 32-bit field containing a non-zero identifier for a session.
      L2TP sessions are named by identifiers that have local
      significance only.  That is, the same logical session will be
      given different Session IDs by each end of the control connection
      for the life of the session.  When the L2TP control connection is
      used for session establishment, Session IDs are selected and
      exchanged as Local Session ID AVPs during the creation of a
      session.  The Session ID alone provides the necessary context for
      all further packet processing, including the presence, size, and
      value of the Cookie, the type of L2-Specific Sublayer, and the
      type of payload being tunneled.






Lau, et al.                 Standards Track                    [Page 16]

RFC 3931                         L2TPv3                       March 2005


   Cookie

      The optional Cookie field contains a variable-length value
      (maximum 64 bits) used to check the association of a received data
      message with the session identified by the Session ID.  The Cookie
      MUST be set to the configured or signaled random value for this
      session.  The Cookie provides an additional level of guarantee
      that a data message has been directed to the proper session by the
      Session ID.  A well-chosen Cookie may prevent inadvertent
      misdirection of stray packets with recently reused Session IDs,
      Session IDs subject to packet corruption, etc.  The Cookie may
      also provide protection against some specific malicious packet
      insertion attacks, as described in Section 8.2.

      When the L2TP control connection is used for session
      establishment, random Cookie values are selected and exchanged as
      Assigned Cookie AVPs during session creation.

4.1.1.  L2TPv3 over IP



   L2TPv3 over IP (both versions) utilizes the IANA-assigned IP protocol
   ID 115.

4.1.1.1.  L2TPv3 Session Header Over IP



   Unlike L2TP over UDP, the L2TPv3 session header over IP is free of
   any restrictions imposed by coexistence with L2TPv2 and L2F.  As
   such, the header format has been designed to optimize packet
   processing.  The following session header format is utilized when
   operating L2TPv3 over IP:

               Figure 4.1.1.1: L2TPv3 Session Header Over IP

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                           Session ID                          |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |               Cookie (optional, maximum 64 bits)...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
                                                                   |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

   The Session ID and Cookie fields are as defined in Section 4.1.  The
   Session ID of zero is reserved for use by L2TP control messages (see
   Section 4.1.1.2).





Lau, et al.                 Standards Track                    [Page 17]

RFC 3931                         L2TPv3                       March 2005


4.1.1.2.  L2TP Control and Data Traffic over IP



   Unlike L2TP over UDP, which uses the T bit to distinguish between
   L2TP control and data packets, L2TP over IP uses the reserved Session
   ID of zero (0) when sending control messages.  It is presumed that
   checking for the zero Session ID is more efficient -- both in header
   size for data packets and in processing speed for distinguishing
   between control and data messages -- than checking a single bit.

   The entire control message header over IP, including the zero session
   ID, appears as follows:

           Figure 4.1.1.2: L2TPv3 Control Message Header Over IP

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                      (32 bits of zeros)                       |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |T|L|x|x|S|x|x|x|x|x|x|x|  Ver  |             Length            |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                     Control Connection ID                     |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |               Ns              |               Nr              |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

   Named fields are as defined in Section 3.2.1.  Note that the Length
   field is still calculated from the beginning of the control message
   header, beginning with the T bit.  It does NOT include the "(32 bits
   of zeros)" depicted above.

   When operating directly over IP, L2TP packets lose the ability to
   take advantage of the UDP checksum as a simple packet integrity
   check, which is of particular concern for L2TP control messages.
   Control Message Authentication (see Section 4.3), even with an empty
   password field, provides for a sufficient packet integrity check and
   SHOULD always be enabled.

4.1.2.  L2TP over UDP



   L2TPv3 over UDP must consider other L2 tunneling protocols that may
   be operating in the same environment, including L2TPv2 [RFC2661] and
   L2F [RFC2341].

   While there are efficiencies gained by running L2TP directly over IP,
   there are possible side effects as well.  For instance, L2TP over IP
   is not as NAT-friendly as L2TP over UDP.




Lau, et al.                 Standards Track                    [Page 18]

RFC 3931                         L2TPv3                       March 2005


4.1.2.1.  L2TP Session Header Over UDP



   The following session header format is utilized when operating L2TPv3
   over UDP:

              Figure 4.1.2.1: L2TPv3 Session Header over UDP

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |T|x|x|x|x|x|x|x|x|x|x|x|  Ver  |          Reserved             |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                           Session ID                          |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |               Cookie (optional, maximum 64 bits)...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
                                                                   |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

   The T bit MUST be set to 0, indicating that this is a data message.

   The x bits and Reserved field are reserved for future extensions.
   All reserved values MUST be set to 0 on outgoing messages and ignored
   on incoming messages.

   The Ver field MUST be set to 3, indicating an L2TPv3 message.

   Note that the initial bits 1, 4, 6, and 7 have meaning in L2TPv2
   [RFC2661], and are deprecated and marked as reserved in L2TPv3.
   Thus, for UDP mode on a system that supports both versions of L2TP,
   it is important that the Ver field be inspected first to determine
   the Version of the header before acting upon any of these bits.

   The Session ID and Cookie fields are as defined in Section 4.1.

4.1.2.2.  UDP Port Selection



   The method for UDP Port Selection defined in this section is
   identical to that defined for L2TPv2 [RFC2661].

   When negotiating a control connection over UDP, control messages MUST
   be sent as UDP datagrams using the registered UDP port 1701
   [RFC1700].  The initiator of an L2TP control connection picks an
   available source UDP port (which may or may not be 1701) and sends to
   the desired destination address at port 1701.  The recipient picks a
   free port on its own system (which may or may not be 1701) and sends
   its reply to the initiator's UDP port and address, setting its own
   source port to the free port it found.



Lau, et al.                 Standards Track                    [Page 19]

RFC 3931                         L2TPv3                       March 2005


   Any subsequent traffic associated with this control connection
   (either control traffic or data traffic from a session established
   through this control connection) must use these same UDP ports.

   It has been suggested that having the recipient choose an arbitrary
   source port (as opposed to using the destination port in the packet
   initiating the control connection, i.e., 1701) may make it more
   difficult for L2TP to traverse some NAT devices.  Implementations
   should consider the potential implication of this capability before
   choosing an arbitrary source port.  A NAT device that can pass TFTP
   traffic with variant UDP ports should be able to pass L2TP UDP
   traffic since both protocols employ similar policies with regard to
   UDP port selection.

4.1.2.3.  UDP Checksum



   The tunneled frames that L2TP carry often have their own checksums or
   integrity checks, rendering the UDP checksum redundant for much of
   the L2TP data message contents.  Thus, UDP checksums MAY be disabled
   in order to reduce the associated packet processing burden at the
   L2TP endpoints.

   The L2TP header itself does not have its own checksum or integrity
   check.  However, use of the L2TP Session ID and Cookie pair guards
   against accepting an L2TP data message if corruption of the Session
   ID or associated Cookie has occurred.  When the L2-Specific Sublayer
   is present in the L2TP header, there is no built-in integrity check
   for the information contained therein if UDP checksums or some other
   integrity check is not employed.  IPsec (see Section 4.1.3) may be
   used for strong integrity protection of the entire contents of L2TP
   data messages.

   UDP checksums MUST be enabled for L2TP control messages.

4.1.3.  L2TP and IPsec



   The L2TP data channel does not provide cryptographic security of any
   kind.  If the L2TP data channel operates over a public or untrusted
   IP network where privacy of the L2TP data is of concern or
   sophisticated attacks against L2TP are expected to occur, IPsec
   [RFC2401] MUST be made available to secure the L2TP traffic.

   Either L2TP over UDP or L2TP over IP may be secured with IPsec.
   [RFC3193] defines the recommended method for securing L2TPv2.  L2TPv3
   possesses identical characteristics to IPsec as L2TPv2 when running
   over UDP and implementations MUST follow the same recommendation.
   When operating over IP directly, [RFC3193] still applies, though
   references to UDP source and destination ports (in particular, those



Lau, et al.                 Standards Track                    [Page 20]

RFC 3931                         L2TPv3                       March 2005


   in Section 4, "IPsec Filtering details when protecting L2TP") may be
   ignored.  Instead, the selectors used to identify L2TPv3 traffic are
   simply the source and destination IP addresses for the tunnel
   endpoints together with the L2TPv3 IP protocol type, 115.

   In addition to IP transport security, IPsec defines a mode of
   operation that allows tunneling of IP packets.  The packet-level
   encryption and authentication provided by IPsec tunnel mode and that
   provided by L2TP secured with IPsec provide an equivalent level of
   security for these requirements.

   IPsec also defines access control features that are required of a
   compliant IPsec implementation.  These features allow filtering of
   packets based upon network and transport layer characteristics such
   as IP address, ports, etc.  In the L2TP tunneling model, analogous
   filtering may be performed at the network layer above L2TP.  These
   network layer access control features may be handled at an LCCE via
   vendor-specific authorization features, or at the network layer
   itself by using IPsec transport mode end-to-end between the
   communicating hosts.  The requirements for access control mechanisms
   are not a part of the L2TP specification, and as such, are outside
   the scope of this document.

   Protecting the L2TP packet stream with IPsec does, in turn, also
   protect the data within the tunneled session packets while
   transported from one LCCE to the other.  Such protection must not be
   considered a substitution for end-to-end security between
   communicating hosts or applications.

4.1.4.  IP Fragmentation Issues



   Fragmentation and reassembly in network equipment generally require
   significantly greater resources than sending or receiving a packet as
   a single unit.  As such, fragmentation and reassembly should be
   avoided whenever possible.  Ideal solutions for avoiding
   fragmentation include proper configuration and management of MTU
   sizes among the Remote System, the LCCE, and the IP network, as well
   as adaptive measures that operate with the originating host (e.g.,
   [RFC1191], [RFC1981]) to reduce the packet sizes at the source.

   An LCCE MAY fragment a packet before encapsulating it in L2TP.  For
   example, if an IPv4 packet arrives at an LCCE from a Remote System
   that, after encapsulation with its associated framing, L2TP, and IP,
   does not fit in the available path MTU towards its LCCE peer, the
   local LCCE may perform IPv4 fragmentation on the packet before tunnel
   encapsulation.  This creates two (or more) L2TP packets, each





Lau, et al.                 Standards Track                    [Page 21]

RFC 3931                         L2TPv3                       March 2005


   carrying an IPv4 fragment with its associated framing.  This
   ultimately has the effect of placing the burden of fragmentation on
   the LCCE, while reassembly occurs on the IPv4 destination host.

   If an IPv6 packet arrives at an LCCE from a Remote System that, after
   encapsulation with associated framing, L2TP and IP, does not fit in
   the available path MTU towards its L2TP peer, the Generic Packet
   Tunneling specification [RFC2473], Section 7.1 SHOULD be followed.
   In this case, the LCCE should either send an ICMP Packet Too Big
   message to the data source, or fragment the resultant L2TP/IP packet
   (for reassembly by the L2TP peer).

   If the amount of traffic requiring fragmentation and reassembly is
   rather light, or there are sufficiently optimized mechanisms at the
   tunnel endpoints, fragmentation of the L2TP/IP packet may be
   sufficient for accommodating mismatched MTUs that cannot be managed
   by more efficient means.  This method effectively emulates a larger
   MTU between tunnel endpoints and should work for any type of L2-
   encapsulated packet.  Note that IPv6 does not support "in-flight"
   fragmentation of data packets.  Thus, unlike IPv4, the MTU of the
   path towards an L2TP peer must be known in advance (or the last
   resort IPv6 minimum MTU of 1280 bytes utilized) so that IPv6
   fragmentation may occur at the LCCE.

   In summary, attempting to control the source MTU by communicating
   with the originating host, forcing that an MTU be sufficiently large
   on the path between LCCE peers to tunnel a frame from any other
   interface without fragmentation, fragmenting IP packets before
   encapsulation with L2TP/IP, or fragmenting the resultant L2TP/IP
   packet between the tunnel endpoints, are all valid methods for
   managing MTU mismatches.  Some are clearly better than others
   depending on the given deployment.  For example, a passive monitoring
   application using L2TP would certainly not wish to have ICMP messages
   sent to a traffic source.  Further, if the links connecting a set of
   LCCEs have a very large MTU (e.g., SDH/SONET) and it is known that
   the MTU of all links being tunneled by L2TP have smaller MTUs (e.g.,
   1500 bytes), then any IP fragmentation and reassembly enabled on the
   participating LCCEs would never be utilized.  An implementation MUST
   implement at least one of the methods described in this section for
   managing mismatched MTUs, based on careful consideration of how the
   final product will be deployed.

   L2TP-specific fragmentation and reassembly methods, which may or may
   not depend on the characteristics of the type of link being tunneled
   (e.g., judicious packing of ATM cells), may be defined as well, but
   these methods are outside the scope of this document.





Lau, et al.                 Standards Track                    [Page 22]

RFC 3931                         L2TPv3                       March 2005


4.2.  Reliable Delivery of Control Messages



   L2TP provides a lower level reliable delivery service for all control
   messages.  The Nr and Ns fields of the control message header (see
   Section 3.2.1) belong to this delivery mechanism.  The upper level
   functions of L2TP are not concerned with retransmission or ordering
   of control messages.  The reliable control messaging mechanism is a
   sliding window mechanism that provides control message retransmission
   and congestion control.  Each peer maintains separate sequence number
   state for each control connection.

   The message sequence number, Ns, begins at 0.  Each subsequent
   message is sent with the next increment of the sequence number.  The
   sequence number is thus a free-running counter represented modulo
   65536.  The sequence number in the header of a received message is
   considered less than or equal to the last received number if its
   value lies in the range of the last received number and the preceding
   32767 values, inclusive.  For example, if the last received sequence
   number was 15, then messages with sequence numbers 0 through 15, as
   well as 32784 through 65535, would be considered less than or equal.
   Such a message would be considered a duplicate of a message already
   received and ignored from processing.  However, in order to ensure
   that all messages are acknowledged properly (particularly in the case
   of a lost ACK message), receipt of duplicate messages MUST be
   acknowledged by the reliable delivery mechanism.  This acknowledgment
   may either piggybacked on a message in queue or sent explicitly via
   an ACK message.

   All control messages take up one slot in the control message sequence
   number space, except the ACK message.  Thus, Ns is not incremented
   after an ACK message is sent.

   The last received message number, Nr, is used to acknowledge messages
   received by an L2TP peer.  It contains the sequence number of the
   message the peer expects to receive next (e.g., the last Ns of a
   non-ACK message received plus 1, modulo 65536).  While the Nr in a
   received ACK message is used to flush messages from the local
   retransmit queue (see below), the Nr of the next message sent is not
   updated by the Ns of the ACK message.  Nr SHOULD be sanity-checked
   before flushing the retransmit queue.  For instance, if the Nr
   received in a control message is greater than the last Ns sent plus 1
   modulo 65536, the control message is clearly invalid.

   The reliable delivery mechanism at a receiving peer is responsible
   for making sure that control messages are delivered in order and
   without duplication to the upper level.  Messages arriving out-of-
   order may be queued for in-order delivery when the missing messages




Lau, et al.                 Standards Track                    [Page 23]

RFC 3931                         L2TPv3                       March 2005


   are received.  Alternatively, they may be discarded, thus requiring a
   retransmission by the peer.  When dropping out-of-order control
   packets, Nr MAY be updated before the packet is discarded.

   Each control connection maintains a queue of control messages to be
   transmitted to its peer.  The message at the front of the queue is
   sent with a given Ns value and is held until a control message
   arrives from the peer in which the Nr field indicates receipt of this
   message.  After a period of time (a recommended default is 1 second
   but SHOULD be configurable) passes without acknowledgment, the
   message is retransmitted.  The retransmitted message contains the
   same Ns value, but the Nr value MUST be updated with the sequence
   number of the next expected message.

   Each subsequent retransmission of a message MUST employ an
   exponential backoff interval.  Thus, if the first retransmission
   occurred after 1 second, the next retransmission should occur after 2
   seconds has elapsed, then 4 seconds, etc.  An implementation MAY
   place a cap upon the maximum interval between retransmissions.  This
   cap SHOULD be no less than 8 seconds per retransmission.  If no peer
   response is detected after several retransmissions (a recommended
   default is 10, but MUST be configurable), the control connection and
   all associated sessions MUST be cleared.  As it is the first message
   to establish a control connection, the SCCRQ MAY employ a different
   retransmission maximum than other control messages in order to help
   facilitate failover to alternate LCCEs in a timely fashion.

   When a control connection is being shut down for reasons other than
   loss of connectivity, the state and reliable delivery mechanisms MUST
   be maintained and operated for the full retransmission interval after
   the final message StopCCN message has been sent (e.g., 1 + 2 + 4 + 8
   + 8... seconds), or until the StopCCN message itself has been
   acknowledged.

   A sliding window mechanism is used for control message transmission
   and retransmission.  Consider two peers, A and B.  Suppose A
   specifies a Receive Window Size AVP with a value of N in the SCCRQ or
   SCCRP message.  B is now allowed to have a maximum of N outstanding
   (i.e., unacknowledged) control messages.  Once N messages have been
   sent, B must wait for an acknowledgment from A that advances the
   window before sending new control messages.  An implementation may
   advertise a non-zero receive window as small or as large as it
   wishes, depending on its own ability to process incoming messages
   before sending an acknowledgement.  Each peer MUST limit the number
   of unacknowledged messages it will send before receiving an
   acknowledgement by this Receive Window Size.  The actual internal





Lau, et al.                 Standards Track                    [Page 24]

RFC 3931                         L2TPv3                       March 2005


   unacknowledged message send-queue depth may be further limited by
   local resource allocation or by dynamic slow-start and congestion-
   avoidance mechanisms.

   When retransmitting control messages, a slow start and congestion
   avoidance window adjustment procedure SHOULD be utilized.  A
   recommended procedure is described in Appendix A.  A peer MAY drop
   messages, but MUST NOT actively delay acknowledgment of messages as a
   technique for flow control of control messages.  Appendix B contains
   examples of control message transmission, acknowledgment, and
   retransmission.

4.3.  Control Message Authentication



   L2TP incorporates an optional authentication and integrity check for
   all control messages.  This mechanism consists of a computed one-way
   hash over the header and body of the L2TP control message, a pre-
   configured shared secret, and a local and remote nonce (random value)
   exchanged via the Control Message Authentication Nonce AVP. This
   per-message authentication and integrity check is designed to perform
   a mutual authentication between L2TP nodes, perform integrity
   checking of all control messages, and guard against control message
   spoofing and replay attacks that would otherwise be trivial to mount.

   At least one shared secret (password) MUST exist between
   communicating L2TP nodes to enable Control Message Authentication.
   See Section 5.4.3 for details on calculation of the Message Digest
   and construction of the Control Message Authentication Nonce and
   Message Digest AVPs.

   L2TPv3 Control Message Authentication is similar to L2TPv2 [RFC2661]
   Tunnel Authentication in its use of a shared secret and one-way hash
   calculation.  The principal difference is that, instead of computing
   the hash over selected contents of a received control message (e.g.,
   the Challenge AVP and Message Type) as in L2TPv2, the entire message
   is used in the hash in L2TPv3.  In addition, instead of including the
   hash digest in just the SCCRP and SCCCN messages, it is now included
   in all L2TP messages.

   The Control Message Authentication mechanism is optional, and may be
   disabled if both peers agree.  For example, if IPsec is already being
   used for security and integrity checking between the LCCEs, the
   function of the L2TP mechanism becomes redundant and may be disabled.

   Presence of the Control Message Authentication Nonce AVP in an SCCRQ
   or SCCRP message serves as indication to a peer that Control Message
   Authentication is enabled.  If an SCCRQ or SCCRP contains a Control
   Message Authentication Nonce AVP, the receiver of the message MUST



Lau, et al.                 Standards Track                    [Page 25]

RFC 3931                         L2TPv3                       March 2005


   respond with a Message Digest AVP in all subsequent messages sent.
   Control Message Authentication is always bidirectional; either both
   sides participate in authentication, or neither does.

   If Control Message Authentication is disabled, the Message Digest AVP
   still MAY be sent as an integrity check of the message.  The
   integrity check is calculated as in Section 5.4.3, with an empty
   zero-length shared secret, local nonce, and remote nonce.  If an
   invalid Message Digest is received, it should be assumed that the
   message has been corrupted in transit and the message dropped
   accordingly.

   Implementations MAY rate-limit control messages, particularly SCCRQ
   messages, upon receipt for performance reasons or for protection
   against denial of service attacks.

4.4.  Keepalive (Hello)



   L2TP employs a keepalive mechanism to detect loss of connectivity
   between a pair of LCCEs.  This is accomplished by injecting Hello
   control messages (see Section 6.5) after a period of time has elapsed
   since the last data message or control message was received on an
   L2TP session or control connection, respectively.  As with any other
   control message, if the Hello message is not reliably delivered, the
   sending LCCE declares that the control connection is down and resets
   its state for the control connection.  This behavior ensures that a
   connectivity failure between the LCCEs is detected independently by
   each end of a control connection.

   Since the control channel is operated in-band with data traffic over
   the PSN, this single mechanism can be used to infer basic data
   connectivity between a pair of LCCEs for all sessions associated with
   the control connection.

   Periodic keepalive for the control connection MUST be implemented by
   sending a Hello if a period of time (a recommended default is 60
   seconds, but MUST be configurable) has passed without receiving any
   message (data or control) from the peer.  An LCCE sending Hello
   messages across multiple control connections between the same LCCE
   endpoints MUST employ a jittered timer mechanism to prevent grouping
   of Hello messages.

4.5.  Forwarding Session Data Frames



   Once session establishment is complete, circuit frames are received
   at an LCCE, encapsulated in L2TP (with appropriate attention to
   framing, as described in documents for the particular pseudowire
   type), and forwarded over the appropriate session.  For every



Lau, et al.                 Standards Track                    [Page 26]

RFC 3931                         L2TPv3                       March 2005


   outgoing data message, the sender places the identifier specified in
   the Local Session ID AVP (received from peer during session
   establishment) in the Session ID field of the L2TP data header.  In
   this manner, session frames are multiplexed and demultiplexed between
   a given pair of LCCEs.  Multiple control connections may exist
   between a given pair of LCCEs, and multiple sessions may be
   associated with a given control connection.

   The peer LCCE receiving the L2TP data packet identifies the session
   with which the packet is associated by the Session ID in the data
   packet's header.  The LCCE then checks the Cookie field in the data
   packet against the Cookie value received in the Assigned Cookie AVP
   during session establishment.  It is important for implementers to
   note that the Cookie field check occurs after looking up the session
   context by the Session ID, and as such, consists merely of a value
   match of the Cookie field and that stored in the retrieved context.
   There is no need to perform a lookup across the Session ID and Cookie
   as a single value.  Any received data packets that contain invalid
   Session IDs or associated Cookie values MUST be dropped.  Finally,
   the LCCE either forwards the network packet within the tunneled frame
   (e.g., as an LNS) or switches the frame to a circuit (e.g., as an
   LAC).

4.6.  Default L2-Specific Sublayer



   This document defines a Default L2-Specific Sublayer format (see
   Section 3.2.2) that a pseudowire may use for features such as
   sequencing support, L2 interworking, OAM, or other per-data-packet
   operations.  The Default L2-Specific Sublayer SHOULD be used by a
   given PW type to support these features if it is adequate, and its
   presence is requested by a peer during session negotiation.
   Alternative sublayers MAY be defined (e.g., an encapsulation with a
   larger Sequence Number field or timing information) and identified
   for use via the L2-Specific Sublayer Type AVP.

              Figure 4.6: Default L2-Specific Sublayer Format

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |x|S|x|x|x|x|x|x|              Sequence Number                  |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

   The S (Sequence) bit is set to 1 when the Sequence Number contains a
   valid number for this sequenced frame.  If the S bit is set to zero,
   the Sequence Number contents are undefined and MUST be ignored by the
   receiver.




Lau, et al.                 Standards Track                    [Page 27]

RFC 3931                         L2TPv3                       March 2005


   The Sequence Number field contains a free-running counter of 2^24
   sequence numbers.  If the number in this field is valid, the S bit
   MUST be set to 1.  The Sequence Number begins at zero, which is a
   valid sequence number.  (In this way, implementations inserting
   sequence numbers do not have to "skip" zero when incrementing.)  The
   sequence number in the header of a received message is considered
   less than or equal to the last received number if its value lies in
   the range of the last received number and the preceding (2^23-1)
   values, inclusive.

4.6.1.  Sequencing Data Packets



   The Sequence Number field may be used to detect lost, duplicate, or
   out-of-order packets within a given session.

   When L2 frames are carried over an L2TP-over-IP or L2TP-over-UDP/IP
   data channel, this part of the link has the characteristic of being
   able to reorder, duplicate, or silently drop packets.  Reordering may
   break some non-IP protocols or L2 control traffic being carried by
   the link.  Silent dropping or duplication of packets may break
   protocols that assume per-packet indications of error, such as TCP
   header compression.  While a common mechanism for packet sequence
   detection is provided, the sequence dependency characteristics of
   individual protocols are outside the scope of this document.

   If any protocol being transported by over L2TP data channels cannot
   tolerate misordering of data packets, packet duplication, or silent
   packet loss, sequencing may be enabled on some or all packets by
   using the S bit and Sequence Number field defined in the Default L2-
   Specific Sublayer (see Section 4.6).  For a given L2TP session, each
   LCCE is responsible for communicating to its peer the level of
   sequencing support that it requires of data packets that it receives.
   Mechanisms to advertise this information during session negotiation
   are provided (see Data Sequencing AVP in Section 5.4.4).

   When determining whether a packet is in or out of sequence, an
   implementation SHOULD utilize a method that is resilient to temporary
   dropouts in connectivity coupled with high per-session packet rates.
   The recommended method is outlined in Appendix C.

4.7.  L2TPv2/v3 Interoperability and Migration



   L2TPv2 and L2TPv3 environments should be able to coexist while a
   migration to L2TPv3 is made.  Migration issues are discussed for each
   media type in this section.  Most issues apply only to
   implementations that require both L2TPv2 and L2TPv3 operation.





Lau, et al.                 Standards Track                    [Page 28]

RFC 3931                         L2TPv3                       March 2005


   However, even L2TPv3-only implementations must at least be mindful of
   these issues in order to interoperate with implementations that
   support both versions.

4.7.1.  L2TPv3 over IP



   L2TPv3 implementations running strictly over IP with no desire to
   interoperate with L2TPv2 implementations may safely disregard most
   migration issues from L2TPv2.  All control messages and data messages
   are sent as described in this document, without normative reference
   to RFC 2661.

   If one wishes to tunnel PPP over L2TPv3, and fallback to L2TPv2 only
   if it is not available, then L2TPv3 over UDP with automatic fallback
   (see Section 4.7.3) MUST be used.  There is no deterministic method
   for automatic fallback from L2TPv3 over IP to either L2TPv2 or L2TPv3
   over UDP.  One could infer whether L2TPv3 over IP is supported by
   sending an SCCRQ and waiting for a response, but this could be
   problematic during periods of packet loss between L2TP nodes.

4.7.2.  L2TPv3 over UDP



   The format of the L2TPv3 over UDP header is defined in Section
   4.1.2.1.

   When operating over UDP, L2TPv3 uses the same port (1701) as L2TPv2
   and shares the first two octets of header format with L2TPv2.  The
   Ver field is used to distinguish L2TPv2 packets from L2TPv3 packets.
   If an implementation is capable of operating in L2TPv2 or L2TPv3
   modes, it is possible to automatically detect whether a peer can
   support L2TPv2 or L2TPv3 and operate accordingly.  The details of
   this fallback capability is defined in the following section.

4.7.3.  Automatic L2TPv2 Fallback



   When running over UDP, an implementation may detect whether a peer is
   L2TPv3-capable by sending a special SCCRQ that is properly formatted
   for both L2TPv2 and L2TPv3.  This is accomplished by sending an SCCRQ
   with its Ver field set to 2 (for L2TPv2), and ensuring that any
   L2TPv3-specific AVPs (i.e., AVPs present within this document and not
   defined within RFC 2661) in the message are sent with each M bit set
   to 0, and that all L2TPv2 AVPs are present as they would be for
   L2TPv2.  This is done so that L2TPv3 AVPs will be ignored by an
   L2TPv2-only implementation.  Note that, in both L2TPv2 and L2TPv3,
   the value contained in the space of the control message header
   utilized by the 32-bit Control Connection ID in L2TPv3, and the 16-
   bit Tunnel ID and




Lau, et al.                 Standards Track                    [Page 29]

RFC 3931                         L2TPv3                       March 2005


   16-bit Session ID in L2TPv2, are always 0 for an SCCRQ.  This
   effectively hides the fact that there are a pair of 16-bit fields in
   L2TPv2, and a single 32-bit field in L2TPv3.

   If the peer implementation is L2TPv3-capable, a control message with
   the Ver field set to 3 and an L2TPv3 header and message format will
   be sent in response to the SCCRQ.  Operation may then continue as
   L2TPv3.  If a message is received with the Ver field set to 2, it
   must be assumed that the peer implementation is L2TPv2-only, thus
   enabling fallback to L2TPv2 mode to safely occur.

   Note Well: The L2TPv2/v3 auto-detection mode requires that all L2TPv3
   implementations over UDP be liberal in accepting an SCCRQ control
   message with the Ver field set to 2 or 3 and the presence of L2TPv2-
   specific AVPs.  An L2TPv3-only implementation MUST ignore all L2TPv2
   AVPs (e.g., those defined in RFC 2661 and not in this document)
   within an SCCRQ with the Ver field set to 2 (even if the M bit is set
   on the L2TPv2-specific AVPs).

5.  Control Message Attribute Value Pairs



   To maximize extensibility while permitting interoperability, a
   uniform method for encoding message types is used throughout L2TP.
   This encoding will be termed AVP (Attribute Value Pair) for the
   remainder of this document.

5.1.  AVP Format



   Each AVP is encoded as follows:

                          Figure 5.1: AVP Format

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |M|H| rsvd  |      Length       |           Vendor ID           |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |         Attribute Type        |        Attribute Value ...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
                       (until Length is reached)                   |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

   The first six bits comprise a bit mask that describes the general
   attributes of the AVP.  Two bits are defined in this document; the
   remaining bits are reserved for future extensions.  Reserved bits
   MUST be set to 0 when sent and ignored upon receipt.





Lau, et al.                 Standards Track                    [Page 30]

RFC 3931                         L2TPv3                       March 2005


   Mandatory (M) bit: Controls the behavior required of an
   implementation that receives an unrecognized AVP.  The M bit of a
   given AVP MUST only be inspected and acted upon if the AVP is
   unrecognized (see Section 5.2).

   Hidden (H) bit: Identifies the hiding of data in the Attribute Value
   field of an AVP.  This capability can be used to avoid the passing of
   sensitive data, such as user passwords, as cleartext in an AVP.
   Section 5.3 describes the procedure for performing AVP hiding.

   Length: Contains the number of octets (including the Overall Length
   and bit mask fields) contained in this AVP.  The Length may be
   calculated as 6 + the length of the Attribute Value field in octets.

   The field itself is 10 bits, permitting a maximum of 1023 octets of
   data in a single AVP.  The minimum Length of an AVP is 6.  If the
   Length is 6, then the Attribute Value field is absent.

   Vendor ID: The IANA-assigned "SMI Network Management Private
   Enterprise Codes" [RFC1700] value.  The value 0, corresponding to
   IETF-adopted attribute values, is used for all AVPs defined within
   this document.  Any vendor wishing to implement its own L2TP
   extensions can use its own Vendor ID along with private Attribute
   values, guaranteeing that they will not collide with any other
   vendor's extensions or future IETF extensions.  Note that there are
   16 bits allocated for the Vendor ID, thus limiting this feature to
   the first 65,535 enterprises.

   Attribute Type: A 2-octet value with a unique interpretation across
   all AVPs defined under a given Vendor ID.

   Attribute Value: This is the actual value as indicated by the Vendor
   ID and Attribute Type.  It follows immediately after the Attribute
   Type field and runs for the remaining octets indicated in the Length
   (i.e., Length minus 6 octets of header).  This field is absent if the
   Length is 6.

   In the event that the 16-bit Vendor ID space is exhausted, vendor-
   specific AVPs with a 32-bit Vendor ID MUST be encapsulated in the
   following manner:











Lau, et al.                 Standards Track                    [Page 31]

RFC 3931                         L2TPv3                       March 2005


                 Figure 5.2: Extended Vendor ID AVP Format

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |M|H| rsvd  |      Length       |               0               |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |              58               |       32-bit Vendor ID     ...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
                                   |        Attribute Type         |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                      Attribute Value                       ...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
                    (until Length is reached)                      |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

   This AVP encodes a vendor-specific AVP with a 32-bit Vendor ID space
   within the Attribute Value field.  Multiple AVPs of this type may
   exist in any message.  The 16-bit Vendor ID MUST be 0, indicating
   that this is an IETF-defined AVP, and the Attribute Type MUST be 58,
   indicating that what follows is a vendor-specific AVP with a 32-bit
   Vendor ID code.  This AVP MAY be hidden (the H bit MAY be 0 or 1).
   The M bit for this AVP MUST be set to 0.  The Length of the AVP is 12
   plus the length of the Attribute Value.

5.2.  Mandatory AVPs and Setting the M Bit



   If the M bit is set on an AVP that is unrecognized by its recipient,
   the session or control connection associated with the control message
   containing the AVP MUST be shut down.  If the control message
   containing the unrecognized AVP is associated with a session (e.g.,
   an ICRQ, ICRP, ICCN, SLI, etc.), then the session MUST be issued a
   CDN with a Result Code of 2 and Error Code of 8 (as defined in
   Section 5.4.2) and shut down.  If the control message containing the
   unrecognized AVP is associated with establishment or maintenance of a
   Control Connection (e.g., SCCRQ, SCCRP, SCCCN, Hello), then the
   associated Control Connection MUST be issued a StopCCN with Result
   Code of 2 and Error Code of 8 (as defined in Section 5.4.2) and shut
   down.  If the M bit is not set on an unrecognized AVP, the AVP MUST
   be ignored when received, processing the control message as if the
   AVP were not present.

   Receipt of an unrecognized AVP that has the M bit set is catastrophic
   to the session or control connection with which it is associated.
   Thus, the M bit should only be set for AVPs that are deemed crucial
   to proper operation of the session or control connection by the
   sender.  AVPs that are considered crucial by the sender may vary by
   application and configured options.  In no case shall a receiver of



Lau, et al.                 Standards Track                    [Page 32]

RFC 3931                         L2TPv3                       March 2005


   an AVP "validate" if the M bit is set on a recognized AVP.  If the
   AVP is recognized (as all AVPs defined in this document MUST be for a
   compliant L2TPv3 specification), then by definition, the M bit is of
   no consequence.

   The sender of an AVP is free to set its M bit to 1 or 0 based on
   whether the configured application strictly requires the value
   contained in the AVP to be recognized or not.  For example,
   "Automatic L2TPv2 Fallback" in Section 4.7.3 requires the setting of
   the M bit on all new L2TPv3 AVPs to zero if fallback to L2TPv2 is
   supported and desired, and 1 if not.

   The M bit is useful as extra assurance for support of critical AVP
   extensions.  However, more explicit methods may be available to
   determine support for a given feature rather than using the M bit
   alone.  For example, if a new AVP is defined in a message for which
   there is always a message reply (i.e., an ICRQ, ICRP, SCCRQ, or SCCRP
   message), rather than simply sending an AVP in the message with the M
   bit set, availability of the extension may be identified by sending
   an AVP in the request message and expecting a corresponding AVP in a
   reply message.  This more explicit method, when possible, is
   preferred.

   The M bit also plays a role in determining whether or not a malformed
   or out-of-range value within an AVP should be ignored or should
   result in termination of a session or control connection (see Section
   7.1 for more details).

5.3.  Hiding of AVP Attribute Values



   The H bit in the header of each AVP provides a mechanism to indicate
   to the receiving peer whether the contents of the AVP are hidden or
   present in cleartext.  This feature can be used to hide sensitive
   control message data such as user passwords, IDs, or other vital
   information.

   The H bit MUST only be set if (1) a shared secret exists between the
   LCCEs and (2) Control Message Authentication is enabled (see Section
   4.3).  If the H bit is set in any AVP(s) in a given control message,
   at least one Random Vector AVP must also be present in the message
   and MUST precede the first AVP having an H bit of 1.










Lau, et al.                 Standards Track                    [Page 33]

RFC 3931                         L2TPv3                       March 2005


   The shared secret between LCCEs is used to derive a unique shared key
   for hiding and unhiding calculations.  The derived shared key is
   obtained via an HMAC-MD5 keyed hash [RFC2104], with the key
   consisting of the shared secret, and with the data being hashed
   consisting of a single octet containing the value 1.

         shared_key = HMAC_MD5 (shared_secret, 1)

   Hiding an AVP value is done in several steps.  The first step is to
   take the length and value fields of the original (cleartext) AVP and
   encode them into the Hidden AVP Subformat, which appears as follows:

                     Figure 5.3: Hidden AVP Subformat

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |   Length of Original Value    |   Original Attribute Value ...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
                  ...              |             Padding ...
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

   Length of Original Attribute Value: This is length of the Original
   Attribute Value to be obscured in octets.  This is necessary to
   determine the original length of the Attribute Value that is lost
   when the additional Padding is added.

   Original Attribute Value: Attribute Value that is to be obscured.

   Padding: Random additional octets used to obscure length of the
   Attribute Value that is being hidden.

   To mask the size of the data being hidden, the resulting subformat
   MAY be padded as shown above.  Padding does NOT alter the value
   placed in the Length of Original Attribute Value field, but does
   alter the length of the resultant AVP that is being created.  For
   example, if an Attribute Value to be hidden is 4 octets in length,
   the unhidden AVP length would be 10 octets (6 + Attribute Value
   length).  After hiding, the length of the AVP would become 6 +
   Attribute Value length + size of the Length of Original Attribute
   Value field + Padding.  Thus, if Padding is 12 octets, the AVP length
   would be 6 + 4 + 2 + 12 = 24 octets.









Lau, et al.                 Standards Track                    [Page 34]

RFC 3931                         L2TPv3                       March 2005


   Next, an MD5 [RFC1321] hash is performed (in network byte order) on
   the concatenation of the following:

         + the 2-octet Attribute number of the AVP
         + the shared key
         + an arbitrary length random vector

   The value of the random vector used in this hash is passed in the
   value field of a Random Vector AVP.  This Random Vector AVP must be
   placed in the message by the sender before any hidden AVPs.  The same
   random vector may be used for more than one hidden AVP in the same
   message, but not for hiding two or more instances of an AVP with the
   same Attribute Type unless the Attribute Values in the two AVPs are
   also identical.  When a different random vector is used for the
   hiding of subsequent AVPs, a new Random Vector AVP MUST be placed in
   the control message before the first AVP to which it applies.

   The MD5 hash value is then XORed with the first 16-octet (or less)
   segment of the Hidden AVP Subformat and placed in the Attribute Value
   field of the Hidden AVP.  If the Hidden AVP Subformat is less than 16
   octets, the Subformat is transformed as if the Attribute Value field
   had been padded to 16 octets before the XOR.  Only the actual octets
   present in the Subformat are modified, and the length of the AVP is
   not altered.

   If the Subformat is longer than 16 octets, a second one-way MD5 hash
   is calculated over a stream of octets consisting of the shared key
   followed by the result of the first XOR.  That hash is XORed with the
   second 16-octet (or less) segment of the Subformat and placed in the
   corresponding octets of the Value field of the Hidden AVP.

   If necessary, this operation is repeated, with the shared key used
   along with each XOR result to generate the next hash to XOR the next
   segment of the value with.

   The hiding method was adapted from [RFC2865], which was taken from
   the "Mixing in the Plaintext" section in the book "Network Security"
   by Kaufman, Perlman and Speciner [KPS].  A detailed explanation of
   the method follows:

   Call the shared key S, the Random Vector RV, and the Attribute Type
   A.  Break the value field into 16-octet chunks p_1, p_2, etc., with
   the last one padded at the end with random data to a 16-octet
   boundary.  Call the ciphertext blocks c_1, c_2, etc.  We will also
   define intermediate values b_1, b_2, etc.






Lau, et al.                 Standards Track                    [Page 35]

RFC 3931                         L2TPv3                       March 2005


      b_1 = MD5 (A + S + RV)   c_1 = p_1 xor b_1
      b_2 = MD5 (S + c_1)      c_2 = p_2 xor b_2
                .                      .
                .                      .
                .                      .
      b_i = MD5 (S + c_i-1)    c_i = p_i xor b_i

   The String will contain c_1 + c_2 +...+ c_i, where "+" denotes
   concatenation.

   On receipt, the random vector is taken from the last Random Vector
   AVP encountered in the message prior to the AVP to be unhidden.  The
   above process is then reversed to yield the original value.

5.4.  AVP Summary



   The following sections contain a list of all L2TP AVPs defined in
   this document.

   Following the name of the AVP is a list indicating the message types
   that utilize each AVP.  After each AVP title follows a short
   description of the purpose of the AVP, a detail (including a graphic)
   of the format for the Attribute Value, and any additional information
   needed for proper use of the AVP.

5.4.1.  General Control Message AVPs



   Message Type (All Messages)

      The Message Type AVP, Attribute Type 0, identifies the control
      message herein and defines the context in which the exact meaning
      of the following AVPs will be determined.

      The Attribute Value field for this AVP has the following format:

       0                   1
       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |         Message Type          |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

      The Message Type is a 2-octet unsigned integer.

      The Message Type AVP MUST be the first AVP in a message,
      immediately following the control message header (defined in
      Section 3.2.1).  See Section 3.1 for the list of defined control
      message types and their identifiers.




Lau, et al.                 Standards Track                    [Page 36]

RFC 3931                         L2TPv3                       March 2005


      The Mandatory (M) bit within the Message Type AVP has special
      meaning.  Rather than an indication as to whether the AVP itself
      should be ignored if not recognized, it is an indication as to
      whether the control message itself should be ignored.  If the M
      bit is set within the Message Type AVP and the Message Type is
      unknown to the implementation, the control connection MUST be
      cleared.  If the M bit is not set, then the implementation may
      ignore an unknown message type.  The M bit MUST be set to 1 for
      all message types defined in this document.  This AVP MUST NOT be
      hidden (the H bit MUST be 0).  The Length of this AVP is 8.

      A vendor-specific control message may be defined by setting the
      Vendor ID of the Message Type AVP to a value other than the IETF
      Vendor ID of 0 (see Section 5.1).  The Message Type AVP MUST still
      be the first AVP in the control message.

   Message Digest (All Messages)

      The Message Digest AVP, Attribute Type 59 is used as an integrity
      and authentication check of the L2TP Control Message header and
      body.

      The Attribute Value field for this AVP has the following format:

       0                   1                   2                   3
       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |  Digest Type  | Message Digest ...
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
                                        ... (16 or 20 octets)         |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

      Digest Type is a one-octet integer indicating the Digest
      calculation algorithm:

         0 HMAC-MD5 [RFC2104]
         1 HMAC-SHA-1 [RFC2104]



      Digest Type 0 (HMAC-MD5) MUST be supported, while Digest Type 1
      (HMAC-SHA-1) SHOULD be supported.

      The Message Digest is of variable length and contains the result
      of the control message authentication and integrity calculation.
      For Digest Type 0 (HMAC-MD5), the length of the digest MUST be 16



Lau, et al.                 Standards Track                    [Page 37]

RFC 3931                         L2TPv3                       March 2005


      bytes.  For Digest Type 1 (HMAC-SHA-1) the length of the digest
      MUST be 20 bytes.

      If Control Message Authentication is enabled, at least one Message
      Digest AVP MUST be present in all messages and MUST be placed
      immediately after the Message Type AVP.  This forces the Message
      Digest AVP to begin at a well-known and fixed offset.  A second
      Message Digest AVP MAY be present in a message and MUST be placed
      directly after the first Message Digest AVP.

      The shared secret between LCCEs is used to derive a unique shared
      key for Control Message Authentication calculations.  The derived
      shared key is obtained via an HMAC-MD5 keyed hash [RFC2104], with
      the key consisting of the shared secret, and with the data being
      hashed consisting of a single octet containing the value 2.

         shared_key = HMAC_MD5 (shared_secret, 2)

      Calculation of the Message Digest is as follows for all messages
      other than the SCCRQ (where "+" refers to concatenation):

         Message Digest = HMAC_Hash (shared_key, local_nonce +
                                     remote_nonce + control_message)

         HMAC_Hash: HMAC Hashing algorithm identified by the Digest Type
         (MD5 or SHA1)

         local_nonce: Nonce chosen locally and advertised to the remote
         LCCE.

         remote_nonce: Nonce received from the remote LCCE

         (The local_nonce and remote_nonce are advertised via the
         Control Message Authentication Nonce AVP, also defined in this
         section.)

         shared_key: Derived shared key for this control connection

         control_message: The entire contents of the L2TP control
         message, including the control message header and all AVPs.
         Note that the control message header in this case begins after
         the all-zero Session ID when running over IP (see Section
         4.1.1.2), and after the UDP header when running over UDP (see
         Section 4.1.2.1).

      When calculating the Message Digest, the Message Digest AVP MUST
      be present within the control message with the Digest Type set to
      its proper value, but the Message Digest itself set to zeros.



Lau, et al.                 Standards Track                    [Page 38]

RFC 3931                         L2TPv3                       March 2005


      When receiving a control message, the contents of the Message
      Digest AVP MUST be compared against the expected digest value
      based on local calculation.  This is done by performing the same
      digest calculation above, with the local_nonce and remote_nonce
      reversed.  This message authenticity and integrity checking MUST
      be performed before utilizing any information contained within the
      control message.  If the calculation fails, the message MUST be
      dropped.

      The SCCRQ has special treatment as it is the initial message
      commencing a new control connection.  As such, there is only one
      nonce available.  Since the nonce is present within the message
      itself as part of the Control Message Authentication Nonce AVP,
      there is no need to use it in the calculation explicitly.
      Calculation of the SCCRQ Message Digest is performed as follows:

         Message Digest = HMAC_Hash (shared_key, control_message)

      To allow for graceful switchover to a new shared secret or hash
      algorithm, two Message Digest AVPs MAY be present in a control
      message, and two shared secrets MAY be configured for a given
      LCCE.  If two Message Digest AVPs are received in a control
      message, the message MUST be accepted if either Message Digest is
      valid.  If two shared secrets are configured, each (separately)
      MUST be used for calculating a digest to be compared to the
      Message Digest(s) received.  When calculating a digest for a
      control message, the Value field for both of the Message Digest
      AVPs MUST be set to zero.

      This AVP MUST NOT be hidden (the H bit MUST be 0).  The M bit for
      this AVP SHOULD be set to 1, but MAY vary (see Section 5.2).  The
      Length is 23 for Digest Type 1 (HMAC-MD5), and 27 for Digest Type
      2 (HMAC-SHA-1).

   Control Message Authentication Nonce (SCCRQ, SCCRP)

      The Control Message Authentication Nonce AVP, Attribute Type 73,
      MUST contain a cryptographically random value [RFC1750].  This
      value is used for Control Message Authentication.

      The Attribute Value field for this AVP has the following format:

       0                   1                   2                   3
       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-++-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      | Nonce ... (arbitrary number of octets)
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-++-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+




Lau, et al.                 Standards Track                    [Page 39]

RFC 3931                         L2TPv3                       March 2005


      The Nonce is of arbitrary length, though at least 16 octets is
      recommended.  The Nonce contains the random value for use in the
      Control Message Authentication hash calculation (see Message
      Digest AVP definition in this section).

      If Control Message Authentication is enabled, this AVP MUST be
      present in the SCCRQ and SCCRP messages.

      This AVP MUST NOT be hidden (the H bit MUST be 0).  The M bit for
      this AVP SHOULD be set to 1, but MAY vary (see Section 5.2).  The
      Length of this AVP is 6 plus the length of the Nonce.

   Random Vector (All Messages)

      The Random Vector AVP, Attribute Type 36, MUST contain a
      cryptographically random value [RFC1750].  This value is used for
      AVP Hiding.

      The Attribute Value field for this AVP has the following format:

       0                   1                   2                   3
       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-++-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      | Random Octet String ... (arbitrary number of octets)
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-++-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

      The Random Octet String is of arbitrary length, though at least 16
      octets is recommended.  The string contains the random vector for
      use in computing the MD5 hash to retrieve or hide the Attribute
      Value of a hidden AVP (see Section 5.3).

      More than one Random Vector AVP may appear in a message, in which
      case a hidden AVP uses the Random Vector AVP most closely
      preceding it.  As such, at least one Random Vector AVP MUST
      precede the first AVP with the H bit set.

      This AVP MUST NOT be hidden (the H bit MUST be 0).  The M bit for
      this AVP SHOULD be set to 1, but MAY vary (see Section 5.2).  The
      Length of this AVP is 6 plus the length of the Random Octet
      String.

5.4.2.  Result and Error Codes



   Result Code (StopCCN, CDN)

      The Result Code AVP, Attribute Type 1, indicates the reason for
      terminating the control connection or session.




Lau, et al.                 Standards Track                    [Page 40]

RFC 3931                         L2TPv3                       March 2005


      The Attribute Value field for this AVP has the following format:

       0                   1                   2                   3
       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |          Result Code          |     Error Code (optional)     |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      | Error Message ... (optional, arbitrary number of octets)      |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

      The Result Code is a 2-octet unsigned integer.  The optional Error
      Code is a 2-octet unsigned integer.  An optional Error Message can
      follow the Error Code field.  Presence of the Error Code and
      Message is indicated by the AVP Length field.  The Error Message
      contains an arbitrary string providing further (human-readable)
      text associated with the condition.  Human-readable text in all
      error messages MUST be provided in the UTF-8 charset [RFC3629]
      using the Default Language [RFC2277].

      This AVP MUST NOT be hidden (the H bit MUST be 0).  The M bit for
      this AVP SHOULD be set to 1, but MAY vary (see Section 5.2).  The
      Length is 8 if there is no Error Code or Message, 10 if there is
      an Error Code and no Error Message, or 10 plus the length of the
      Error Message if there is an Error Code and Message.

      Defined Result Code values for the StopCCN message are as follows:

         0 - Reserved.
         1 - General request to clear control connection.
         2 - General error, Error Code indicates the problem.
         3 - Control connection already exists.
         4 - Requester is not authorized to establish a control
             connection.
         5 - The protocol version of the requester is not supported,
             Error Code indicates highest version supported.
         6 - Requester is being shut down.
         7 - Finite state machine error or timeout

      General Result Code values for the CDN message are as follows:

         0 - Reserved.
         1 - Session disconnected due to loss of carrier or
             circuit disconnect.
         2 - Session disconnected for the reason indicated in Error
             Code.
         3 - Session disconnected for administrative reasons.
         4 - Session establishment failed due to lack of appropriate
             facilities being available (temporary condition).



Lau, et al.                 Standards Track                    [Page 41]

RFC 3931                         L2TPv3                       March 2005


         5 - Session establishment failed due to lack of appropriate
             facilities being available (permanent condition).
        13 - Session not established due to losing tie breaker.
        14 - Session not established due to unsupported PW type.
        15 - Session not established, sequencing required without
             valid L2-Specific Sublayer.
        16 - Finite state machine error or timeout.

      Additional service-specific Result Codes are defined outside this
      document.

      The Error Codes defined below pertain to types of errors that are
      not specific to any particular L2TP request, but rather to
      protocol or message format errors.  If an L2TP reply indicates in
      its Result Code that a General Error occurred, the General Error
      value should be examined to determine what the error was.  The
      currently defined General Error codes and their meanings are as
      follows:

      0 - No General Error.
      1 - No control connection exists yet for this pair of LCCEs.
      2 - Length is wrong.
      3 - One of the field values was out of range.
      4 - Insufficient resources to handle this operation now.
      5 - Invalid Session ID.
      6 - A generic vendor-specific error occurred.
      7 - Try another.  If initiator is aware of other possible
          responder destinations, it should try one of them.  This can
          be used to guide an LAC or LNS based on policy.
      8 - The session or control connection was shut down due to receipt
          of an unknown AVP with the M bit set (see Section 5.2).  The
          Error Message SHOULD contain the attribute of the offending
          AVP in (human-readable) text form.
      9 - Try another directed.  If an LAC or LNS is aware of other
          possible destinations, it should inform the initiator of the
          control connection or session.  The Error Message MUST contain
          a comma-separated list of addresses from which the initiator
          may choose.  If the L2TP data channel runs over IPv4, then
          this would be a comma-separated list of IP addresses in the
          canonical dotted-decimal format (e.g., "192.0.2.1, 192.0.2.2,
          192.0.2.3") in the UTF-8 charset [RFC3629] using the Default
          Language [RFC2277].  If there are no servers for the LAC or
          LNS to suggest, then Error Code 7 should be used.  For IPv4,
          the delimiter between addresses MUST be precisely a single
          comma and a single space.  For IPv6, each literal address MUST
          be enclosed in "[" and "]" characters, following the encoding
          described in [RFC2732].




Lau, et al.                 Standards Track                    [Page 42]

RFC 3931                         L2TPv3                       March 2005


      When a General Error Code of 6 is used, additional information
      about the error SHOULD be included in the Error Message field.  A
      vendor-specific AVP MAY be sent to more precisely detail a
      vendor-specific problem.

5.4.3.  Control Connection Management AVPs



   Control Connection Tie Breaker (SCCRQ)

      The Control Connection Tie Breaker AVP, Attribute Type 5,
      indicates that the sender desires a single control connection to
      exist between a given pair of LCCEs.

      The Attribute Value field for this AVP has the following format:

       0                   1                   2                   3
       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      | Control Connection Tie Breaker Value ...
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
                                                 ... (64 bits)        |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

      The Control Connection Tie Breaker Value is an 8-octet random
      value that is used to choose a single control connection when two
      LCCEs request a control connection concurrently.  The recipient of
      a SCCRQ must check to see if a SCCRQ has been sent to the peer; if
      so, a tie has been detected.  In this case, the LCCE must compare
      its Control Connection Tie Breaker value with the one received in
      the SCCRQ.  The lower value "wins", and the "loser" MUST discard
      its control connection.  A StopCCN SHOULD be sent by the winner as
      an explicit rejection for the losing SCCRQ.  In the case in which
      a tie breaker is present on both sides and the value is equal,
      both sides MUST discard their control connections and restart
      control connection negotiation with a new, random tie breaker
      value.

      If a tie breaker is received and an outstanding SCCRQ has no tie
      breaker value, the initiator that included the Control Connection
      Tie Breaker AVP "wins".  If neither side issues a tie breaker,
      then two separate control connections are opened.

      Applications that employ a distinct and well-known initiator have
      no need for tie breaking, and MAY omit this AVP or disable tie
      breaking functionality.  Applications that require tie breaking
      also require that an LCCE be uniquely identifiable upon receipt of
      an SCCRQ.  For L2TP over IP, this MUST be accomplished via the
      Router ID AVP.



Lau, et al.                 Standards Track                    [Page 43]

RFC 3931                         L2TPv3                       March 2005


      Note that in [RFC2661], this AVP is referred to as the "Tie
      Breaker AVP" and is applicable only to a control connection.  In
      L2TPv3, the AVP serves the same purpose of tie breaking, but is
      applicable to a control connection or a session.  The Control
      Connection Tie Breaker AVP (present only in Control Connection
      messages) and Session Tie Breaker AVP (present only in Session
      messages), are described separately in this document, but share
      the same Attribute type of 5.

      This AVP MUST NOT be hidden (the H bit MUST be 0).  The M bit for
      this AVP SHOULD be set to 1, but MAY vary (see Section 5.2).  The
      length of this AVP is 14.

   Host Name (SCCRQ, SCCRP)

      The Host Name AVP, Attribute Type 7, indicates the name of the
      issuing LAC or LNS, encoded in the US-ASCII charset.

      The Attribute Value field for this AVP has the following format:

       0                   1                   2                   3
       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      | Host Name ... (arbitrary number of octets)
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

      The Host Name is of arbitrary length, but MUST be at least 1
      octet.

      This name should be as broadly unique as possible; for hosts
      participating in DNS [RFC1034], a host name with fully qualified
      domain would be appropriate.  The Host Name AVP and/or Router ID
      AVP MUST be used to identify an LCCE as described in Section 3.3.

      This AVP MUST NOT be hidden (the H bit MUST be 0).  The M bit for
      this AVP SHOULD be set to 1, but MAY vary (see Section 5.2).  The
      Length of this AVP is 6 plus the length of the Host Name.

   Router ID (SCCRQ, SCCRP)

      The Router ID AVP, Attribute Type 60, is an identifier used to
      identify an LCCE for control connection setup, tie breaking,
      and/or tunnel authentication.








Lau, et al.                 Standards Track                    [Page 44]

RFC 3931                         L2TPv3                       March 2005


      The Attribute Value field for this AVP has the following format:

       0                   1                   2                   3
       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                      Router Identifier                        |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

      The Router Identifier is a 4-octet unsigned integer.  Its value is
      unique for a given LCCE, per Section 8.1 of [RFC2072].  The Host
      Name AVP and/or Router ID AVP MUST be used to identify an LCCE as
      described in Section 3.3.

      Implementations MUST NOT assume that Router Identifier is a valid
      IP address.  The Router Identifier for L2TP over IPv6 can be
      obtained from an IPv4 address (if available) or via unspecified
      implementation-specific means.

      This AVP MUST NOT be hidden (the H bit MUST be 0).  The M bit for
      this AVP SHOULD be set to 1, but MAY vary (see Section 5.2).  The
      Length of this AVP is 10.

   Vendor Name (SCCRQ, SCCRP)

      The Vendor Name AVP, Attribute Type 8, contains a vendor-specific
      (possibly human-readable) string describing the type of LAC or LNS
      being used.

      The Attribute Value field for this AVP has the following format:

       0                   1                   2                   3
       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |  Vendor Name ... (arbitrary number of octets)
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

      The Vendor Name is the indicated number of octets representing the
      vendor string.  Human-readable text for this AVP MUST be provided
      in the US-ASCII charset [RFC1958, RFC2277].

      This AVP MAY be hidden (the H bit MAY be 0 or 1).  The M bit for
      this AVP SHOULD be set to 0, but MAY vary (see Section 5.2).  The
      Length (before hiding) of this AVP is 6 plus the length of the
      Vendor Name.







Lau, et al.                 Standards Track                    [Page 45]

RFC 3931                         L2TPv3                       March 2005


   Assigned Control Connection ID (SCCRQ, SCCRP, StopCCN)

      The Assigned Control Connection ID AVP, Attribute Type 61,
      contains the ID being assigned to this control connection by the
      sender.

      The Attribute Value field for this AVP has the following format:

       0                   1                   2                   3
       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                Assigned Control Connection ID                 |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

      The Assigned Control Connection ID is a 4-octet non-zero unsigned
      integer.

      The Assigned Control Connection ID AVP establishes the identifier
      used to multiplex and demultiplex multiple control connections
      between a pair of LCCEs.  Once the Assigned Control Connection ID
      AVP has been received by an LCCE, the Control Connection ID
      specified in the AVP MUST be included in the Control Connection ID
      field of all control packets sent to the peer for the lifetime of
      the control connection.  Before the Assigned Control Connection ID
      AVP is received from a peer, all control messages MUST be sent to
      that peer with a Control Connection ID value of 0 in the header.
      Because a Control Connection ID value of 0 is used in this special
      manner, the zero value MUST NOT be sent as an Assigned Control
      Connection ID value.

      Under certain circumstances, an LCCE may need to send a StopCCN to
      a peer without having yet received an Assigned Control Connection
      ID AVP from the peer (i.e., SCCRQ sent, no SCCRP received yet).
      In this case, the Assigned Control Connection ID AVP that had been
      sent to the peer earlier (i.e., in the SCCRQ) MUST be sent as the
      Assigned Control Connection ID AVP in the StopCCN.  This policy
      allows the peer to try to identify the appropriate control
      connection via a reverse lookup.

      This AVP MAY be hidden (the H bit MAY be 0 or 1).  The M bit for
      this AVP SHOULD be set to 1, but MAY vary (see Section 5.2).  The
      Length (before hiding) of this AVP is 10.

   Receive Window Size (SCCRQ, SCCRP)

      The Receive Window Size AVP, Attribute Type 10, specifies the
      receive window size being offered to the remote peer.




Lau, et al.                 Standards Track                    [Page 46]

RFC 3931                         L2TPv3                       March 2005


      The Attribute Value field for this AVP has the following format:

       0                   1
       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |         Window Size           |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

      The Window Size is a 2-octet unsigned integer.

      If absent, the peer must assume a Window Size of 4 for its
      transmit window.

      The remote peer may send the specified number of control messages
      before it must wait for an acknowledgment.  See Section 4.2 for
      more information on reliable control message delivery.

      This AVP MUST NOT be hidden (the H bit MUST be 0).  The M bit for
      this AVP SHOULD be set to 1, but MAY vary (see Section 5.2).  The
      Length of this AVP is 8.

   Pseudowire Capabilities List (SCCRQ, SCCRP)

      The Pseudowire Capabilities List (PW Capabilities List) AVP,
      Attribute Type 62, indicates the L2 payload types the sender can
      support.  The specific payload type of a given session is
      identified by the Pseudowire Type AVP.

      The Attribute Value field for this AVP has the following format:

       0                   1                   2                   3
       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |           PW Type 0           |             ...               |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |              ...              |          PW Type N            |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

      Defined PW types that may appear in this list are managed by IANA
      and will appear in associated pseudowire-specific documents for
      each PW type.

      If a sender includes a given PW type in the PW Capabilities List
      AVP, the sender assumes full responsibility for supporting that
      particular payload, such as any payload-specific AVPs, L2-Specific
      Sublayer, or control messages that may be defined in the
      appropriate companion document.




Lau, et al.                 Standards Track                    [Page 47]

RFC 3931                         L2TPv3                       March 2005


      This AVP MAY be hidden (the H bit MAY be 0 or 1).  The M bit for
      this AVP SHOULD be set to 1, but MAY vary (see Section 5.2).  The
      Length (before hiding) of this AVP is 8 octets with one PW type
      specified, plus 2 octets for each additional PW type.

   Preferred Language (SCCRQ, SCCRP)

      The Preferred Language AVP, Attribute Type 72, provides a method
      for an LCCE to indicate to the peer the language in which human-
      readable messages it sends SHOULD be composed.  This AVP contains
      a single language tag or language range [RFC3066].

      The Attribute Value field for this AVP has the following format:

      0                   1                   2                   3
      0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |  Preferred Language... (arbitrary number of octets)
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

      The Preferred Language is the indicated number of octets
      representing the language tag or language range, encoded in the
      US-ASCII charset.

      It is not required to send a Preferred Language AVP.  If (1) an
      LCCE does not signify a language preference by the inclusion of
      this AVP in the SCCRQ or SCCRP, (2) the Preferred Language AVP is
      unrecognized, or (3) the requested language is not supported by
      the peer LCCE, the default language [RFC2277] MUST be used for all
      internationalized strings sent by the peer.

      This AVP MAY be hidden (the H bit MAY be 0 or 1).  The M bit for
      this AVP SHOULD be set to 0, but MAY vary (see Section 5.2).  The
      Length (before hiding) of this AVP is 6 plus the length of the
      Preferred Language.

5.4.4.  Session Management AVPs



   Local Session ID (ICRQ, ICRP, ICCN, OCRQ, OCRP, OCCN, CDN, WEN, SLI)

      The Local Session ID AVP (analogous to the Assigned Session ID in
      L2TPv2), Attribute Type 63, contains the identifier being assigned
      to this session by the sender.








Lau, et al.                 Standards Track                    [Page 48]

RFC 3931                         L2TPv3                       March 2005


      The Attribute Value field for this AVP has the following format:

       0                   1                   2                   3
       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                       Local Session ID                        |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

      The Local Session ID is a 4-octet non-zero unsigned integer.

      The Local Session ID AVP establishes the two identifiers used to
      multiplex and demultiplex sessions between two LCCEs.  Each LCCE
      chooses any free value it desires, and sends it to the remote LCCE
      using this AVP.  The remote LCCE MUST then send all data packets
      associated with this session using this value.  Additionally, for
      all session-oriented control messages sent after this AVP is
      received (e.g., ICRP, ICCN, CDN, SLI, etc.), the remote LCCE MUST
      echo this value in the Remote Session ID AVP.

      Note that a Session ID value is unidirectional.  Because each LCCE
      chooses its Session ID independent of its peer LCCE, the value
      does not have to match in each direction for a given session.

      See Section 4.1 for additional information about the Session ID.

      This AVP MAY be hidden (the H bit MAY be 0 or 1).  The M bit for
      this AVP SHOULD be 1 set to 1, but MAY vary (see Section 5.2).
      The Length (before hiding) of this AVP is 10.

   Remote Session ID (ICRQ, ICRP, ICCN, OCRQ, OCRP, OCCN, CDN, WEN, SLI)

      The Remote Session ID AVP, Attribute Type 64, contains the
      identifier that was assigned to this session by the peer.

      The Attribute Value field for this AVP has the following format:

       0                   1                   2                   3
       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                      Remote Session ID                        |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

      The Remote Session ID is a 4-octet non-zero unsigned integer.

      The Remote Session ID AVP MUST be present in all session-level
      control messages.  The AVP's value echoes the session identifier
      advertised by the peer via the Local Session ID AVP.  It is the
      same value that will be used in all transmitted data messages by



Lau, et al.                 Standards Track                    [Page 49]

RFC 3931                         L2TPv3                       March 2005


      this side of the session.  In most cases, this identifier is
      sufficient for the peer to look up session-level context for this
      control message.

      When a session-level control message must be sent to the peer
      before the Local Session ID AVP has been received, the value of
      the Remote Session ID AVP MUST be set to zero.  Additionally, the
      Local Session ID AVP (sent in a previous control message for this
      session) MUST be included in the control message.  The peer must
      then use the Local Session ID AVP to perform a reverse lookup to
      find its session context.  Session-level control messages defined
      in this document that might be subject to a reverse lookup by a
      receiving peer include the CDN, WEN, and SLI.

      This AVP MAY be hidden (the H bit MAY be 0 or 1).  The M bit for
      this AVP SHOULD be set to 1, but MAY vary (see Section 5.2).  The
      Length (before hiding) of this AVP is 10.

   Assigned Cookie (ICRQ, ICRP, OCRQ, OCRP)

      The Assigned Cookie AVP, Attribute Type 65, contains the Cookie
      value being assigned to this session by the sender.

      The Attribute Value field for this AVP has the following format:

       0                   1                   2                   3
       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |               Assigned Cookie (32 or 64 bits) ...
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

      The Assigned Cookie is a 4-octet or 8-octet random value.

      The Assigned Cookie AVP contains the value used to check the
      association of a received data message with the session identified
      by the Session ID.  All data messages sent to a peer MUST use the
      Assigned Cookie sent by the peer in this AVP.  The value's length
      (0, 32, or 64 bits) is obtained by the length of the AVP.

      A missing Assigned Cookie AVP or Assigned Cookie Value of zero
      length indicates that the Cookie field should not be present in
      any data packets sent to the LCCE sending this AVP.

      See Section 4.1 for additional information about the Assigned
      Cookie.






Lau, et al.                 Standards Track                    [Page 50]

RFC 3931                         L2TPv3                       March 2005


      This AVP MAY be hidden (the H bit MAY be 0 or 1).  The M bit for
      this AVP SHOULD be set to 1, but MAY vary (see Section 5.2).  The
      Length (before hiding) of this AVP may be 6, 10, or 14 octets.

   Serial Number (ICRQ, OCRQ)

      The Serial Number AVP, Attribute Type 15, contains an identifier
      assigned by the LAC or LNS to this session.

      The Attribute Value field for this AVP has the following format:

       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                        Serial Number                          |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

      The Serial Number is a 32-bit value.

      The Serial Number is intended to be an easy reference for
      administrators on both ends of a control connection to use when
      investigating session failure problems.  Serial Numbers should be
      set to progressively increasing values, which are likely to be
      unique for a significant period of time across all interconnected
      LNSs and LACs.

      Note that in RFC 2661, this value was referred to as the "Call
      Serial Number AVP".  It serves the same purpose and has the same
      attribute value and composition.

      This AVP MAY be hidden (the H bit MAY be 0 or 1).  The M bit for
      this AVP SHOULD be set to 0, but MAY vary (see Section 5.2).  The
      Length (before hiding) of this AVP is 10.

   Remote End ID (ICRQ, OCRQ)

      The Remote End ID AVP, Attribute Type 66, contains an identifier
      used to bind L2TP sessions to a given circuit, interface, or
      bridging instance.  It also may be used to detect session-level
      ties.

      The Attribute Value field for this AVP has the following format:

       0                   1                   2                   3
       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      | Remote End Identifier ... (arbitrary number of octets)
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+




Lau, et al.                 Standards Track                    [Page 51]

RFC 3931                         L2TPv3                       March 2005


      The Remote End Identifier field is a variable-length field whose
      value is unique for a given LCCE peer, as described in Section
      3.3.

      A session-level tie is detected if an LCCE receives an ICRQ or
      OCRQ with an End ID AVP whose value matches that which was just
      sent in an outgoing ICRQ or OCRQ to the same peer.  If the two
      values match, an LCCE recognizes that a tie exists (i.e., both
      LCCEs are attempting to establish sessions for the same circuit).
      The tie is broken by the Session Tie Breaker AVP.

      By default, the LAC-LAC cross-connect application (see Section
      2(b)) of L2TP over an IP network MUST utilize the Router ID AVP
      and Remote End ID AVP to associate a circuit to an L2TP session.
      Other AVPs MAY be used for LCCE or circuit identification as
      specified in companion documents.

      This AVP MAY be hidden (the H bit MAY be 0 or 1).  The M bit for
      this AVP SHOULD be set to 1, but MAY vary (see Section 5.2).  The
      Length (before hiding) of this AVP is 6 plus the length of the
      Remote End Identifier value.

   Session Tie Breaker (ICRQ, OCRQ)

      The Session Tie Breaker AVP, Attribute Type 5, is used to break
      ties when two peers concurrently attempt to establish a session
      for the same circuit.

      The Attribute Value field for this AVP has the following format:

      0                   1                   2                   3
      0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      | Session Tie Breaker Value ...
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
                                                 ... (64 bits)        |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

      The Session Tie Breaker Value is an 8-octet random value that is
      used to choose a session when two LCCEs concurrently request a
      session for the same circuit.  A tie is detected by examining the
      peer's identity (described in Section 3.3) plus the per-session
      shared value communicated via the End ID AVP.  In the case of a
      tie, the recipient of an ICRQ or OCRQ must compare the received
      tie breaker value with the one that it sent earlier.  The LCCE
      with the lower value "wins" and MUST send a CDN with result code
      set to 13 (as defined in Section 5.4.2) in response to the losing
      ICRQ or OCRQ.  In the case in which a tie is detected, tie



Lau, et al.                 Standards Track                    [Page 52]

RFC 3931                         L2TPv3                       March 2005


      breakers are sent by both sides, and the tie breaker values are
      equal, both sides MUST discard their sessions and restart session
      negotiation with new random tie breaker values.

      If a tie is detected but only one side sends a Session Tie Breaker
      AVP, the session initiator that included the Session Tie Breaker
      AVP "wins".  If neither side issues a tie breaker, then both sides
      MUST tear down the session.

      This AVP MUST NOT be hidden (the H bit MUST be 0).  The M bit for
      this AVP SHOULD be set to 1, but MAY vary (see Section 5.2).  The
      Length of this AVP is 14.

   Pseudowire Type (ICRQ, OCRQ)

      The Pseudowire Type (PW Type) AVP, Attribute Type 68, indicates
      the L2 payload type of the packets that will be tunneled using
      this L2TP session.

      The Attribute Value field for this AVP has the following format:

       0                   1
       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |           PW Type             |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

      A peer MUST NOT request an incoming or outgoing call with a PW
      Type AVP specifying a value not advertised in the PW Capabilities
      List AVP it received during control connection establishment.
      Attempts to do so MUST result in the call being rejected via a CDN
      with the Result Code set to 14 (see Section 5.4.2).

      This AVP MAY be hidden (the H bit MAY be 0 or 1).  The M bit for
      this AVP SHOULD be set to 1, but MAY vary (see Section 5.2).  The
      Length (before hiding) of this AVP is 8.

   L2-Specific Sublayer (ICRQ, ICRP, ICCN, OCRQ, OCRP, OCCN)

      The L2-Specific Sublayer AVP, Attribute Type 69, indicates the
      presence and format of the L2-Specific Sublayer the sender of this
      AVP requires on all incoming data packets for this L2TP session.

       0                   1
       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |   L2-Specific Sublayer Type   |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+



Lau, et al.                 Standards Track                    [Page 53]

RFC 3931                         L2TPv3                       March 2005


      The L2-Specific Sublayer Type is a 2-octet unsigned integer with
      the following values defined in this document:

         0 - There is no L2-Specific Sublayer present.
         1 - The Default L2-Specific Sublayer (defined in Section 4.6)
             is used.

      If this AVP is received and has a value other than zero, the
      receiving LCCE MUST include the identified L2-Specific Sublayer in
      its outgoing data messages.  If the AVP is not received, it is
      assumed that there is no sublayer present.

      This AVP MAY be hidden (the H bit MAY be 0 or 1).  The M bit for
      this AVP SHOULD be set to 1, but MAY vary (see Section 5.2).  The
      Length (before hiding) of this AVP is 8.

   Data Sequencing (ICRQ, ICRP, ICCN, OCRQ, OCRP, OCCN)

      The Data Sequencing AVP, Attribute Type 70, indicates that the
      sender requires some or all of the data packets that it receives
      to be sequenced.

      The Attribute Value field for this AVP has the following format:

       0                   1
       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |     Data Sequencing Level     |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

      The Data Sequencing Level is a 2-octet unsigned integer indicating
      the degree of incoming data traffic that the sender of this AVP
      wishes to be marked with sequence numbers.

      Defined Data Sequencing Levels are as follows:

         0 - No incoming data packets require sequencing.
         1 - Only non-IP data packets require sequencing.
         2 - All incoming data packets require sequencing.

      If a Data Sequencing Level of 0 is specified, there is no need to
      send packets with sequence numbers.  If sequence numbers are sent,
      they will be ignored upon receipt.  If no Data Sequencing AVP is
      received, a Data Sequencing Level of 0 is assumed.

      If a Data Sequencing Level of 1 is specified, only non-IP traffic
      carried within the tunneled L2 frame should have sequence numbers
      applied.  Non-IP traffic here refers to any packets that cannot be



Lau, et al.                 Standards Track                    [Page 54]

RFC 3931                         L2TPv3                       March 2005


      classified as an IP packet within their respective L2 framing
      (e.g., a PPP control packet or NETBIOS frame encapsulated by Frame
      Relay before being tunneled).  All traffic that can be classified
      as IP MUST be sent with no sequencing (i.e., the S bit in the L2-
      Specific Sublayer is set to zero).  If a packet is unable to be
      classified at all (e.g., because it has been compressed or
      encrypted at layer 2) or if an implementation is unable to perform
      such classification within L2 frames, all packets MUST be provided
      with sequence numbers (essentially falling back to a Data
      Sequencing Level of 2).

      If a Data Sequencing Level of 2 is specified, all traffic MUST be
      sequenced.

      Data sequencing may only be requested when there is an L2-Specific
      Sublayer present that can provide sequence numbers.  If sequencing
      is requested without requesting a L2-Specific Sublayer AVP, the
      session MUST be disconnected with a Result Code of 15 (see Section
      5.4.2).

      This AVP MAY be hidden (the H bit MAY be 0 or 1).  The M bit for
      this AVP SHOULD be set to 1, but MAY vary (see Section 5.2).  The
      Length (before hiding) of this AVP is 8.

   Tx Connect Speed (ICRQ, ICRP, ICCN, OCRQ, OCRP, OCCN)

      The Tx Connect Speed BPS AVP, Attribute Type 74, contains the
      speed of the facility chosen for the connection attempt.

      The Attribute Value field for this AVP has the following format:

       0                   1                   2                   3
       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                      Connect Speed in bps...
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
                        ...Connect Speed in bps (64 bits)             |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

      The Tx Connect Speed BPS is an 8-octet value indicating the speed
      in bits per second.  A value of zero indicates that the speed is
      indeterminable or that there is no physical point-to-point link.

      When the optional Rx Connect Speed AVP is present, the value in
      this AVP represents the transmit connect speed from the
      perspective of the LAC (i.e., data flowing from the LAC to the
      remote system).  When the optional Rx Connect Speed AVP is NOT
      present, the connection speed between the remote system and LAC is



Lau, et al.                 Standards Track                    [Page 55]

RFC 3931                         L2TPv3                       March 2005


      assumed to be symmetric and is represented by the single value in
      this AVP.

      This AVP MAY be hidden (the H bit MAY be 0 or 1).  The M bit for
      this AVP SHOULD be set to 0, but MAY vary (see Section 5.2).  The
      Length (before hiding) of this AVP is 14.

   Rx Connect Speed (ICRQ, ICRP, ICCN, OCRQ, OCRP, OCCN)

      The Rx Connect Speed AVP, Attribute Type 75, represents the speed
      of the connection from the perspective of the LAC (i.e., data
      flowing from the remote system to the LAC).

      The Attribute Value field for this AVP has the following format:

       0                   1                   2                   3
       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                      Connect Speed in bps...
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
                        ...Connect Speed in bps (64 bits)             |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

      Connect Speed BPS is an 8-octet value indicating the speed in bits
      per second.  A value of zero indicates that the speed is
      indeterminable or that there is no physical point-to-point link.

      Presence of this AVP implies that the connection speed may be
      asymmetric with respect to the transmit connect speed given in the
      Tx Connect Speed AVP.

      This AVP MAY be hidden (the H bit MAY be 0 or 1).  The M bit for
      this AVP SHOULD be set to 0, but MAY vary (see Section 5.2).  The
      Length (before hiding) of this AVP is 14.

   Physical Channel ID (ICRQ, ICRP, OCRP)

      The Physical Channel ID AVP, Attribute Type 25, contains the
      vendor-specific physical channel number used for a call.

      The Attribute Value field for this AVP has the following format:

       0                   1                   2                   3
       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                      Physical Channel ID                      |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+




Lau, et al.                 Standards Track                    [Page 56]

RFC 3931                         L2TPv3                       March 2005


      Physical Channel ID is a 4-octet value intended to be used for
      logging purposes only.

      This AVP MAY be hidden (the H bit MAY be 0 or 1).  The M bit for
      this AVP SHOULD be set to 0, but MAY vary (see Section 5.2).  The
      Length (before hiding) of this AVP is 10.

5.4.5.  Circuit Status AVPs



   Circuit Status (ICRQ, ICRP, ICCN, OCRQ, OCRP, OCCN, SLI)

      The Circuit Status AVP, Attribute Type 71, indicates the initial
      status of or a status change in the circuit to which the session
      is bound.

      The Attribute Value field for this AVP has the following format:

       0                   1
       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |         Reserved          |N|A|
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

      The A (Active) bit indicates whether the circuit is
      up/active/ready (1) or down/inactive/not-ready (0).

      The N (New) bit indicates whether the circuit status indication is
      for a new circuit (1) or an existing circuit (0).  Links that have
      a similar mechanism available (e.g., Frame Relay) MUST map the
      setting of this bit to the associated signaling for that link.
      Otherwise, the New bit SHOULD still be set the first time the L2TP
      session is established after provisioning.

      The remaining bits are reserved for future use.  Reserved bits
      MUST be set to 0 when sending and ignored upon receipt.

      The Circuit Status AVP is used to advertise whether a circuit or
      interface bound to an L2TP session is up and ready to send and/or
      receive traffic.  Different circuit types have different names for
      status types.  For example, HDLC primary and secondary stations
      refer to a circuit as being "Receive Ready" or "Receive Not
      Ready", while Frame Relay refers to a circuit as "Active" or
      "Inactive".  This AVP adopts the latter terminology, though the
      concept remains the same regardless of the PW type for the L2TP
      session.






Lau, et al.                 Standards Track                    [Page 57]

RFC 3931                         L2TPv3                       March 2005


      In the simplest case, the circuit to which this AVP refers is a
      single physical interface, port, or circuit, depending on the
      application and the session setup.  The status indication in this
      AVP may then be used to provide simple ILMI interworking for a
      variety of circuit types.  For virtual or multipoint interfaces,
      the Circuit Status AVP is still utilized, but in this case, it
      refers to the state of an internal structure or a logical set of
      circuits.  Each PW-specific companion document MUST specify
      precisely how this AVP is translated for each circuit type.

      If this AVP is received with a Not Active notification for a given
      L2TP session, all data traffic for that session MUST cease (or not
      begin) in the direction of the sender of the Circuit Status AVP
      until the circuit is advertised as Active.

      The Circuit Status MUST be advertised by this AVP in ICRQ, ICRP,
      OCRQ, and OCRP messages.  Often, the circuit type will be marked
      Active when initiated, but subsequently MAY be advertised as
      Inactive.  This indicates that an L2TP session is to be created,
      but that the interface or circuit is still not ready to pass
      traffic.  The ICCN, OCCN, and SLI control messages all MAY contain
      this AVP to update the status of the circuit after establishment
      of the L2TP session is requested.

      If additional circuit status information is needed for a given PW
      type, any new PW-specific AVPs MUST be defined in a separate
      document.  This AVP is only for general circuit status information
      generally applicable to all circuit/interface types.

      This AVP MAY be hidden (the H bit MAY be 0 or 1).  The M bit for
      this AVP SHOULD be set to 1, but MAY vary (see Section 5.2).  The
      Length (before hiding) of this AVP is 8.

   Circuit Errors (WEN)

      The Circuit Errors AVP, Attribute Type 34, conveys circuit error
      information to the peer.














Lau, et al.                 Standards Track                    [Page 58]

RFC 3931                         L2TPv3                       March 2005


      The Attribute Value field for this AVP has the following format:

       0                   1                   2                   3
       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
                                     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-
                                     |             Reserved           |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                        Hardware Overruns                      |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                         Buffer Overruns                       |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                         Timeout Errors                        |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |                        Alignment Errors                       |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

      The following fields are defined:

      Reserved: 2 octets of Reserved data is present (providing longword
         alignment within the AVP of the following values).  Reserved
         data MUST be zero on sending and ignored upon receipt.
      Hardware Overruns: Number of receive buffer overruns since call
         was established.
      Buffer Overruns: Number of buffer overruns detected since call was
         established.
      Timeout Errors: Number of timeouts since call was established.
      Alignment Errors: Number of alignment errors since call was
         established.

      This AVP MAY be hidden (the H bit MAY be 0 or 1).  The M bit for
      this AVP SHOULD be set to 0, but MAY vary (see Section 5.2).  The
      Length (before hiding) of this AVP is 32.

6.  Control Connection Protocol Specification



   The following control messages are used to establish, maintain, and
   tear down L2TP control connections.  All data packets are sent in
   network order (high-order octets first).  Any "reserved" or "empty"
   fields MUST be sent as 0 values to allow for protocol extensibility.

   The exchanges in which these messages are involved are outlined in
   Section 3.3.









Lau, et al.                 Standards Track                    [Page 59]

RFC 3931                         L2TPv3                       March 2005


6.1.  Start-Control-Connection-Request (SCCRQ)



   Start-Control-Connection-Request (SCCRQ) is a control message used to
   initiate a control connection between two LCCEs.  It is sent by
   either the LAC or the LNS to begin the control connection
   establishment process.

   The following AVPs MUST be present in the SCCRQ:

      Message Type
      Host Name
      Router ID
      Assigned Control Connection ID
      Pseudowire Capabilities List

   The following AVPs MAY be present in the SCCRQ:

      Random Vector
      Control Message Authentication Nonce
      Message Digest
      Control Connection Tie Breaker
      Vendor Name
      Receive Window Size
      Preferred Language

6.2.  Start-Control-Connection-Reply (SCCRP)



   Start-Control-Connection-Reply (SCCRP) is the control message sent in
   reply to a received SCCRQ message.  The SCCRP is used to indicate
   that the SCCRQ was accepted and that establishment of the control
   connection should continue.

   The following AVPs MUST be present in the SCCRP:

      Message Type
      Host Name
      Router ID
      Assigned Control Connection ID
      Pseudowire Capabilities List

   The following AVPs MAY be present in the SCCRP:

      Random Vector
      Control Message Authentication Nonce
      Message Digest
      Vendor Name
      Receive Window Size
      Preferred Language



Lau, et al.                 Standards Track                    [Page 60]

RFC 3931                         L2TPv3                       March 2005


6.3.  Start-Control-Connection-Connected (SCCCN)



   Start-Control-Connection-Connected (SCCCN) is the control message
   sent in reply to an SCCRP.  The SCCCN completes the control
   connection establishment process.

   The following AVP MUST be present in the SCCCN:

      Message Type

   The following AVP MAY be present in the SCCCN:

      Random Vector
      Message Digest

6.4.  Stop-Control-Connection-Notification (StopCCN)



   Stop-Control-Connection-Notification (StopCCN) is the control message
   sent by either LCCE to inform its peer that the control connection is
   being shut down and that the control connection should be closed.  In
   addition, all active sessions are implicitly cleared (without sending
   any explicit session control messages).  The reason for issuing this
   request is indicated in the Result Code AVP.  There is no explicit
   reply to the message, only the implicit ACK that is received by the
   reliable control message delivery layer.

   The following AVPs MUST be present in the StopCCN:

      Message Type
      Result Code

   The following AVPs MAY be present in the StopCCN:

      Random Vector
      Message Digest
      Assigned Control Connection ID

   Note that the Assigned Control Connection ID MUST be present if the
   StopCCN is sent after an SCCRQ or SCCRP message has been sent.

6.5.  Hello (HELLO)



   The Hello (HELLO) message is an L2TP control message sent by either
   peer of a control connection.  This control message is used as a
   "keepalive" for the control connection.  See Section 4.2 for a
   description of the keepalive mechanism.





Lau, et al.                 Standards Track                    [Page 61]

RFC 3931                         L2TPv3                       March 2005


   HELLO messages are global to the control connection.  The Session ID
   in a HELLO message MUST be 0.

   The following AVP MUST be present in the HELLO:

      Message Type

   The following AVP MAY be present in the HELLO:

      Random Vector
      Message Digest

6.6.  Incoming-Call-Request (ICRQ)



   Incoming-Call-Request (ICRQ) is the control message sent by an LCCE
   to a peer when an incoming call is detected (although the ICRQ may
   also be sent as a result of a local event).  It is the first in a
   three-message exchange used for establishing a session via an L2TP
   control connection.

   The ICRQ is used to indicate that a session is to be established
   between an LCCE and a peer.  The sender of an ICRQ provides the peer
   with parameter information for the session.  However, the sender
   makes no demands about how the session is terminated at the peer
   (i.e., whether the L2 traffic is processed locally, forwarded, etc.).

   The following AVPs MUST be present in the ICRQ:

      Message Type
      Local Session ID
      Remote Session ID
      Serial Number
      Pseudowire Type
      Remote End ID
      Circuit Status

   The following AVPs MAY be present in the ICRQ:

      Random Vector
      Message Digest
      Assigned Cookie
      Session Tie Breaker
      L2-Specific Sublayer
      Data Sequencing
      Tx Connect Speed
      Rx Connect Speed
      Physical Channel ID




Lau, et al.                 Standards Track                    [Page 62]

RFC 3931                         L2TPv3                       March 2005


6.7.  Incoming-Call-Reply (ICRP)



   Incoming-Call-Reply (ICRP) is the control message sent by an LCCE in
   response to a received ICRQ.  It is the second in the three-message
   exchange used for establishing sessions within an L2TP control
   connection.

   The ICRP is used to indicate that the ICRQ was successful and that
   the peer should establish (i.e., answer) the incoming call if it has
   not already done so.  It also allows the sender to indicate specific
   parameters about the L2TP session.

   The following AVPs MUST be present in the ICRP:

      Message Type
      Local Session ID
      Remote Session ID
      Circuit Status

   The following AVPs MAY be present in the ICRP:

      Random Vector
      Message Digest
      Assigned Cookie
      L2-Specific Sublayer
      Data Sequencing
      Tx Connect Speed
      Rx Connect Speed
      Physical Channel ID

6.8.  Incoming-Call-Connected (ICCN)



   Incoming-Call-Connected (ICCN) is the control message sent by the
   LCCE that originally sent an ICRQ upon receiving an ICRP from its
   peer.  It is the final message in the three-message exchange used for
   establishing L2TP sessions.

   The ICCN is used to indicate that the ICRP was accepted, that the
   call has been established, and that the L2TP session should move to
   the established state.  It also allows the sender to indicate
   specific parameters about the established call (parameters that may
   not have been available at the time the ICRQ was issued).

   The following AVPs MUST be present in the ICCN:

      Message Type
      Local Session ID
      Remote Session ID



Lau, et al.                 Standards Track                    [Page 63]

RFC 3931                         L2TPv3                       March 2005


   The following AVPs MAY be present in the ICCN:

      Random Vector
      Message Digest
      L2-Specific Sublayer
      Data Sequencing
      Tx Connect Speed
      Rx Connect Speed
      Circuit Status

6.9.  Outgoing-Call-Request (OCRQ)



   Outgoing-Call-Request (OCRQ) is the control message sent by an LCCE
   to an LAC to indicate that an outbound call at the LAC is to be
   established based on specific destination information sent in this
   message.  It is the first in a three-message exchange used for
   establishing a session and placing a call on behalf of the initiating
   LCCE.

   Note that a call may be any L2 connection requiring well-known
   destination information to be sent from an LCCE to an LAC.  This call
   could be a dialup connection to the PSTN, an SVC connection, the IP
   address of another LCCE, or any other destination dictated by the
   sender of this message.

   The following AVPs MUST be present in the OCRQ:

      Message Type
      Local Session ID
      Remote Session ID
      Serial Number
      Pseudowire Type
      Remote End ID
      Circuit Status

   The following AVPs MAY be present in the OCRQ:

      Random Vector
      Message Digest
      Assigned Cookie
      Tx Connect Speed
      Rx Connect Speed
      Session Tie Breaker
      L2-Specific Sublayer
      Data Sequencing






Lau, et al.                 Standards Track                    [Page 64]

RFC 3931                         L2TPv3                       March 2005


6.10.  Outgoing-Call-Reply (OCRP)



   Outgoing-Call-Reply (OCRP) is the control message sent by an LAC to
   an LCCE in response to a received OCRQ.  It is the second in a
   three-message exchange used for establishing a session within an L2TP
   control connection.

   OCRP is used to indicate that the LAC has been able to attempt the
   outbound call.  The message returns any relevant parameters regarding
   the call attempt.  Data MUST NOT be forwarded until the OCCN is
   received, which indicates that the call has been placed.

   The following AVPs MUST be present in the OCRP:

      Message Type
      Local Session ID
      Remote Session ID
      Circuit Status

   The following AVPs MAY be present in the OCRP:

      Random Vector
      Message Digest
      Assigned Cookie
      L2-Specific Sublayer
      Tx Connect Speed
      Rx Connect Speed
      Data Sequencing
      Physical Channel ID

6.11.  Outgoing-Call-Connected (OCCN)



   Outgoing-Call-Connected (OCCN) is the control message sent by an LAC
   to another LCCE after the OCRP and after the outgoing call has been
   completed.  It is the final message in a three-message exchange used
   for establishing a session.

   OCCN is used to indicate that the result of a requested outgoing call
   was successful.  It also provides information to the LCCE who
   requested the call about the particular parameters obtained after the
   call was established.

   The following AVPs MUST be present in the OCCN:

      Message Type
      Local Session ID
      Remote Session ID




Lau, et al.                 Standards Track                    [Page 65]

RFC 3931                         L2TPv3                       March 2005


   The following AVPs MAY be present in the OCCN:

      Random Vector
      Message Digest
      L2-Specific Sublayer
      Tx Connect Speed
      Rx Connect Speed
      Data Sequencing
      Circuit Status

6.12.  Call-Disconnect-Notify (CDN)



   The Call-Disconnect-Notify (CDN) is a control message sent by an LCCE
   to request disconnection of a specific session.  Its purpose is to
   inform the peer of the disconnection and the reason for the
   disconnection.  The peer MUST clean up any resources, and does not
   send back any indication of success or failure for such cleanup.

   The following AVPs MUST be present in the CDN:

      Message Type
      Result Code
      Local Session ID
      Remote Session ID

   The following AVP MAY be present in the CDN:

      Random Vector
      Message Digest

6.13.  WAN-Error-Notify (WEN)



   The WAN-Error-Notify (WEN) is a control message sent from an LAC to
   an LNS to indicate WAN error conditions.  The counters in this
   message are cumulative.  This message should only be sent when an
   error occurs, and not more than once every 60 seconds.  The counters
   are reset when a new call is established.

   The following AVPs MUST be present in the WEN:

      Message Type
      Local Session ID
      Remote Session ID
      Circuit Errors







Lau, et al.                 Standards Track                    [Page 66]

RFC 3931                         L2TPv3                       March 2005


   The following AVP MAY be present in the WEN:

      Random Vector
      Message Digest

6.14.  Set-Link-Info (SLI)



   The Set-Link-Info control message is sent by an LCCE to convey link
   or circuit status change information regarding the circuit associated
   with this L2TP session.  For example, if PPP renegotiates LCP at an
   LNS or between an LAC and a Remote System, or if a forwarded Frame
   Relay VC transitions to Active or Inactive at an LAC, an SLI message
   SHOULD be sent to indicate this event.  Precise details of when the
   SLI is sent, what PW type-specific AVPs must be present, and how
   those AVPs should be interpreted by the receiving peer are outside
   the scope of this document.  These details should be described in the
   associated pseudowire-specific documents that require use of this
   message.

   The following AVPs MUST be present in the SLI:

      Message Type
      Local Session ID
      Remote Session ID

   The following AVPs MAY be present in the SLI:

      Random Vector
      Message Digest
      Circuit Status

6.15.  Explicit-Acknowledgement (ACK)



   The Explicit Acknowledgement (ACK) message is used only to
   acknowledge receipt of a message or messages on the control
   connection (e.g., for purposes of updating Ns and Nr values).
   Receipt of this message does not trigger an event for the L2TP
   protocol state machine.

   A message received without any AVPs (including the Message Type AVP),
   is referred to as a Zero Length Body (ZLB) message, and serves the
   same function as the Explicit Acknowledgement.  ZLB messages are only
   permitted when Control Message Authentication defined in Section 4.3
   is not enabled.







Lau, et al.                 Standards Track                    [Page 67]

RFC 3931                         L2TPv3                       March 2005


   The following AVPs MAY be present in the ACK message:

      Message Type
      Message Digest

7.  Control Connection State Machines



   The state tables defined in this section govern the exchange of
   control messages defined in Section 6.  Tables are defined for
   incoming call placement and outgoing call placement, as well as for
   initiation of the control connection itself.  The state tables do not
   encode timeout and retransmission behavior, as this is handled in the
   underlying reliable control message delivery mechanism (see Section
   4.2).

7.1.  Malformed AVPs and Control Messages



   Receipt of an invalid or unrecoverable malformed control message
   SHOULD be logged appropriately and the control connection cleared to
   ensure recovery to a known state.  The control connection may then be
   restarted by the initiator.

   An invalid control message is defined as (1) a message that contains
   a Message Type marked as mandatory (see Section 5.4.1) but that is
   unknown to the implementation, or (2) a control message that is
   received in the wrong state.

   Examples of malformed control messages include (1) a message that has
   an invalid value in its header, (2) a message that contains an AVP
   that is formatted incorrectly or whose value is out of range, and (3)
   a message that is missing a required AVP.  A control message with a
   malformed header MUST be discarded.

   When possible, a malformed AVP should be treated as an unrecognized
   AVP (see Section 5.2).  Thus, an attempt to inspect the M bit SHOULD
   be made to determine the importance of the malformed AVP, and thus,
   the severity of the malformation to the entire control message.  If
   the M bit can be reasonably inspected within the malformed AVP and is
   determined to be set, then as with an unrecognized AVP, the
   associated session or control connection MUST be shut down.  If the M
   bit is inspected and is found to be 0, the AVP MUST be ignored
   (assuming recovery from the AVP malformation is indeed possible).

   This policy must not be considered as a license to send malformed
   AVPs, but rather, as a guide towards how to handle an improperly
   formatted message if one is received.  It is impossible to list all
   potential malformations of a given message and give advice for each.
   One example of a malformed AVP situation that should be recoverable



Lau, et al.                 Standards Track                    [Page 68]

RFC 3931                         L2TPv3                       March 2005


   is if the Rx Connect Speed AVP is received with a length of 10 rather
   than 14, implying that the connect speed bits-per-second is being
   formatted in 4 octets rather than 8.  If the AVP does not have its M
   bit set (as would typically be the case), this condition is not
   considered catastrophic.  As such, the control message should be
   accepted as though the AVP were not present (though a local error
   message may be logged).


   In several cases in the following tables, a protocol message is sent,
   and then a "clean up" occurs.  Note that, regardless of the initiator
   of the control connection destruction, the reliable delivery
   mechanism must be allowed to run (see Section 4.2) before destroying
   the control connection.  This permits the control connection
   management messages to be reliably delivered to the peer.

   Appendix B.1 contains an example of lock-step control connection
   establishment.

7.2.  Control Connection States



   The L2TP control connection protocol is not distinguishable between
   the two LCCEs but is distinguishable between the originator and
   receiver.  The originating peer is the one that first initiates
   establishment of the control connection.  (In a tie breaker
   situation, this is the winner of the tie.)  Since either the LAC or
   the LNS can be the originator, a collision can occur.  See the
   Control Connection Tie Breaker AVP in Section 5.4.3 for a description
   of this and its resolution.

   State           Event              Action              New State
   -----           -----              ------              ---------
   idle            Local open         Send SCCRQ          wait-ctl-reply
                   request

   idle            Receive SCCRQ,     Send SCCRP          wait-ctl-conn
                   acceptable

   idle            Receive SCCRQ,     Send StopCCN,       idle
                   not acceptable     clean up

   idle            Receive SCCRP      Send StopCCN,       idle
                                      clean up

   idle            Receive SCCCN      Send StopCCN,       idle
                                      clean up





Lau, et al.                 Standards Track                    [Page 69]

RFC 3931                         L2TPv3                       March 2005


   wait-ctl-reply  Receive SCCRP,     Send SCCCN,         established
                   acceptable         send control-conn
                                      open event to
                                      waiting sessions

   wait-ctl-reply  Receive SCCRP,     Send StopCCN,       idle
                   not acceptable     clean up

   wait-ctl-reply  Receive SCCRQ,     Send SCCRP,         wait-ctl-conn
                   lose tie breaker,  Clean up losing
                   SCCRQ acceptable   connection

   wait-ctl-reply  Receive SCCRQ,     Send StopCCN,       idle
                   lose tie breaker,  Clean up losing
                   SCCRQ unacceptable connection

   wait-ctl-reply  Receive SCCRQ,     Send StopCCN for    wait-ctl-reply
                   win tie breaker    losing connection

   wait-ctl-reply  Receive SCCCN      Send StopCCN,       idle
                                      clean up

   wait-ctl-conn   Receive SCCCN,     Send control-conn   established
                   acceptable         open event to
                                      waiting sessions

   wait-ctl-conn   Receive SCCCN,     Send StopCCN,       idle
                   not acceptable     clean up

   wait-ctl-conn   Receive SCCRQ,     Send StopCCN,       idle
                   SCCRP              clean up

   established     Local open         Send control-conn   established
                   request            open event to
                   (new call)         waiting sessions

   established     Administrative     Send StopCCN,       idle
                   control-conn       clean up
                   close event

   established     Receive SCCRQ,     Send StopCCN,       idle
                   SCCRP, SCCCN       clean up

   idle,           Receive StopCCN    Clean up            idle
   wait-ctl-reply,
   wait-ctl-conn,
   established




Lau, et al.                 Standards Track                    [Page 70]

RFC 3931                         L2TPv3                       March 2005


   The states associated with an LCCE for control connection
   establishment are as follows:

   idle
      Both initiator and recipient start from this state.  An initiator
      transmits an SCCRQ, while a recipient remains in the idle state
      until receiving an SCCRQ.

   wait-ctl-reply
      The originator checks to see if another connection has been
      requested from the same peer, and if so, handles the collision
      situation described in Section 5.4.3.

   wait-ctl-conn
      Awaiting an SCCCN.  If the SCCCN is valid, the control connection
      is established; otherwise, it is torn down (sending a StopCCN with
      the proper result and/or error code).

   established
      An established connection may be terminated by either a local
      condition or the receipt of a StopCCN.  In the event of a local
      termination, the originator MUST send a StopCCN and clean up the
      control connection.  If the originator receives a StopCCN, it MUST
      also clean up the control connection.

7.3.  Incoming Calls



   An ICRQ is generated by an LCCE, typically in response to an incoming
   call or a local event.  Once the LCCE sends the ICRQ, it waits for a
   response from the peer.  However, it may choose to postpone
   establishment of the call (e.g., answering the call, bringing up the
   circuit) until the peer has indicated with an ICRP that it will
   accept the call.  The peer may choose not to accept the call if, for
   instance, there are insufficient resources to handle an additional
   session.

   If the peer chooses to accept the call, it responds with an ICRP.
   When the local LCCE receives the ICRP, it attempts to establish the
   call.  A final call connected message, the ICCN, is sent from the
   local LCCE to the peer to indicate that the call states for both
   LCCEs should enter the established state.  If the call is terminated
   before the peer can accept it, a CDN is sent by the local LCCE to
   indicate this condition.

   When a call transitions to a "disconnected" or "down" state, the call
   is cleared normally, and the local LCCE sends a CDN.  Similarly, if
   the peer wishes to clear a call, it sends a CDN and cleans up its
   session.



Lau, et al.                 Standards Track                    [Page 71]

RFC 3931                         L2TPv3                       March 2005


7.3.1.  ICRQ Sender States



   State           Event              Action           New State
   -----           -----              ------           ---------

   idle            Call signal or     Initiate local   wait-control-conn
                   ready to receive   control-conn
                   incoming conn      open

   idle            Receive ICCN,      Clean up         idle
                   ICRP, CDN

   wait-control-   Bearer line drop   Clean up         idle
   conn            or local close
                   request

   wait-control-   control-conn-open  Send ICRQ        wait-reply
   conn

   wait-reply      Receive ICRP,      Send ICCN        established
                   acceptable

   wait-reply      Receive ICRP,      Send CDN,        idle
                   Not acceptable     clean up

   wait-reply      Receive ICRQ,      Process as       idle
                   lose tie breaker   ICRQ Recipient
                                      (Section 7.3.2)

   wait-reply      Receive ICRQ,      Send CDN         wait-reply
                   win tie breaker    for losing
                                      session

   wait-reply      Receive CDN,       Clean up         idle
                   ICCN

   wait-reply      Local close        Send CDN,        idle
                   request            clean up

   established     Receive CDN        Clean up         idle

   established     Receive ICRQ,      Send CDN,        idle
                   ICRP, ICCN         clean up

   established     Local close        Send CDN,        idle
                   request            clean up





Lau, et al.                 Standards Track                    [Page 72]

RFC 3931                         L2TPv3                       March 2005


   The states associated with the ICRQ sender are as follows:

   idle
      The LCCE detects an incoming call on one of its interfaces (e.g.,
      an analog PSTN line rings, or an ATM PVC is provisioned), or a
      local event occurs.  The LCCE initiates its control connection
      establishment state machine and moves to a state waiting for
      confirmation of the existence of a control connection.

   wait-control-conn
      In this state, the session is waiting for either the control
      connection to be opened or for verification that the control
      connection is already open.  Once an indication that the control
      connection has been opened is received, session control messages
      may be exchanged.  The first of these messages is the ICRQ.

   wait-reply
      The ICRQ sender receives either (1) a CDN indicating the peer is
      not willing to accept the call (general error or do not accept)
      and moves back into the idle state, or (2) an ICRP indicating the
      call is accepted.  In the latter case, the LCCE sends an ICCN and
      enters the established state.

   established
      Data is exchanged over the session.  The call may be cleared by
      any of the following:
         + An event on the connected interface: The LCCE sends a CDN.
         + Receipt of a CDN: The LCCE cleans up, disconnecting the call.
         + A local reason: The LCCE sends a CDN.

7.3.2.  ICRQ Recipient States



   State           Event              Action            New State
   -----           -----              ------            ---------
   idle            Receive ICRQ,      Send ICRP         wait-connect
                   acceptable

   idle            Receive ICRQ,      Send CDN,         idle
                   not acceptable     clean up

   idle            Receive ICRP       Send CDN          idle
                                      clean up

   idle            Receive ICCN       Clean up          idle

   wait-connect    Receive ICCN,      Prepare for       established
                   acceptable         data




Lau, et al.                 Standards Track                    [Page 73]

RFC 3931                         L2TPv3                       March 2005


   wait-connect    Receive ICCN,      Send CDN,         idle
                   not acceptable     clean up

   wait-connect    Receive ICRQ,      Send CDN,         idle
                   ICRP               clean up

   idle,           Receive CDN        Clean up          idle
   wait-connect,
   established

   wait-connect    Local close        Send CDN,         idle
   established     request            clean up

   established     Receive ICRQ,      Send CDN,         idle
                   ICRP, ICCN         clean up

   The states associated with the ICRQ recipient are as follows:

   idle
      An ICRQ is received.  If the request is not acceptable, a CDN is
      sent back to the peer LCCE, and the local LCCE remains in the idle
      state.  If the ICRQ is acceptable, an ICRP is sent.  The session
      moves to the wait-connect state.

   wait-connect
      The local LCCE is waiting for an ICCN from the peer.  Upon receipt
      of the ICCN, the local LCCE moves to established state.

   established
      The session is terminated either by sending a CDN or by receiving
      a CDN from the peer.  Clean up follows on both sides regardless of
      the initiator.

7.4.  Outgoing Calls



   Outgoing calls instruct an LAC to place a call.  There are three
   messages for outgoing calls: OCRQ, OCRP, and OCCN.  An LCCE first
   sends an OCRQ to an LAC to request an outgoing call.  The LAC MUST
   respond to the OCRQ with an OCRP once it determines that the proper
   facilities exist to place the call and that the call is
   administratively authorized.  Once the outbound call is connected,
   the LAC sends an OCCN to the peer indicating the final result of the
   call attempt.








Lau, et al.                 Standards Track                    [Page 74]

RFC 3931                         L2TPv3                       March 2005


7.4.1.  OCRQ Sender States



   State          Event              Action            New State
   -----          -----              ------            ---------
   idle           Local open         Initiate local    wait-control-conn
                  request            control-conn-open

   idle           Receive OCCN,      Clean up          idle
                  OCRP

   wait-control-  control-conn-open  Send OCRQ         wait-reply
   conn

   wait-reply     Receive OCRP,      none              wait-connect
                  acceptable

   wait-reply     Receive OCRP,      Send CDN,         idle
                  not acceptable     clean up

   wait-reply     Receive OCCN       Send CDN,         idle
                                     clean up

   wait-reply     Receive OCRQ,      Process as        idle
                  lose tie breaker   OCRQ Recipient
                                     (Section 7.4.2)

   wait-reply     Receive OCRQ,      Send CDN          wait-reply
                  win tie breaker    for losing
                                     session

   wait-connect   Receive OCCN       none              established

   wait-connect   Receive OCRQ,      Send CDN,         idle
                  OCRP               clean up

   idle,          Receive CDN        Clean up          idle
   wait-reply,
   wait-connect,
   established

   established    Receive OCRQ,      Send CDN,         idle
                  OCRP, OCCN         clean up

   wait-reply,    Local close        Send CDN,         idle
   wait-connect,  request            clean up
   established





Lau, et al.                 Standards Track                    [Page 75]

RFC 3931                         L2TPv3                       March 2005


   wait-control-  Local close        Clean up          idle
   conn           request

   The states associated with the OCRQ sender are as follows:

   idle, wait-control-conn
      When an outgoing call request is initiated, a control connection
      is created as described above, if not already present.  Once the
      control connection is established, an OCRQ is sent to the LAC, and
      the session moves into the wait-reply state.

   wait-reply
      If a CDN is received, the session is cleaned up and returns to
      idle state.  If an OCRP is received, the call is in progress, and
      the session moves to the wait-connect state.

   wait-connect
      If a CDN is received, the session is cleaned up and returns to
      idle state.  If an OCCN is received, the call has succeeded, and
      the session may now exchange data.

   established
      If a CDN is received, the session is cleaned up and returns to
      idle state.  Alternatively, if the LCCE chooses to terminate the
      session, it sends a CDN to the LAC, cleans up the session, and
      moves the session to idle state.

7.4.2.  OCRQ Recipient (LAC) States



   State           Event              Action            New State
   -----           -----              ------            ---------
   idle            Receive OCRQ,      Send OCRP,        wait-cs-answer
                   acceptable         Place call

   idle            Receive OCRQ,      Send CDN,         idle
                   not acceptable     clean up

   idle            Receive OCRP       Send CDN,         idle
                                      clean up

   idle            Receive OCCN,      Clean up          idle
                   CDN

   wait-cs-answer  Call placement     Send OCCN         established
                   successful

   wait-cs-answer  Call placement     Send CDN,         idle
                   failed             clean up



Lau, et al.                 Standards Track                    [Page 76]

RFC 3931                         L2TPv3                       March 2005


   wait-cs-answer  Receive OCRQ,      Send CDN,         idle
                   OCRP, OCCN         clean up

   established     Receive OCRQ,      Send CDN,         idle
                   OCRP, OCCN         clean up

   wait-cs-answer, Receive CDN        Clean up          idle
   established

   wait-cs-answer, Local close        Send CDN,         idle
   established     request            clean up

   The states associated with the LAC for outgoing calls are as follows:

   idle
      If the OCRQ is received in error, respond with a CDN.  Otherwise,
      place the call, send an OCRP, and move to the wait-cs-answer
      state.

   wait-cs-answer
      If the call is not completed or a timer expires while waiting for
      the call to complete, send a CDN with the appropriate error
      condition set, and go to idle state.  If a circuit-switched
      connection is established, send an OCCN indicating success, and go
      to established state.

   established
      If the LAC receives a CDN from the peer, the call MUST be released
      via appropriate mechanisms, and the session cleaned up.  If the
      call is disconnected because the circuit transitions to a
      "disconnected" or "down" state, the LAC MUST send a CDN to the
      peer and return to idle state.

7.5.  Termination of a Control Connection



   The termination of a control connection consists of either peer
   issuing a StopCCN.  The sender of this message SHOULD wait a full
   control message retransmission cycle (e.g., 1 + 2 + 4 + 8 ...
   seconds) for the acknowledgment of this message before releasing the
   control information associated with the control connection.  The
   recipient of this message should send an acknowledgment of the
   message to the peer, then release the associated control information.

   When to release a control connection is an implementation issue and
   is not specified in this document.  A particular implementation may
   use whatever policy is appropriate for determining when to release a
   control connection.  Some implementations may leave a control
   connection open for a period of time or perhaps indefinitely after



Lau, et al.                 Standards Track                    [Page 77]

RFC 3931                         L2TPv3                       March 2005


   the last session for that control connection is cleared.  Others may
   choose to disconnect the control connection immediately after the
   last call on the control connection disconnects.

8.  Security Considerations



   This section addresses some of the security issues that L2TP
   encounters in its operation.

8.1.  Control Connection Endpoint and Message Security



   If a shared secret (password) exists between two LCCEs, it may be
   used to perform a mutual authentication between the two LCCEs, and
   construct an authentication and integrity check of arriving L2TP
   control messages.  The mechanism provided by L2TPv3 is described in
   Section 4.3 and in the definition of the Message Digest and Control
   Message Authentication Nonce AVPs in Section 5.4.1.

   This control message security mechanism provides for (1) mutual
   endpoint authentication, and (2) individual control message integrity
   and authenticity checking.  Mutual endpoint authentication ensures
   that an L2TPv3 control connection is only established between two
   endpoints that are configured with the proper password.  The
   individual control message and integrity check guards against
   accidental or intentional packet corruption (i.e., those caused by a
   control message spoofing or man-in-the-middle attack).

   The shared secret that is used for all control connection, control
   message, and AVP security features defined in this document never
   needs to be sent in the clear between L2TP tunnel endpoints.

8.2.  Data Packet Spoofing



   Packet spoofing for any type of Virtual Private Network (VPN)
   protocol is of particular concern as insertion of carefully
   constructed rogue packets into the VPN transit network could result
   in a violation of VPN traffic separation, leaking data into a
   customer VPN.  This is complicated by the fact that it may be
   particularly difficult for the operator of the VPN to even be aware
   that it has become a point of transit into or between customer VPNs.

   L2TPv3 provides traffic separation for its VPNs via a 32-bit Session
   ID in the L2TPv3 data header.  When present, the L2TPv3 Cookie
   (described in Section 4.1), provides an additional check to ensure
   that an arriving packet is intended for the identified session.
   Thus, use of a Cookie with the Session ID provides an extra guarantee
   that the Session ID lookup was performed properly and that the
   Session ID itself was not corrupted in transit.



Lau, et al.                 Standards Track                    [Page 78]

RFC 3931                         L2TPv3                       March 2005


   In the presence of a blind packet spoofing attack, the Cookie may
   also provide security against inadvertent leaking of frames into a
   customer VPN.  To illustrate the type of security that it is provided
   in this case, consider comparing the validation of a 64-bit Cookie in
   the L2TPv3 header to the admission of packets that match a given
   source and destination IP address pair.  Both the source and
   destination IP address pair validation and Cookie validation consist
   of a fast check on cleartext header information on all arriving
   packets.  However, since L2TPv3 uses its own value, it removes the
   requirement for one to maintain a list of (potentially several)
   permitted or denied IP addresses, and moreover, to guard knowledge of
   the permitted IP addresses from hackers who may obtain and spoof
   them.  Further, it is far easier to change a compromised L2TPv3
   Cookie than a compromised IP address," and a cryptographically random
   [RFC1750] value is far less likely to be discovered by brute-force
   attacks compared to an IP address.

   For protection against brute-force, blind, insertion attacks, a 64-
   bit Cookie MUST be used with all sessions.  A 32-bit Cookie is
   vulnerable to brute-force guessing at high packet rates, and as such,
   should not be considered an effective barrier to blind insertion
   attacks (though it is still useful as an additional verification of a
   successful Session ID lookup).  The Cookie provides no protection
   against a sophisticated man-in-the-middle attacker who can sniff and
   correlate captured data between nodes for use in a coordinated
   attack.

   The Assigned Cookie AVP is used to signal the value and size of the
   Cookie that must be present in all data packets for a given session.
   Each Assigned Cookie MUST be selected in a cryptographically random
   manner [RFC1750] such that a series of Assigned Cookies does not
   provide any indication of what a future Cookie will be.

   The L2TPv3 Cookie must not be regarded as a substitute for security
   such as that provided by IPsec when operating over an open or
   untrusted network where packets may be sniffed, decoded, and
   correlated for use in a coordinated attack.  See Section 4.1.3 for
   more information on running L2TP over IPsec.

9.  Internationalization Considerations



   The Host Name and Vendor Name AVPs are not internationalized.  The
   Vendor Name AVP, although intended to be human-readable, would seem
   to fit in the category of "globally visible names" [RFC2277] and so
   is represented in US-ASCII.

   If (1) an LCCE does not signify a language preference by the
   inclusion of a Preferred Language AVP (see Section 5.4.3) in the



Lau, et al.                 Standards Track                    [Page 79]

RFC 3931                         L2TPv3                       March 2005


   SCCRQ or SCCRP, (2) the Preferred Language AVP is unrecognized, or
   (3) the requested language is not supported by the peer LCCE, the
   default language [RFC2277] MUST be used for all internationalized
   strings sent by the peer.

10.  IANA Considerations



   This document defines a number of "magic" numbers to be maintained by
   the IANA.  This section explains the criteria used by the IANA to
   assign additional numbers in each of these lists.  The following
   subsections describe the assignment policy for the namespaces defined
   elsewhere in this document.

   Sections 10.1 through 10.3 are requests for new values already
   managed by IANA according to [RFC3438].

   The remaining sections are for new registries that have been added to
   the existing L2TP registry and are maintained by IANA accordingly.

10.1.  Control Message Attribute Value Pairs (AVPs)



   This number space is managed by IANA as per [RFC3438].

   A summary of the new AVPs follows:

   Control Message Attribute Value Pairs

      Attribute
      Type        Description
      ---------   ------------------

         58       Extended Vendor ID AVP
         59       Message Digest
         60       Router ID
         61       Assigned Control Connection ID
         62       Pseudowire Capabilities List
         63       Local Session ID
         64       Remote Session ID
         65       Assigned Cookie
         66       Remote End ID
         68       Pseudowire Type
         69       L2-Specific Sublayer
         70       Data Sequencing
         71       Circuit Status
         72       Preferred Language
         73       Control Message Authentication Nonce
         74       Tx Connect Speed
         75       Rx Connect Speed



Lau, et al.                 Standards Track                    [Page 80]

RFC 3931                         L2TPv3                       March 2005


10.2.  Message Type AVP Values



   This number space is managed by IANA as per [RFC3438].  There is one
   new message type, defined in Section 3.1, that was allocated for this
   specification:

   Message Type AVP (Attribute Type 0) Values
   ------------------------------------------

     Control Connection Management

         20 (ACK)  Explicit Acknowledgement



10.3.  Result Code AVP Values



   This number space is managed by IANA as per [RFC3438].

   New Result Code values for the CDN message are defined in Section
   5.4.  The following is a summary:

   Result Code AVP (Attribute Type 1) Values
   -----------------------------------------

      General Error Codes

         13 - Session not established due to losing
              tie breaker (L2TPv3).
         14 - Session not established due to unsupported
              PW type (L2TPv3).
         15 - Session not established, sequencing required
              without valid L2-Specific Sublayer (L2TPv3).
         16 - Finite state machine error or timeout.



















Lau, et al.                 Standards Track                    [Page 81]

RFC 3931                         L2TPv3                       March 2005


10.4.  AVP Header Bits



   This is a new registry for IANA to maintain.

   Leading Bits of the L2TP AVP Header
   -----------------------------------

   There six bits at the beginning of the L2TP AVP header.  New bits are
   assigned via Standards Action [RFC2434].

   Bit 0 - Mandatory (M bit)
   Bit 1 - Hidden (H bit)
   Bit 2 - Reserved
   Bit 3 - Reserved
   Bit 4 - Reserved
   Bit 5 - Reserved

10.5.  L2TP Control Message Header Bits



   This is a new registry for IANA to maintain.

   Leading Bits of the L2TP Control Message Header
   -----------------------------------------------

   There are 12 bits at the beginning of the L2TP Control Message
   Header.  Reserved bits should only be defined by Standard
   Action [RFC2434].

   Bit  0 - Message Type (T bit)
   Bit  1 - Length Field is Present (L bit)
   Bit  2 - Reserved
   Bit  3 - Reserved
   Bit  4 - Sequence Numbers Present (S bit)
   Bit  5 - Reserved
   Bit  6 - Offset Field is Present [RFC2661]
   Bit  7 - Priority Bit (P bit) [RFC2661]
   Bit  8 - Reserved
   Bit  9 - Reserved
   Bit 10 - Reserved
   Bit 11 - Reserved











Lau, et al.                 Standards Track                    [Page 82]

RFC 3931                         L2TPv3                       March 2005


10.6.  Pseudowire Types



   This is a new registry for IANA to maintain, there are no values
   assigned within this document to maintain.

   L2TPv3 Pseudowire Types
   -----------------------

   The Pseudowire Type (PW Type, see Section 5.4) is a 2-octet value
   used in the Pseudowire Type AVP and Pseudowire Capabilities List AVP
   defined in Section 5.4.3.  0 to 32767 are assignable by Expert Review
   [RFC2434], while 32768 to 65535 are assigned by a First Come First
   Served policy [RFC2434].  There are no specific pseudowire types
   assigned within this document.  Each pseudowire-specific document
   must allocate its own PW types from IANA as necessary.

10.7.  Circuit Status Bits



   This is a new registry for IANA to maintain.

   Circuit Status Bits
   -------------------

   The Circuit Status field is a 16-bit mask, with the two low order
   bits assigned.  Additional bits may be assigned by IETF Consensus
   [RFC2434].

   Bit 14 - New (N bit)
   Bit 15 - Active (A bit)






















Lau, et al.                 Standards Track                    [Page 83]

RFC 3931                         L2TPv3                       March 2005


10.8.  Default L2-Specific Sublayer bits



   This is a new registry for IANA to maintain.

   Default L2-Specific Sublayer Bits
   ---------------------------------

   The Default L2-Specific Sublayer contains 8 bits in the low-order
   portion of the header.  Reserved bits may be assigned by IETF
   Consensus [RFC2434].

   Bit 0 - Reserved
   Bit 1 - Sequence (S bit)
   Bit 2 - Reserved
   Bit 3 - Reserved
   Bit 4 - Reserved
   Bit 5 - Reserved
   Bit 6 - Reserved
   Bit 7 - Reserved

10.9.  L2-Specific Sublayer Type



   This is a new registry for IANA to maintain.

   L2-Specific Sublayer Type
   -------------------------

   The L2-Specific Sublayer Type is a 2-octet unsigned integer.
   Additional values may be assigned by Expert Review [RFC2434].

   0 - No L2-Specific Sublayer
   1 - Default L2-Specific Sublayer present

10.10.  Data Sequencing Level



   This is a new registry for IANA to maintain.

   Data Sequencing Level
   ---------------------

   The Data Sequencing Level is a 2-octet unsigned integer
   Additional values may be assigned by Expert Review [RFC2434].

   0 - No incoming data packets require sequencing.
   1 - Only non-IP data packets require sequencing.
   2 - All incoming data packets require sequencing.





Lau, et al.                 Standards Track                    [Page 84]

RFC 3931                         L2TPv3                       March 2005


11.  References



11.1.  Normative References



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

   [RFC2277] Alvestrand, H., "IETF Policy on Character Sets and
             Languages", BCP 18, RFC 2277, January 1998.

   [RFC2434] Narten, T. and H. Alvestrand, "Guidelines for Writing an
             IANA Considerations section in RFCs", BCP 26, RFC 2434,
             October 1998.

   [RFC2473] Conta, A. and S. Deering, "Generic Packet Tunneling in IPv6
             Specification", RFC 2473, December 1998.

   [RFC2661] Townsley, W., Valencia, A., Rubens, A., Pall, G., Zorn, G.,
             and Palter, B., "Layer Two Tunneling Layer Two Tunneling
             Protocol (L2TP)", RFC 2661, August 1999.

   [RFC2865] Rigney, C., Willens, S., Rubens, A., and W. Simpson,
             "Remote Authentication Dial In User Service (RADIUS)", RFC
             2865, June 2000.

   [RFC3066] Alvestrand, H., "Tags for the Identification of Languages",
             BCP 47, RFC 3066, January 2001.

   [RFC3193] Patel, B., Aboba, B., Dixon, W., Zorn, G., and Booth, S.,
             "Securing L2TP using IPsec", RFC 3193, November 2001.

   [RFC3438] Townsley, W., "Layer Two Tunneling Protocol (L2TP) Internet
             Assigned Numbers Authority (IANA) Considerations Update",
             BCP 68, RFC 3438, December 2002.

   [RFC3629] Yergeau, F., "UTF-8, a transformation format of ISO 10646",
             STD 63, RFC 3629, November 2003.

11.2.  Informative References



   [RFC1034] Mockapetris, P., "Domain Names - Concepts and Facilities",
             STD 13, RFC 1034, November 1987.

   [RFC1191] Mogul, J. and S. Deering, "Path MTU Discovery", RFC 1191,
             November 1990.

   [RFC1321] Rivest, R., "The MD5 Message-Digest Algorithm", RFC 1321,
             April 1992.



Lau, et al.                 Standards Track                    [Page 85]

RFC 3931                         L2TPv3                       March 2005


   [RFC1661] Simpson, W., Ed., "The Point-to-Point Protocol (PPP)", STD
             51, RFC 1661, July 1994.

   [RFC1700] Reynolds, J. and Postel, J., "Assigned Numbers", STD 2, RFC
             1700, October 1994.

   [RFC1750] Eastlake, D., Crocker, S., and Schiller, J., "Randomness
             Recommendations for Security", RFC 1750, December 1994.

   [RFC1958] Carpenter, B., Ed., "Architectural Principles of the
             Internet", RFC 1958, June 1996.

   [RFC1981] McCann, J., Deering, S., and Mogul, J., "Path MTU Discovery
             for IP version 6", RFC 1981, August 1996.

   [RFC2072] Berkowitz, H., "Router Renumbering Guide", RFC 2072,
             January 1997.

   [RFC2104] Krawczyk, H., Bellare, M., and Canetti, R., "HMAC:  Keyed-
             Hashing for Message Authentication", RFC 2104, February
             1997.

   [RFC2341] Valencia, A., Littlewood, M., and Kolar, T., "Cisco Layer
             Two Forwarding (Protocol) L2F", RFC 2341, May 1998.

   [RFC2401] Kent, S. and Atkinson, R., "Security Architecture for the
             Internet Protocol", RFC 2401, November 1998.

   [RFC2581] Allman, M., Paxson, V., and Stevens, W., "TCP Congestion
             Control", RFC 2581, April 1999.

   [RFC2637] Hamzeh, K., Pall, G., Verthein, W., Taarud, J., Little, W.,
             and Zorn, G., "Point-to-Point Tunneling Protocol (PPTP)",
             RFC 2637, July 1999.

   [RFC2732] Hinden, R., Carpenter, B., and Masinter, L., "Format for
             Literal IPv6 Addresses in URL's", RFC 2732, December 1999.

   [RFC2809] Aboba, B. and Zorn, G., "Implementation of L2TP Compulsory
             Tunneling via RADIUS", RFC 2809, April 2000.

   [RFC3070] Rawat, V., Tio, R., Nanji, S., and Verma, R., "Layer Two
             Tunneling Protocol (L2TP) over Frame Relay", RFC 3070,
             February 2001.







Lau, et al.                 Standards Track                    [Page 86]

RFC 3931                         L2TPv3                       March 2005


   [RFC3355] Singh, A., Turner, R., Tio, R., and Nanji, S., "Layer Two
             Tunnelling Protocol (L2TP) Over ATM Adaptation Layer 5
             (AAL5)", RFC 3355, August 2002.

   [KPS]     Kaufman, C., Perlman, R., and Speciner, M., "Network
             Security:  Private Communications in a Public World",
             Prentice Hall, March 1995, ISBN 0-13-061466-1.

   [STEVENS] Stevens, W. Richard, "TCP/IP Illustrated, Volume I: The
             Protocols", Addison-Wesley Publishing Company, Inc., March
             1996, ISBN 0-201-63346-9.

12.  Acknowledgments



   Many of the protocol constructs were originally defined in, and the
   text of this document began with, RFC 2661, "L2TPv2".  RFC 2661
   authors are W. Townsley, A. Valencia, A. Rubens, G. Pall, G. Zorn and
   B. Palter.

   The basic concept for L2TP and many of its protocol constructs were
   adopted from L2F [RFC2341] and PPTP [RFC2637].  Authors of these
   versions are A. Valencia, M. Littlewood, T. Kolar, K. Hamzeh, G.
   Pall, W. Verthein, J. Taarud, W. Little, and G. Zorn.

   Danny Mcpherson and Suhail Nanji published the first "L2TP Service
   Type" version, which defined the use of L2TP for tunneling of various
   L2 payload types (initially, Ethernet and Frame Relay).

   The team for splitting RFC 2661 into this base document and the
   companion PPP document consisted of Ignacio Goyret, Jed Lau, Bill
   Palter, Mark Townsley, and Madhvi Verma.  Skip Booth also provided
   very helpful review and comment.

   Some constructs of L2TPv3 were based in part on UTI (Universal
   Transport Interface), which was originally conceived by Peter
   Lothberg and Tony Bates.

   Stewart Bryant and Simon Barber provided valuable input for the
   L2TPv3 over IP header.

   Juha Heinanen provided helpful review in the early stages of this
   effort.

   Jan Vilhuber, Scott Fluhrer, David McGrew, Scott Wainner, Skip Booth
   and Maria Dos Santos contributed to the Control Message
   Authentication Mechanism as well as general discussions of security.





Lau, et al.                 Standards Track                    [Page 87]

RFC 3931                         L2TPv3                       March 2005


   James Carlson, Thomas Narten, Maria Dos Santos, Steven Bellovin, Ted
   Hardie, and Pekka Savola provided very helpful review of the final
   versions of text.

   Russ Housley provided valuable review and comment on security,
   particularly with respect to the Control Message Authentication
   mechanism.

   Pekka Savola contributed to proper alignment with IPv6 and inspired
   much of Section 4.1.4 on fragmentation.

   Aside of his original influence and co-authorship of RFC 2661, Glen
   Zorn helped get all of the language and character references straight
   in this document.

   A number of people provided valuable input and effort for RFC 2661,
   on which this document was based:

   John Bray, Greg Burns, Rich Garrett, Don Grosser, Matt Holdrege,
   Terry Johnson, Dory Leifer, and Rich Shea provided valuable input and
   review at the 43rd IETF in Orlando, FL, which led to improvement of
   the overall readability and clarity of RFC 2661.

   Thomas Narten provided a great deal of critical review and
   formatting.  He wrote the first version of the IANA Considerations
   section.

   Dory Leifer made valuable refinements to the protocol definition of
   L2TP and contributed to the editing of early versions leading to RFC
   2661.

   Steve Cobb and Evan Caves redesigned the state machine tables.
   Barney Wolff provided a great deal of design input on the original
   endpoint authentication mechanism.

















Lau, et al.                 Standards Track                    [Page 88]

RFC 3931                         L2TPv3                       March 2005


Appendix A: Control Slow Start and Congestion Avoidance



   Although each side has indicated the maximum size of its receive
   window, it is recommended that a slow start and congestion avoidance
   method be used to transmit control packets.  The methods described
   here are based upon the TCP congestion avoidance algorithm as
   described in Section 21.6 of TCP/IP Illustrated, Volume I, by W.
   Richard Stevens [STEVENS] (this algorithm is also described in
   [RFC2581]).

   Slow start and congestion avoidance make use of several variables.
   The congestion window (CWND) defines the number of packets a sender
   may send before waiting for an acknowledgment.  The size of CWND
   expands and contracts as described below.  Note, however, that CWND
   is never allowed to exceed the size of the advertised window obtained
   from the Receive Window AVP.  (In the text below, it is assumed any
   increase will be limited by the Receive Window Size.)  The variable
   SSTHRESH determines when the sender switches from slow start to
   congestion avoidance.  Slow start is used while CWND is less than
   SSHTRESH.

   A sender starts out in the slow start phase.  CWND is initialized to
   one packet, and SSHTRESH is initialized to the advertised window
   (obtained from the Receive Window AVP).  The sender then transmits
   one packet and waits for its acknowledgment (either explicit or
   piggybacked).  When the acknowledgment is received, the congestion
   window is incremented from one to two.  During slow start, CWND is
   increased by one packet each time an ACK (explicit ACK message or
   piggybacked) is received.  Increasing CWND by one on each ACK has the
   effect of doubling CWND with each round trip, resulting in an
   exponential increase.  When the value of CWND reaches SSHTRESH, the
   slow start phase ends and the congestion avoidance phase begins.

   During congestion avoidance, CWND expands more slowly.  Specifically,
   it increases by 1/CWND for every new ACK received.  That is, CWND is
   increased by one packet after CWND new ACKs have been received.
   Window expansion during the congestion avoidance phase is effectively
   linear, with CWND increasing by one packet each round trip.

   When congestion occurs (indicated by the triggering of a
   retransmission) one-half of the CWND is saved in SSTHRESH, and CWND
   is set to one.  The sender then reenters the slow start phase.









Lau, et al.                 Standards Track                    [Page 89]

RFC 3931                         L2TPv3                       March 2005


Appendix B: Control Message Examples



B.1: Lock-Step Control Connection Establishment



   In this example, an LCCE establishes a control connection, with the
   exchange involving each side alternating in sending messages.  This
   example shows the final acknowledgment explicitly sent within an ACK
   message.  An alternative would be to piggyback the acknowledgment
   within a message sent as a reply to the ICRQ or OCRQ that will likely
   follow from the side that initiated the control connection.

      LCCE A                   LCCE B
      ------                   ------
      SCCRQ     ->
      Nr: 0, Ns: 0
                               <-     SCCRP
                               Nr: 1, Ns: 0
      SCCCN     ->
      Nr: 1, Ns: 1
                               <-       ACK
                               Nr: 2, Ns: 1

B.2: Lost Packet with Retransmission



   An existing control connection has a new session requested by LCCE A.
   The ICRP is lost and must be retransmitted by LCCE B.  Note that loss
   of the ICRP has two effects: It not only keeps the upper level state
   machine from progressing, but also keeps LCCE A from seeing a timely
   lower level acknowledgment of its ICRQ.

        LCCE A                           LCCE B
        ------                           ------
        ICRQ      ->
        Nr: 1, Ns: 2
                         (packet lost)   <-      ICRP
                                         Nr: 3, Ns: 1

      (pause; LCCE A's timer started first, so fires first)

       ICRQ      ->
       Nr: 1, Ns: 2

      (Realizing that it has already seen this packet,
       LCCE B discards the packet and sends an ACK message)

                                         <-       ACK
                                         Nr: 3, Ns: 2




Lau, et al.                 Standards Track                    [Page 90]

RFC 3931                         L2TPv3                       March 2005


      (LCCE B's retransmit timer fires)

                                         <-      ICRP
                                         Nr: 3, Ns: 1
       ICCN      ->
       Nr: 2, Ns: 3

                                         <-       ACK
                                         Nr: 4, Ns: 2

Appendix C: Processing Sequence Numbers



   The Default L2-Specific Sublayer, defined in Section 4.6, provides a
   24-bit field for sequencing of data packets within an L2TP session.
   L2TP data packets are never retransmitted, so this sequence is used
   only to detect packet order, duplicate packets, or lost packets.

   The 24-bit Sequence Number field of the Default L2-Specific Sublayer
   contains a packet sequence number for the associated session.  Each
   sequenced data packet that is sent must contain the sequence number,
   incremented by one, of the previous sequenced packet sent on a given
   L2TP session.  Upon receipt, any packet with a sequence number equal
   to or greater than the current expected packet (the last received
   in-order packet plus one) should be considered "new" and accepted.
   All other packets are considered "old" or "duplicate" and discarded.
   Note that the 24-bit sequence number space includes zero as a valid
   sequence number (as such, it may be implemented with a masked 32-bit
   counter if desired).  All new sessions MUST begin sending sequence
   numbers at zero.

   Larger or smaller sequence number fields are possible with L2TP if an
   alternative format to the Default L2-Specific Sublayer defined in
   this document is used.  While 24 bits may be adequate in a number of
   circumstances, a larger sequence number space will be less
   susceptible to sequence number wrapping problems for very high
   session data rates across long dropout periods.  The sequence number
   processing recommendations below should hold for any size sequence
   number field.

   When detecting whether a packet sequence number is "greater" or
   "less" than a given sequence number value, wrapping of the sequence
   number must be considered.  This is typically accomplished by keeping
   a window of sequence numbers beyond the current expected sequence
   number for determination of whether a packet is "new" or not.  The
   window may be sized based on the link speed and sequence number space
   and SHOULD be configurable with a default equal to one half the size
   of the available number space (e.g., 2^(n-1), where n is the number
   of bits available in the sequence number).



Lau, et al.                 Standards Track                    [Page 91]

RFC 3931                         L2TPv3                       March 2005


   Upon receipt, packets that exactly match the expected sequence number
   are processed immediately and the next expected sequence number
   incremented.  Packets that fall within the window for new packets may
   either be processed immediately and the next expected sequence number
   updated to one plus that received in the new packet, or held for a
   very short period of time in hopes of receiving the missing
   packet(s).  This "very short period" should be configurable, with a
   default corresponding to a time lapse that is at least an order of
   magnitude less than the retransmission timeout periods of higher
   layer protocols such as TCP.

   For typical transient packet mis-orderings, dropping out-of-order
   packets alone should suffice and generally requires far less
   resources than actively reordering packets within L2TP.  An exception
   is a case in which a pair of packet fragments are persistently
   retransmitted and sent out-of-order.  For example, if an IP packet
   has been fragmented into a very small packet followed by a very large
   packet before being tunneled by L2TP, it is possible (though
   admittedly wrong) that the two resulting L2TP packets may be
   consistently mis-ordered by the PSN in transit between L2TP nodes.
   If sequence numbers were being enforced at the receiving node without
   any buffering of out-of-order packets, then the fragmented IP packet
   may never reach its destination.  It may be worth noting here that
   this condition is true for any tunneling mechanism of IP packets that
   includes sequence number checking on receipt (i.e., GRE [RFC2890]).

   Utilization of a Data Sequencing Level (see Section 5.4.3) of 1 (only
   non-IP data packets require sequencing) allows IP data packets being
   tunneled by L2TP to not utilize sequence numbers, while utilizing
   sequence numbers and enforcing packet order for any remaining non-IP
   data packets.  Depending on the requirements of the link layer being
   tunneled and the network data traversing the data link, this is
   sufficient in many cases to enforce packet order on frames that
   require it (such as end-to-end data link control messages), while not
   on IP packets that are known to be resilient to packet reordering.

   If a large number of packets (i.e., more than one new packet window)
   are dropped due to an extended outage or loss of sequence number
   state on one side of the connection (perhaps as part of a forwarding
   plane reset or failover to a standby node), it is possible that a
   large number of packets will be sent in-order, but be wrongly
   detected by the peer as out-of-order.  This can be generally
   characterized for a window size, w, sequence number space, s, and
   number of packets lost in transit between L2TP endpoints, p, as
   follows:






Lau, et al.                 Standards Track                    [Page 92]

RFC 3931                         L2TPv3                       March 2005


   If s > p > w, then an additional (s - p) packets that were otherwise
   received in-order, will be incorrectly classified as out-of-order and
   dropped.  Thus, for a sequence number space, s = 128, window size, w
   = 64, and number of lost packets, p = 70; 128 - 70 = 58 additional
   packets would be dropped after the outage until the sequence number
   wrapped back to the current expected next sequence number.

   To mitigate this additional packet loss, one MUST inspect the
   sequence numbers of packets dropped due to being classified as "old"
   and reset the expected sequence number accordingly.  This may be
   accomplished by counting the number of "old" packets dropped that
   were in sequence among themselves and, upon reaching a threshold,
   resetting the next expected sequence number to that seen in the
   arriving data packets.  Packet timestamps may also be used as an
   indicator to reset the expected sequence number by detecting a period
   of time over which "old" packets have been received in-sequence.  The
   ideal thresholds will vary depending on link speed, sequence number
   space, and link tolerance to out-of-order packets, and MUST be
   configurable.

Editors' Addresses



   Jed Lau
   cisco Systems
   170 W. Tasman Drive
   San Jose, CA  95134

   EMail: jedlau@cisco.com


   W. Mark Townsley
   cisco Systems



   EMail: mark@townsley.net


   Ignacio Goyret
   Lucent Technologies

   EMail: igoyret@lucent.com











Lau, et al.                 Standards Track                    [Page 93]

RFC 3931                         L2TPv3                       March 2005


Full Copyright Statement



   Copyright (C) The Internet Society (2005).

   This document is subject to the rights, licenses and restrictions
   contained in BCP 78, and except as set forth therein, the authors
   retain all their rights.

   This document and the information contained herein are provided on an
   "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS
   OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY AND THE INTERNET
   ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, EXPRESS OR IMPLIED,
   INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE
   INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED
   WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.

Intellectual Property



   The IETF takes no position regarding the validity or scope of any
   Intellectual Property Rights or other rights that might be claimed to
   pertain to the implementation or use of the technology described in
   this document or the extent to which any license under such rights
   might or might not be available; nor does it represent that it has
   made any independent effort to identify any such rights.  Information
   on the procedures with respect to rights in RFC documents can be
   found in BCP 78 and BCP 79.

   Copies of IPR disclosures made to the IETF Secretariat and any
   assurances of licenses to be made available, or the result of an
   attempt made to obtain a general license or permission for the use of
   such proprietary rights by implementers or users of this
   specification can be obtained from the IETF on-line IPR repository at
   http://www.ietf.org/ipr.

   The IETF invites any interested party to bring to its attention any
   copyrights, patents or patent applications, or other proprietary
   rights that may cover technology that may be required to implement
   this standard.  Please address the information to the IETF at ietf-
   ipr@ietf.org.

Acknowledgement



   Funding for the RFC Editor function is currently provided by the
   Internet Society.







Lau, et al.                 Standards Track                    [Page 94]