RFC 8489






Internet Engineering Task Force (IETF)                 M. Petit-Huguenin
Request for Comments: 8489                            Impedance Mismatch
Obsoletes: 5389                                             G. Salgueiro
Category: Standards Track                                          Cisco
ISSN: 2070-1721                                             J. Rosenberg
                                                                   Five9
                                                                 D. Wing
                                                                  Citrix
                                                                 R. Mahy
                                                            Unaffiliated
                                                             P. Matthews
                                                                   Nokia
                                                           February 2020


               Session Traversal Utilities for NAT (STUN)

Abstract



   Session Traversal Utilities for NAT (STUN) is a protocol that serves
   as a tool for other protocols in dealing with NAT traversal.  It can
   be used by an endpoint to determine the IP address and port allocated
   to it by a NAT.  It can also be used to check connectivity between
   two endpoints and as a keep-alive protocol to maintain NAT bindings.
   STUN works with many existing NATs and does not require any special
   behavior from them.

   STUN is not a NAT traversal solution by itself.  Rather, it is a tool
   to be used in the context of a NAT traversal solution.

   This document obsoletes RFC 5389.

Status of This Memo



   This is an Internet Standards Track document.

   This document is a product of the Internet Engineering Task Force
   (IETF).  It represents the consensus of the IETF community.  It has
   received public review and has been approved for publication by the
   Internet Engineering Steering Group (IESG).  Further information on
   Internet Standards is available in Section 2 of RFC 7841.

   Information about the current status of this document, any errata,
   and how to provide feedback on it may be obtained at
   https://www.rfc-editor.org/info/rfc8489.






Petit-Huguenin, et al.       Standards Track                    [Page 1]

RFC 8489                          STUN                     February 2020


Copyright Notice



   Copyright (c) 2020 IETF Trust and the persons identified as the
   document authors.  All rights reserved.

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

Table of Contents



   1. Introduction ....................................................4
   2. Overview of Operation ...........................................5
   3. Terminology .....................................................7
   4. Definitions .....................................................7
   5. STUN Message Structure ..........................................9
   6. Base Protocol Procedures .......................................11
      6.1. Forming a Request or an Indication ........................11
      6.2. Sending the Request or Indication .........................12
           6.2.1. Sending over UDP or DTLS-over-UDP ..................13
           6.2.2. Sending over TCP or TLS-over-TCP ...................14
           6.2.3. Sending over TLS-over-TCP or DTLS-over-UDP .........15
      6.3. Receiving a STUN Message ..................................16
           6.3.1. Processing a Request ...............................17
                  6.3.1.1. Forming a Success or Error Response .......17
                  6.3.1.2. Sending the Success or Error Response .....18
           6.3.2. Processing an Indication ...........................18
           6.3.3. Processing a Success Response ......................19
           6.3.4. Processing an Error Response .......................19
   7. FINGERPRINT Mechanism ..........................................20
   8. DNS Discovery of a Server ......................................20
      8.1. STUN URI Scheme Semantics .................................21
   9. Authentication and Message-Integrity Mechanisms ................22
      9.1. Short-Term Credential Mechanism ...........................23
           9.1.1. HMAC Key ...........................................23
           9.1.2. Forming a Request or Indication ....................23
           9.1.3. Receiving a Request or Indication ..................23
           9.1.4. Receiving a Response ...............................25
           9.1.5. Sending Subsequent Requests ........................25
      9.2. Long-Term Credential Mechanism ............................26
           9.2.1. Bid-Down Attack Prevention .........................27
           9.2.2. HMAC Key ...........................................27



Petit-Huguenin, et al.       Standards Track                    [Page 2]

RFC 8489                          STUN                     February 2020


           9.2.3. Forming a Request ..................................28
                  9.2.3.1. First Request .............................28
                  9.2.3.2. Subsequent Requests .......................29
           9.2.4. Receiving a Request ................................29
           9.2.5. Receiving a Response ...............................31
   10. ALTERNATE-SERVER Mechanism ....................................33
   11. Backwards Compatibility with RFC 3489 .........................34
   12. Basic Server Behavior .........................................34
   13. STUN Usages ...................................................35
   14. STUN Attributes ...............................................36
      14.1. MAPPED-ADDRESS ...........................................37
      14.2. XOR-MAPPED-ADDRESS .......................................38
      14.3. USERNAME .................................................39
      14.4. USERHASH .................................................40
      14.5. MESSAGE-INTEGRITY ........................................40
      14.6. MESSAGE-INTEGRITY-SHA256 .................................41
      14.7. FINGERPRINT ..............................................41
      14.8. ERROR-CODE ...............................................42
      14.9. REALM ....................................................44
      14.10. NONCE ...................................................44
      14.11. PASSWORD-ALGORITHMS .....................................44
      14.12. PASSWORD-ALGORITHM ......................................45
      14.13. UNKNOWN-ATTRIBUTES ......................................45
      14.14. SOFTWARE ................................................46
      14.15. ALTERNATE-SERVER ........................................46
      14.16. ALTERNATE-DOMAIN ........................................46
   15. Operational Considerations ....................................47
   16. Security Considerations .......................................47
      16.1. Attacks against the Protocol .............................47
           16.1.1. Outside Attacks ...................................47
           16.1.2. Inside Attacks ....................................48
           16.1.3. Bid-Down Attacks ..................................48
      16.2. Attacks Affecting the Usage ..............................50
           16.2.1. Attack I: Distributed DoS (DDoS) against a
                   Target ............................................51
           16.2.2. Attack II: Silencing a Client .....................51
           16.2.3. Attack III: Assuming the Identity of a Client .....52
           16.2.4. Attack IV: Eavesdropping ..........................52
      16.3. Hash Agility Plan ........................................52
   17. IAB Considerations ............................................53
   18. IANA Considerations ...........................................53
      18.1. STUN Security Features Registry ..........................53
      18.2. STUN Methods Registry ....................................54
      18.3. STUN Attributes Registry .................................54
           18.3.1. Updated Attributes ................................55
           18.3.2. New Attributes ....................................55
      18.4. STUN Error Codes Registry ................................56
      18.5. STUN Password Algorithms Registry ........................56



Petit-Huguenin, et al.       Standards Track                    [Page 3]

RFC 8489                          STUN                     February 2020


           18.5.1. Password Algorithms ...............................57
                  18.5.1.1. MD5 ......................................57
                  18.5.1.2. SHA-256 ..................................57
      18.6. STUN UDP and TCP Port Numbers ............................57
   19. Changes since RFC 5389 ........................................57
   20. References ....................................................58
      20.1. Normative References .....................................58
      20.2. Informative References ...................................61
   Appendix A.  C Snippet to Determine STUN Message Types ............64
   Appendix B.  Test Vectors .........................................64
     B.1.  Sample Request with Long-Term Authentication with
           MESSAGE-INTEGRITY-SHA256 and USERHASH .....................65
   Acknowledgements ..................................................66
   Contributors ......................................................66
   Authors' Addresses ................................................67

1.  Introduction



   The protocol defined in this specification, Session Traversal
   Utilities for NAT (STUN), provides a tool for dealing with Network
   Address Translators (NATs).  It provides a means for an endpoint to
   determine the IP address and port allocated by a NAT that corresponds
   to its private IP address and port.  It also provides a way for an
   endpoint to keep a NAT binding alive.  With some extensions, the
   protocol can be used to do connectivity checks between two endpoints
   [RFC8445] or to relay packets between two endpoints [RFC5766].

   In keeping with its tool nature, this specification defines an
   extensible packet format, defines operation over several transport
   protocols, and provides for two forms of authentication.

   STUN is intended to be used in the context of one or more NAT
   traversal solutions.  These solutions are known as "STUN Usages".
   Each usage describes how STUN is utilized to achieve the NAT
   traversal solution.  Typically, a usage indicates when STUN messages
   get sent, which optional attributes to include, what server is used,
   and what authentication mechanism is to be used.  Interactive
   Connectivity Establishment (ICE) [RFC8445] is one usage of STUN.  SIP
   Outbound [RFC5626] is another usage of STUN.  In some cases, a usage
   will require extensions to STUN.  A STUN extension can be in the form
   of new methods, attributes, or error response codes.  More
   information on STUN Usages can be found in Section 13.









Petit-Huguenin, et al.       Standards Track                    [Page 4]

RFC 8489                          STUN                     February 2020


2.  Overview of Operation



   This section is descriptive only.

                           /-----\
                         // STUN  \\
                        |   Server  |
                         \\       //
                           \-----/




                      +--------------+             Public Internet
      ................|     NAT 2    |.......................
                      +--------------+



                      +--------------+             Private Network 2
      ................|     NAT 1    |.......................
                      +--------------+




                           /-----\
                         // STUN  \\
                        |   Client  |
                         \\       //               Private Network 1
                           \-----/

                 Figure 1: One Possible STUN Configuration

   One possible STUN configuration is shown in Figure 1.  In this
   configuration, there are two entities (called STUN agents) that
   implement the STUN protocol.  The lower agent in the figure is the
   client, which is connected to private network 1.  This network
   connects to private network 2 through NAT 1.  Private network 2
   connects to the public Internet through NAT 2.  The upper agent in
   the figure is the server, which resides on the public Internet.

   STUN is a client-server protocol.  It supports two types of
   transactions.  One is a request/response transaction in which a
   client sends a request to a server, and the server returns a
   response.  The second is an indication transaction in which either
   agent -- client or server -- sends an indication that generates no
   response.  Both types of transactions include a transaction ID, which



Petit-Huguenin, et al.       Standards Track                    [Page 5]

RFC 8489                          STUN                     February 2020


   is a randomly selected 96-bit number.  For request/response
   transactions, this transaction ID allows the client to associate the
   response with the request that generated it; for indications, the
   transaction ID serves as a debugging aid.

   All STUN messages start with a fixed header that includes a method, a
   class, and the transaction ID.  The method indicates which of the
   various requests or indications this is; this specification defines
   just one method, Binding, but other methods are expected to be
   defined in other documents.  The class indicates whether this is a
   request, a success response, an error response, or an indication.
   Following the fixed header comes zero or more attributes, which are
   Type-Length-Value extensions that convey additional information for
   the specific message.

   This document defines a single method called "Binding".  The Binding
   method can be used either in request/response transactions or in
   indication transactions.  When used in request/response transactions,
   the Binding method can be used to determine the particular binding a
   NAT has allocated to a STUN client.  When used in either request/
   response or in indication transactions, the Binding method can also
   be used to keep these bindings alive.

   In the Binding request/response transaction, a Binding request is
   sent from a STUN client to a STUN server.  When the Binding request
   arrives at the STUN server, it may have passed through one or more
   NATs between the STUN client and the STUN server (in Figure 1, there
   are two such NATs).  As the Binding request message passes through a
   NAT, the NAT will modify the source transport address (that is, the
   source IP address and the source port) of the packet.  As a result,
   the source transport address of the request received by the server
   will be the public IP address and port created by the NAT closest to
   the server.  This is called a "reflexive transport address".  The
   STUN server copies that source transport address into an XOR-MAPPED-
   ADDRESS attribute in the STUN Binding response and sends the Binding
   response back to the STUN client.  As this packet passes back through
   a NAT, the NAT will modify the destination transport address in the
   IP header, but the transport address in the XOR-MAPPED-ADDRESS
   attribute within the body of the STUN response will remain untouched.
   In this way, the client can learn its reflexive transport address
   allocated by the outermost NAT with respect to the STUN server.

   In some usages, STUN must be multiplexed with other protocols (e.g.,
   [RFC8445] and [RFC5626]).  In these usages, there must be a way to
   inspect a packet and determine if it is a STUN packet or not.  STUN
   provides three fields in the STUN header with fixed values that can





Petit-Huguenin, et al.       Standards Track                    [Page 6]

RFC 8489                          STUN                     February 2020


   be used for this purpose.  If this is not sufficient, then STUN
   packets can also contain a FINGERPRINT value, which can further be
   used to distinguish the packets.

   STUN defines a set of optional procedures that a usage can decide to
   use, called "mechanisms".  These mechanisms include DNS discovery, a
   redirection technique to an alternate server, a fingerprint attribute
   for demultiplexing, and two authentication and message-integrity
   exchanges.  The authentication mechanisms revolve around the use of a
   username, password, and message-integrity value.  Two authentication
   mechanisms, the long-term credential mechanism and the short-term
   credential mechanism, are defined in this specification.  Each usage
   specifies the mechanisms allowed with that usage.

   In the long-term credential mechanism, the client and server share a
   pre-provisioned username and password and perform a digest challenge/
   response exchange inspired by the one defined for HTTP [RFC7616] but
   differing in details.  In the short-term credential mechanism, the
   client and the server exchange a username and password through some
   out-of-band method prior to the STUN exchange.  For example, in the
   ICE usage [RFC8445], the two endpoints use out-of-band signaling to
   exchange a username and password.  These are used to integrity
   protect and authenticate the request and response.  There is no
   challenge or nonce used.

3.  Terminology



   The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
   "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and
   "OPTIONAL" in this document are to be interpreted as described in
   BCP 14 [RFC2119] [RFC8174] when, and only when, they appear in all
   capitals, as shown here.

4.  Definitions



   STUN Agent:  A STUN agent is an entity that implements the STUN
      protocol.  The entity can be either a STUN client or a STUN
      server.

   STUN Client:  A STUN client is an entity that sends STUN requests and
      receives STUN responses and STUN indications.  A STUN client can
      also send indications.  In this specification, the terms "STUN
      client" and "client" are synonymous.

   STUN Server:  A STUN server is an entity that receives STUN requests
      and STUN indications and that sends STUN responses.  A STUN server
      can also send indications.  In this specification, the terms "STUN
      server" and "server" are synonymous.



Petit-Huguenin, et al.       Standards Track                    [Page 7]

RFC 8489                          STUN                     February 2020


   Transport Address:  The combination of an IP address and port number
      (such as a UDP or TCP port number).

   Reflexive Transport Address:  A transport address learned by a client
      that identifies that client as seen by another host on an IP
      network, typically a STUN server.  When there is an intervening
      NAT between the client and the other host, the reflexive transport
      address represents the mapped address allocated to the client on
      the public side of the NAT.  Reflexive transport addresses are
      learned from the mapped address attribute (MAPPED-ADDRESS or XOR-
      MAPPED-ADDRESS) in STUN responses.

   Mapped Address:  Same meaning as reflexive address.  This term is
      retained only for historic reasons and due to the naming of the
      MAPPED-ADDRESS and XOR-MAPPED-ADDRESS attributes.

   Long-Term Credential:  A username and associated password that
      represent a shared secret between client and server.  Long-term
      credentials are generally granted to the client when a subscriber
      enrolls in a service and persist until the subscriber leaves the
      service or explicitly changes the credential.

   Long-Term Password:  The password from a long-term credential.

   Short-Term Credential:  A temporary username and associated password
      that represent a shared secret between client and server.  Short-
      term credentials are obtained through some kind of protocol
      mechanism between the client and server, preceding the STUN
      exchange.  A short-term credential has an explicit temporal scope,
      which may be based on a specific amount of time (such as 5
      minutes) or on an event (such as termination of a Session
      Initiation Protocol (SIP) [RFC3261] dialog).  The specific scope
      of a short-term credential is defined by the application usage.

   Short-Term Password:  The password component of a short-term
      credential.

   STUN Indication:  A STUN message that does not receive a response.

   Attribute:  The STUN term for a Type-Length-Value (TLV) object that
      can be added to a STUN message.  Attributes are divided into two
      types: comprehension-required and comprehension-optional.  STUN
      agents can safely ignore comprehension-optional attributes they
      don't understand but cannot successfully process a message if it
      contains comprehension-required attributes that are not
      understood.





Petit-Huguenin, et al.       Standards Track                    [Page 8]

RFC 8489                          STUN                     February 2020


   RTO:  Retransmission TimeOut, which defines the initial period of
      time between transmission of a request and the first retransmit of
      that request.

5.  STUN Message Structure



   STUN messages are encoded in binary using network-oriented format
   (most significant byte or octet first, also commonly known as big-
   endian).  The transmission order is described in detail in Appendix B
   of [RFC0791].  Unless otherwise noted, numeric constants are in
   decimal (base 10).

   All STUN messages comprise a 20-byte header followed by zero or more
   attributes.  The STUN header contains a STUN message type, message
   length, magic cookie, and transaction ID.

      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
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |0 0|     STUN Message Type     |         Message Length        |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |                         Magic Cookie                          |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |                                                               |
     |                     Transaction ID (96 bits)                  |
     |                                                               |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

                  Figure 2: Format of STUN Message Header

   The most significant 2 bits of every STUN message MUST be zeroes.
   This can be used to differentiate STUN packets from other protocols
   when STUN is multiplexed with other protocols on the same port.

   The message type defines the message class (request, success
   response, error response, or indication) and the message method (the
   primary function) of the STUN message.  Although there are four
   message classes, there are only two types of transactions in STUN:
   request/response transactions (which consist of a request message and
   a response message) and indication transactions (which consist of a
   single indication message).  Response classes are split into error
   and success responses to aid in quickly processing the STUN message.









Petit-Huguenin, et al.       Standards Track                    [Page 9]

RFC 8489                          STUN                     February 2020


   The STUN Message Type field is decomposed further into the following
   structure:

                       0                 1
                       2  3  4 5 6 7 8 9 0 1 2 3 4 5
                      +--+--+-+-+-+-+-+-+-+-+-+-+-+-+
                      |M |M |M|M|M|C|M|M|M|C|M|M|M|M|
                      |11|10|9|8|7|1|6|5|4|0|3|2|1|0|
                      +--+--+-+-+-+-+-+-+-+-+-+-+-+-+

                Figure 3: Format of STUN Message Type Field

   Here the bits in the STUN Message Type field are shown as most
   significant (M11) through least significant (M0).  M11 through M0
   represent a 12-bit encoding of the method.  C1 and C0 represent a
   2-bit encoding of the class.  A class of 0b00 is a request, a class
   of 0b01 is an indication, a class of 0b10 is a success response, and
   a class of 0b11 is an error response.  This specification defines a
   single method, Binding.  The method and class are orthogonal, so that
   for each method, a request, success response, error response, and
   indication are possible for that method.  Extensions defining new
   methods MUST indicate which classes are permitted for that method.

   For example, a Binding request has class=0b00 (request) and
   method=0b000000000001 (Binding) and is encoded into the first 16 bits
   as 0x0001.  A Binding response has class=0b10 (success response) and
   method=0b000000000001 and is encoded into the first 16 bits as
   0x0101.

      Note: This unfortunate encoding is due to assignment of values in
      [RFC3489] that did not consider encoding indication messages,
      success responses, and errors responses using bit fields.

   The Magic Cookie field MUST contain the fixed value 0x2112A442 in
   network byte order.  In [RFC3489], the 32 bits comprising the Magic
   Cookie field were part of the transaction ID; placing the magic
   cookie in this location allows a server to detect if the client will
   understand certain attributes that were added to STUN by [RFC5389].
   In addition, it aids in distinguishing STUN packets from packets of
   other protocols when STUN is multiplexed with those other protocols
   on the same port.

   The transaction ID is a 96-bit identifier, used to uniquely identify
   STUN transactions.  For request/response transactions, the
   transaction ID is chosen by the STUN client for the request and
   echoed by the server in the response.  For indications, it is chosen
   by the agent sending the indication.  It primarily serves to
   correlate requests with responses, though it also plays a small role



Petit-Huguenin, et al.       Standards Track                   [Page 10]

RFC 8489                          STUN                     February 2020


   in helping to prevent certain types of attacks.  The server also uses
   the transaction ID as a key to identify each transaction uniquely
   across all clients.  As such, the transaction ID MUST be uniformly
   and randomly chosen from the interval 0 .. 2**96-1 and MUST be
   cryptographically random.  Resends of the same request reuse the same
   transaction ID, but the client MUST choose a new transaction ID for
   new transactions unless the new request is bit-wise identical to the
   previous request and sent from the same transport address to the same
   IP address.  Success and error responses MUST carry the same
   transaction ID as their corresponding request.  When an agent is
   acting as a STUN server and STUN client on the same port, the
   transaction IDs in requests sent by the agent have no relationship to
   the transaction IDs in requests received by the agent.

   The message length MUST contain the size of the message in bytes, not
   including the 20-byte STUN header.  Since all STUN attributes are
   padded to a multiple of 4 bytes, the last 2 bits of this field are
   always zero.  This provides another way to distinguish STUN packets
   from packets of other protocols.

   Following the STUN fixed portion of the header are zero or more
   attributes.  Each attribute is TLV (Type-Length-Value) encoded.
   Details of the encoding and the attributes themselves are given in
   Section 14.

6.  Base Protocol Procedures



   This section defines the base procedures of the STUN protocol.  It
   describes how messages are formed, how they are sent, and how they
   are processed when they are received.  It also defines the detailed
   processing of the Binding method.  Other sections in this document
   describe optional procedures that a usage may elect to use in certain
   situations.  Other documents may define other extensions to STUN, by
   adding new methods, new attributes, or new error response codes.

6.1.  Forming a Request or an Indication



   When formulating a request or indication message, the agent MUST
   follow the rules in Section 5 when creating the header.  In addition,
   the message class MUST be either "Request" or "Indication" (as
   appropriate), and the method must be either Binding or some method
   defined in another document.

   The agent then adds any attributes specified by the method or the
   usage.  For example, some usages may specify that the agent use an
   authentication method (Section 9) or the FINGERPRINT attribute
   (Section 7).




Petit-Huguenin, et al.       Standards Track                   [Page 11]

RFC 8489                          STUN                     February 2020


   If the agent is sending a request, it SHOULD add a SOFTWARE attribute
   to the request.  Agents MAY include a SOFTWARE attribute in
   indications, depending on the method.  Extensions to STUN should
   discuss whether SOFTWARE is useful in new indications.  Note that the
   inclusion of a SOFTWARE attribute may have security implications; see
   Section 16.1.2 for details.

   For the Binding method with no authentication, no attributes are
   required unless the usage specifies otherwise.

   All STUN messages sent over UDP or DTLS-over-UDP [RFC6347] SHOULD be
   less than the path MTU, if known.

   If the path MTU is unknown for UDP, messages SHOULD be the smaller of
   576 bytes and the first-hop MTU for IPv4 [RFC1122] and 1280 bytes for
   IPv6 [RFC8200].  This value corresponds to the overall size of the IP
   packet.  Consequently, for IPv4, the actual STUN message would need
   to be less than 548 bytes (576 minus 20-byte IP header, minus 8-byte
   UDP header, assuming no IP options are used).

   If the path MTU is unknown for DTLS-over-UDP, the rules described in
   the previous paragraph need to be adjusted to take into account the
   size of the (13-byte) DTLS Record header, the Message Authentication
   Code (MAC) size, and the padding size.

   STUN provides no ability to handle the case where the request is
   smaller than the MTU but the response is larger than the MTU.  It is
   not envisioned that this limitation will be an issue for STUN.  The
   MTU limitation is a SHOULD, not a MUST, to account for cases where
   STUN itself is being used to probe for MTU characteristics [RFC5780].
   See also [STUN-PMTUD] for a framework that uses STUN to add Path MTU
   Discovery to protocols that lack such a mechanism.  Outside of this
   or similar applications, the MTU constraint MUST be followed.

6.2.  Sending the Request or Indication



   The agent then sends the request or indication.  This document
   specifies how to send STUN messages over UDP, TCP, TLS-over-TCP, or
   DTLS-over-UDP; other transport protocols may be added in the future.
   The STUN Usage must specify which transport protocol is used and how
   the agent determines the IP address and port of the recipient.
   Section 8 describes a DNS-based method of determining the IP address
   and port of a server that a usage may elect to use.

   At any time, a client MAY have multiple outstanding STUN requests
   with the same STUN server (that is, multiple transactions in
   progress, with different transaction IDs).  Absent other limits to




Petit-Huguenin, et al.       Standards Track                   [Page 12]

RFC 8489                          STUN                     February 2020


   the rate of new transactions (such as those specified by ICE for
   connectivity checks or when STUN is run over TCP), a client SHOULD
   limit itself to ten outstanding transactions to the same server.

6.2.1.  Sending over UDP or DTLS-over-UDP



   When running STUN over UDP or STUN over DTLS-over-UDP [RFC7350], it
   is possible that the STUN message might be dropped by the network.
   Reliability of STUN request/response transactions is accomplished
   through retransmissions of the request message by the client
   application itself.  STUN indications are not retransmitted; thus,
   indication transactions over UDP or DTLS-over-UDP are not reliable.

   A client SHOULD retransmit a STUN request message starting with an
   interval of RTO ("Retransmission TimeOut"), doubling after each
   retransmission.  The RTO is an estimate of the round-trip time (RTT)
   and is computed as described in [RFC6298], with two exceptions.
   First, the initial value for RTO SHOULD be greater than or equal to
   500 ms.  The exception cases for this "SHOULD" are when other
   mechanisms are used to derive congestion thresholds (such as the ones
   defined in ICE for fixed-rate streams) or when STUN is used in non-
   Internet environments with known network capacities.  In fixed-line
   access links, a value of 500 ms is RECOMMENDED.  Second, the value of
   RTO SHOULD NOT be rounded up to the nearest second.  Rather, a 1 ms
   accuracy SHOULD be maintained.  As with TCP, the usage of Karn's
   algorithm is RECOMMENDED [KARN87].  When applied to STUN, it means
   that RTT estimates SHOULD NOT be computed from STUN transactions that
   result in the retransmission of a request.

   The value for RTO SHOULD be cached by a client after the completion
   of the transaction and used as the starting value for RTO for the
   next transaction to the same server (based on equality of IP
   address).  The value SHOULD be considered stale and discarded if no
   transactions have occurred to the same server in the last 10 minutes.

   Retransmissions continue until a response is received or until a
   total of Rc requests have been sent.  Rc SHOULD be configurable and
   SHOULD have a default of 7.  If, after the last request, a duration
   equal to Rm times the RTO has passed without a response (providing
   ample time to get a response if only this final request actually
   succeeds), the client SHOULD consider the transaction to have failed.
   Rm SHOULD be configurable and SHOULD have a default of 16.  A STUN
   transaction over UDP or DTLS-over-UDP is also considered failed if
   there has been a hard ICMP error [RFC1122].  For example, assuming an
   RTO of 500 ms, requests would be sent at times 0 ms, 500 ms, 1500 ms,
   3500 ms, 7500 ms, 15500 ms, and 31500 ms.  If the client has not
   received a response after 39500 ms, the client will consider the
   transaction to have timed out.



Petit-Huguenin, et al.       Standards Track                   [Page 13]

RFC 8489                          STUN                     February 2020


6.2.2.  Sending over TCP or TLS-over-TCP



   For TCP and TLS-over-TCP [RFC8446], the client opens a TCP connection
   to the server.

   In some usages of STUN, STUN is the only protocol over the TCP
   connection.  In this case, it can be sent without the aid of any
   additional framing or demultiplexing.  In other usages, or with other
   extensions, it may be multiplexed with other data over a TCP
   connection.  In that case, STUN MUST be run on top of some kind of
   framing protocol, specified by the usage or extension, which allows
   for the agent to extract complete STUN messages and complete
   application-layer messages.  The STUN service running on the well-
   known port or ports discovered through the DNS procedures in
   Section 8 is for STUN alone, and not for STUN multiplexed with other
   data.  Consequently, no framing protocols are used in connections to
   those servers.  When additional framing is utilized, the usage will
   specify how the client knows to apply it and what port to connect to.
   For example, in the case of ICE connectivity checks, this information
   is learned through out-of-band negotiation between client and server.

   Reliability of STUN over TCP and TLS-over-TCP is handled by TCP
   itself, and there are no retransmissions at the STUN protocol level.
   However, for a request/response transaction, if the client has not
   received a response by Ti seconds after it sent the request message,
   it considers the transaction to have timed out.  Ti SHOULD be
   configurable and SHOULD have a default of 39.5 s.  This value has
   been chosen to equalize the TCP and UDP timeouts for the default
   initial RTO.

   In addition, if the client is unable to establish the TCP connection,
   or the TCP connection is reset or fails before a response is
   received, any request/response transaction in progress is considered
   to have failed.

   The client MAY send multiple transactions over a single TCP (or TLS-
   over-TCP) connection, and it MAY send another request before
   receiving a response to the previous request.  The client SHOULD keep
   the connection open until it:

   o  has no further STUN requests or indications to send over that
      connection,

   o  has no plans to use any resources (such as a mapped address
      (MAPPED-ADDRESS or XOR-MAPPED-ADDRESS) or relayed address
      [RFC5766]) that were learned though STUN requests sent over that
      connection,




Petit-Huguenin, et al.       Standards Track                   [Page 14]

RFC 8489                          STUN                     February 2020


   o  if multiplexing other application protocols over that port, has
      finished using those other protocols,

   o  if using that learned port with a remote peer, has established
      communications with that remote peer, as is required by some TCP
      NAT traversal techniques (e.g., [RFC6544]).

   The details of an eventual keep-alive mechanism are left to each STUN
   Usage.  In any case, if a transaction fails because an idle TCP
   connection doesn't work anymore, the client SHOULD send a RST and try
   to open a new TCP connection.

   At the server end, the server SHOULD keep the connection open and let
   the client close it, unless the server has determined that the
   connection has timed out (for example, due to the client
   disconnecting from the network).  Bindings learned by the client will
   remain valid in intervening NATs only while the connection remains
   open.  Only the client knows how long it needs the binding.  The
   server SHOULD NOT close a connection if a request was received over
   that connection for which a response was not sent.  A server MUST NOT
   ever open a connection back towards the client in order to send a
   response.  Servers SHOULD follow best practices regarding connection
   management in cases of overload.

6.2.3.  Sending over TLS-over-TCP or DTLS-over-UDP



   When STUN is run by itself over TLS-over-TCP or DTLS-over-UDP, the
   TLS_DHE_RSA_WITH_AES_128_GCM_SHA256 and
   TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256 ciphersuites MUST be
   implemented (for compatibility with older versions of this protocol),
   except if deprecated by rules of a specific STUN usage.  Other
   ciphersuites MAY be implemented.  Note that STUN clients and servers
   that implement TLS version 1.3 [RFC8446] or subsequent versions are
   also required to implement mandatory ciphersuites from those
   specifications and SHOULD disable usage of deprecated ciphersuites
   when they detect support for those specifications.  Perfect Forward
   Secrecy (PFS) ciphersuites MUST be preferred over non-PFS
   ciphersuites.  Ciphersuites with known weaknesses, such as those
   based on (single) DES and RC4, MUST NOT be used.  Implementations
   MUST disable TLS-level compression.

   These recommendations are just a part of the recommendations in
   [BCP195] that implementations and deployments of a STUN Usage using
   TLS or DTLS MUST follow.

   When it receives the TLS Certificate message, the client MUST verify
   the certificate and inspect the site identified by the certificate.
   If the certificate is invalid or revoked, or if it does not identify



Petit-Huguenin, et al.       Standards Track                   [Page 15]

RFC 8489                          STUN                     February 2020


   the appropriate party, the client MUST NOT send the STUN message or
   otherwise proceed with the STUN transaction.  The client MUST verify
   the identity of the server.  To do that, it follows the
   identification procedures defined in [RFC6125], with a certificate
   containing an identifier of type DNS-ID or CN-ID, optionally with a
   wildcard character as the leftmost label, but not of type SRV-ID or
   URI-ID.

   When STUN is run multiplexed with other protocols over a TLS-over-TCP
   connection or a DTLS-over-UDP association, the mandatory ciphersuites
   and TLS handling procedures operate as defined by those protocols.

6.3.  Receiving a STUN Message



   This section specifies the processing of a STUN message.  The
   processing specified here is for STUN messages as defined in this
   specification; additional rules for backwards compatibility are
   defined in Section 11.  Those additional procedures are optional, and
   usages can elect to utilize them.  First, a set of processing
   operations is applied that is independent of the class.  This is
   followed by class-specific processing, described in the subsections
   that follow.

   When a STUN agent receives a STUN message, it first checks that the
   message obeys the rules of Section 5.  It checks that the first two
   bits are 0, that the Magic Cookie field has the correct value, that
   the message length is sensible, and that the method value is a
   supported method.  It checks that the message class is allowed for
   the particular method.  If the message class is "Success Response" or
   "Error Response", the agent checks that the transaction ID matches a
   transaction that is still in progress.  If the FINGERPRINT extension
   is being used, the agent checks that the FINGERPRINT attribute is
   present and contains the correct value.  If any errors are detected,
   the message is silently discarded.  In the case when STUN is being
   multiplexed with another protocol, an error may indicate that this is
   not really a STUN message; in this case, the agent should try to
   parse the message as a different protocol.

   The STUN agent then does any checks that are required by a
   authentication mechanism that the usage has specified (see
   Section 9).

   Once the authentication checks are done, the STUN agent checks for
   unknown attributes and known-but-unexpected attributes in the
   message.  Unknown comprehension-optional attributes MUST be ignored
   by the agent.  Known-but-unexpected attributes SHOULD be ignored by
   the agent.  Unknown comprehension-required attributes cause
   processing that depends on the message class and is described below.



Petit-Huguenin, et al.       Standards Track                   [Page 16]

RFC 8489                          STUN                     February 2020


   At this point, further processing depends on the message class of the
   request.

6.3.1.  Processing a Request



   If the request contains one or more unknown comprehension-required
   attributes, the server replies with an error response with an error
   code of 420 (Unknown Attribute) and includes an UNKNOWN-ATTRIBUTES
   attribute in the response that lists the unknown comprehension-
   required attributes.

   Otherwise, the server then does any additional checking that the
   method or the specific usage requires.  If all the checks succeed,
   the server formulates a success response as described below.

   When run over UDP or DTLS-over-UDP, a request received by the server
   could be the first request of a transaction or could be a
   retransmission.  The server MUST respond to retransmissions such that
   the following property is preserved: if the client receives the
   response to the retransmission and not the response that was sent to
   the original request, the overall state on the client and server is
   identical to the case where only the response to the original
   retransmission is received or where both responses are received (in
   which case the client will use the first).  The easiest way to meet
   this requirement is for the server to remember all transaction IDs
   received over UDP or DTLS-over-UDP and their corresponding responses
   in the last 40 seconds.  However, this requires the server to hold
   state and is inappropriate for any requests that are not
   authenticated.  Another way is to reprocess the request and recompute
   the response.  The latter technique MUST only be applied to requests
   that are idempotent (a request is considered idempotent when the same
   request can be safely repeated without impacting the overall state of
   the system) and result in the same success response for the same
   request.  The Binding method is considered to be idempotent.  Note
   that there are certain rare network events that could cause the
   reflexive transport address value to change, resulting in a different
   mapped address in different success responses.  Extensions to STUN
   MUST discuss the implications of request retransmissions on servers
   that do not store transaction state.

6.3.1.1.  Forming a Success or Error Response



   When forming the response (success or error), the server follows the
   rules of Section 6.  The method of the response is the same as that
   of the request, and the message class is either "Success Response" or
   "Error Response".





Petit-Huguenin, et al.       Standards Track                   [Page 17]

RFC 8489                          STUN                     February 2020


   For an error response, the server MUST add an ERROR-CODE attribute
   containing the error code specified in the processing above.  The
   reason phrase is not fixed but SHOULD be something suitable for the
   error code.  For certain errors, additional attributes are added to
   the message.  These attributes are spelled out in the description
   where the error code is specified.  For example, for an error code of
   420 (Unknown Attribute), the server MUST include an UNKNOWN-
   ATTRIBUTES attribute.  Certain authentication errors also cause
   attributes to be added (see Section 9).  Extensions may define other
   errors and/or additional attributes to add in error cases.

   If the server authenticated the request using an authentication
   mechanism, then the server SHOULD add the appropriate authentication
   attributes to the response (see Section 9).

   The server also adds any attributes required by the specific method
   or usage.  In addition, the server SHOULD add a SOFTWARE attribute to
   the message.

   For the Binding method, no additional checking is required unless the
   usage specifies otherwise.  When forming the success response, the
   server adds an XOR-MAPPED-ADDRESS attribute to the response; this
   attribute contains the source transport address of the request
   message.  For UDP or DTLS-over-UDP, this is the source IP address and
   source UDP port of the request message.  For TCP and TLS-over-TCP,
   this is the source IP address and source TCP port of the TCP
   connection as seen by the server.

6.3.1.2.  Sending the Success or Error Response



   The response (success or error) is sent over the same transport as
   the request was received on.  If the request was received over UDP or
   DTLS-over-UDP, the destination IP address and port of the response
   are the source IP address and port of the received request message,
   and the source IP address and port of the response are equal to the
   destination IP address and port of the received request message.  If
   the request was received over TCP or TLS-over-TCP, the response is
   sent back on the same TCP connection as the request was received on.

   The server is allowed to send responses in a different order than it
   received the requests.

6.3.2.  Processing an Indication



   If the indication contains unknown comprehension-required attributes,
   the indication is discarded and processing ceases.





Petit-Huguenin, et al.       Standards Track                   [Page 18]

RFC 8489                          STUN                     February 2020


   Otherwise, the agent then does any additional checking that the
   method or the specific usage requires.  If all the checks succeed,
   the agent then processes the indication.  No response is generated
   for an indication.

   For the Binding method, no additional checking or processing is
   required, unless the usage specifies otherwise.  The mere receipt of
   the message by the agent has refreshed the bindings in the
   intervening NATs.

   Since indications are not re-transmitted over UDP or DTLS-over-UDP
   (unlike requests), there is no need to handle re-transmissions of
   indications at the sending agent.

6.3.3.  Processing a Success Response



   If the success response contains unknown comprehension-required
   attributes, the response is discarded and the transaction is
   considered to have failed.

   Otherwise, the client then does any additional checking that the
   method or the specific usage requires.  If all the checks succeed,
   the client then processes the success response.

   For the Binding method, the client checks that the XOR-MAPPED-ADDRESS
   attribute is present in the response.  The client checks the address
   family specified.  If it is an unsupported address family, the
   attribute SHOULD be ignored.  If it is an unexpected but supported
   address family (for example, the Binding transaction was sent over
   IPv4, but the address family specified is IPv6), then the client MAY
   accept and use the value.

6.3.4.  Processing an Error Response



   If the error response contains unknown comprehension-required
   attributes, or if the error response does not contain an ERROR-CODE
   attribute, then the transaction is simply considered to have failed.

   Otherwise, the client then does any processing specified by the
   authentication mechanism (see Section 9).  This may result in a new
   transaction attempt.

   The processing at this point depends on the error code, the method,
   and the usage; the following are the default rules:

   o  If the error code is 300 through 399, the client SHOULD consider
      the transaction as failed unless the ALTERNATE-SERVER extension
      (Section 10) is being used.



Petit-Huguenin, et al.       Standards Track                   [Page 19]

RFC 8489                          STUN                     February 2020


   o  If the error code is 400 through 499, the client declares the
      transaction failed; in the case of 420 (Unknown Attribute), the
      response should contain a UNKNOWN-ATTRIBUTES attribute that gives
      additional information.

   o  If the error code is 500 through 599, the client MAY resend the
      request; clients that do so MUST limit the number of times they do
      this.  Unless a specific error code specifies a different value,
      the number of retransmissions SHOULD be limited to 4.

   Any other error code causes the client to consider the transaction
   failed.

7.  FINGERPRINT Mechanism



   This section describes an optional mechanism for STUN that aids in
   distinguishing STUN messages from packets of other protocols when the
   two are multiplexed on the same transport address.  This mechanism is
   optional, and a STUN Usage must describe if and when it is used.  The
   FINGERPRINT mechanism is not backwards compatible with RFC 3489 and
   cannot be used in environments where such compatibility is required.

   In some usages, STUN messages are multiplexed on the same transport
   address as other protocols, such as the Real-Time Transport Protocol
   (RTP).  In order to apply the processing described in Section 6, STUN
   messages must first be separated from the application packets.

   Section 5 describes three fixed fields in the STUN header that can be
   used for this purpose.  However, in some cases, these three fixed
   fields may not be sufficient.

   When the FINGERPRINT extension is used, an agent includes the
   FINGERPRINT attribute in messages it sends to another agent.
   Section 14.7 describes the placement and value of this attribute.

   When the agent receives what it believes is a STUN message, then, in
   addition to other basic checks, the agent also checks that the
   message contains a FINGERPRINT attribute and that the attribute
   contains the correct value.  Section 6.3 describes when in the
   overall processing of a STUN message the FINGERPRINT check is
   performed.  This additional check helps the agent detect messages of
   other protocols that might otherwise seem to be STUN messages.

8.  DNS Discovery of a Server



   This section describes an optional procedure for STUN that allows a
   client to use DNS to determine the IP address and port of a server.
   A STUN Usage must describe if and when this extension is used.  To



Petit-Huguenin, et al.       Standards Track                   [Page 20]

RFC 8489                          STUN                     February 2020


   use this procedure, the client must know a STUN URI [RFC7064]; the
   usage must also describe how the client obtains this URI.  Hard-
   coding a STUN URI into software is NOT RECOMMENDED in case the domain
   name is lost or needs to change for legal or other reasons.

   When a client wishes to locate a STUN server on the public Internet
   that accepts Binding request/response transactions, the STUN URI
   scheme is "stun".  When it wishes to locate a STUN server that
   accepts Binding request/response transactions over a TLS or DTLS
   session, the URI scheme is "stuns".

   The syntax of the "stun" and "stuns" URIs is defined in Section 3.1
   of [RFC7064].  STUN Usages MAY define additional URI schemes.

8.1.  STUN URI Scheme Semantics



   If the <host> part of a "stun" URI contains an IP address, then this
   IP address is used directly to contact the server.  A "stuns" URI
   containing an IP address MUST be rejected.  A future STUN extension
   or usage may relax this requirement, provided it demonstrates how to
   authenticate the STUN server and prevent man-in-the-middle attacks.

   If the URI does not contain an IP address, the domain name contained
   in the <host> part is resolved to a transport address using the SRV
   procedures specified in [RFC2782].  The DNS SRV service name is the
   content of the <scheme> part.  The protocol in the SRV lookup is the
   transport protocol the client will run STUN over: "udp" for UDP and
   "tcp" for TCP.

   The procedures of RFC 2782 are followed to determine the server to
   contact.  RFC 2782 spells out the details of how a set of SRV records
   is sorted and then tried.  However, RFC 2782 only states that the
   client should "try to connect to the (protocol, address, service)"
   without giving any details on what happens in the event of failure.
   When following these procedures, if the STUN transaction times out
   without receipt of a response, the client SHOULD retry the request to
   the next server in the order defined by RFC 2782.  Such a retry is
   only possible for request/response transmissions, since indication
   transactions generate no response or timeout.

   In addition, instead of querying either the A or the AAAA resource
   records for a domain name, a dual-stack IPv4/IPv6 client MUST query
   both and try the requests with all the IP addresses received, as
   specified in [RFC8305].

   The default port for STUN requests is 3478, for both TCP and UDP.
   The default port for STUN over TLS and STUN over DTLS requests is
   5349.  Servers can run STUN over DTLS on the same port as STUN over



Petit-Huguenin, et al.       Standards Track                   [Page 21]

RFC 8489                          STUN                     February 2020


   UDP if the server software supports determining whether the initial
   message is a DTLS or STUN message.  Servers can run STUN over TLS on
   the same port as STUN over TCP if the server software supports
   determining whether the initial message is a TLS or STUN message.

   Administrators of STUN servers SHOULD use these ports in their SRV
   records for UDP and TCP.  In all cases, the port in DNS MUST reflect
   the one on which the server is listening.

   If no SRV records are found, the client performs both an A and AAAA
   record lookup of the domain name, as described in [RFC8305].  The
   result will be a list of IP addresses, each of which can be
   simultaneously contacted at the default port using UDP or TCP,
   independent of the STUN Usage.  For usages that require TLS, the
   client connects to the IP addresses using the default STUN over TLS
   port.  For usages that require DTLS, the client connects to the IP
   addresses using the default STUN over DTLS port.

9.  Authentication and Message-Integrity Mechanisms



   This section defines two mechanisms for STUN that a client and server
   can use to provide authentication and message integrity; these two
   mechanisms are known as the short-term credential mechanism and the
   long-term credential mechanism.  These two mechanisms are optional,
   and each usage must specify if and when these mechanisms are used.
   Consequently, both clients and servers will know which mechanism (if
   any) to follow based on knowledge of which usage applies.  For
   example, a STUN server on the public Internet supporting ICE would
   have no authentication, whereas the STUN server functionality in an
   agent supporting connectivity checks would utilize short-term
   credentials.  An overview of these two mechanisms is given in
   Section 2.

   Each mechanism specifies the additional processing required to use
   that mechanism, extending the processing specified in Section 6.  The
   additional processing occurs in three different places: when forming
   a message, when receiving a message immediately after the basic
   checks have been performed, and when doing the detailed processing of
   error responses.

   Note that agents MUST ignore all attributes that follow MESSAGE-
   INTEGRITY, with the exception of the MESSAGE-INTEGRITY-SHA256 and
   FINGERPRINT attributes.  Similarly, agents MUST ignore all attributes
   that follow the MESSAGE-INTEGRITY-SHA256 attribute if the MESSAGE-
   INTEGRITY attribute is not present, with the exception of the
   FINGERPRINT attribute.





Petit-Huguenin, et al.       Standards Track                   [Page 22]

RFC 8489                          STUN                     February 2020


9.1.  Short-Term Credential Mechanism



   The short-term credential mechanism assumes that, prior to the STUN
   transaction, the client and server have used some other protocol to
   exchange a credential in the form of a username and password.  This
   credential is time-limited.  The time limit is defined by the usage.
   As an example, in the ICE usage [RFC8445], the two endpoints use out-
   of-band signaling to agree on a username and password, and this
   username and password are applicable for the duration of the media
   session.

   This credential is used to form a message-integrity check in each
   request and in many responses.  There is no challenge and response as
   in the long-term mechanism; consequently, replay is limited by virtue
   of the time-limited nature of the credential.

9.1.1.  HMAC Key



   For short-term credentials, the Hash-Based Message Authentication
   Code (HMAC) key is defined as follow:

                       key = OpaqueString(password)

   where the OpaqueString profile is defined in [RFC8265].  The encoding
   used is UTF-8 [RFC3629].

9.1.2.  Forming a Request or Indication



   For a request or indication message, the agent MUST include the
   USERNAME, MESSAGE-INTEGRITY-SHA256, and MESSAGE-INTEGRITY attributes
   in the message unless the agent knows from an external mechanism
   which message integrity algorithm is supported by both agents.  In
   this case, either MESSAGE-INTEGRITY or MESSAGE-INTEGRITY-SHA256 MUST
   be included in addition to USERNAME.  The HMAC for the MESSAGE-
   INTEGRITY attribute is computed as described in Section 14.5, and the
   HMAC for the MESSAGE-INTEGRITY-SHA256 attributes is computed as
   described in Section 14.6.  Note that the password is never included
   in the request or indication.

9.1.3.  Receiving a Request or Indication



   After the agent has done the basic processing of a message, the agent
   performs the checks listed below in the order specified:

   o  If the message does not contain 1) a MESSAGE-INTEGRITY or a
      MESSAGE-INTEGRITY-SHA256 attribute and 2) a USERNAME attribute:





Petit-Huguenin, et al.       Standards Track                   [Page 23]

RFC 8489                          STUN                     February 2020


      *  If the message is a request, the server MUST reject the request
         with an error response.  This response MUST use an error code
         of 400 (Bad Request).

      *  If the message is an indication, the agent MUST silently
         discard the indication.

   o  If the USERNAME does not contain a username value currently valid
      within the server:

      *  If the message is a request, the server MUST reject the request
         with an error response.  This response MUST use an error code
         of 401 (Unauthenticated).

      *  If the message is an indication, the agent MUST silently
         discard the indication.

   o  If the MESSAGE-INTEGRITY-SHA256 attribute is present, compute the
      value for the message integrity as described in Section 14.6,
      using the password associated with the username.  If the MESSAGE-
      INTEGRITY-SHA256 attribute is not present, then use the same
      password to compute the value for the message integrity as
      described in Section 14.5.  If the resulting value does not match
      the contents of the corresponding attribute (MESSAGE-INTEGRITY-
      SHA256 or MESSAGE-INTEGRITY):

      *  If the message is a request, the server MUST reject the request
         with an error response.  This response MUST use an error code
         of 401 (Unauthenticated).

      *  If the message is an indication, the agent MUST silently
         discard the indication.

   If these checks pass, the agent continues to process the request or
   indication.  Any response generated by a server to a request that
   contains a MESSAGE-INTEGRITY-SHA256 attribute MUST include the
   MESSAGE-INTEGRITY-SHA256 attribute, computed using the password
   utilized to authenticate the request.  Any response generated by a
   server to a request that contains only a MESSAGE-INTEGRITY attribute
   MUST include the MESSAGE-INTEGRITY attribute, computed using the
   password utilized to authenticate the request.  This means that only
   one of these attributes can appear in a response.  The response MUST
   NOT
contain the USERNAME attribute.








Petit-Huguenin, et al.       Standards Track                   [Page 24]

RFC 8489                          STUN                     February 2020


   If any of the checks fail, a server MUST NOT include a MESSAGE-
   INTEGRITY-SHA256, MESSAGE-INTEGRITY, or USERNAME attribute in the
   error response.  This is because, in these failure cases, the server
   cannot determine the shared secret necessary to compute the MESSAGE-
   INTEGRITY-SHA256 or MESSAGE-INTEGRITY attributes.

9.1.4.  Receiving a Response



   The client looks for the MESSAGE-INTEGRITY or the MESSAGE-INTEGRITY-
   SHA256 attribute in the response.  If present and if the client only
   sent one of the MESSAGE-INTEGRITY or MESSAGE-INTEGRITY-SHA256
   attributes in the request (because of the external indication in
   Section 9.1.2 or because this is a subsequent request as defined in
   Section 9.1.5), the algorithm in the response has to match;
   otherwise, the response MUST be discarded.

   The client then computes the message integrity over the response as
   defined in Section 14.5 for the MESSAGE-INTEGRITY attribute or
   Section 14.6 for the MESSAGE-INTEGRITY-SHA256 attribute, using the
   same password it utilized for the request.  If the resulting value
   matches the contents of the MESSAGE-INTEGRITY or MESSAGE-INTEGRITY-
   SHA256 attribute, respectively, the response is considered
   authenticated.  If the value does not match, or if both MESSAGE-
   INTEGRITY and MESSAGE-INTEGRITY-SHA256 are absent, the processing
   depends on whether the request was sent over a reliable or an
   unreliable transport.

   If the request was sent over an unreliable transport, the response
   MUST be discarded, as if it had never been received.  This means that
   retransmits, if applicable, will continue.  If all the responses
   received are discarded, then instead of signaling a timeout after
   ending the transaction, the layer MUST signal that the integrity
   protection was violated.

   If the request was sent over a reliable transport, the response MUST
   be discarded, and the layer MUST immediately end the transaction and
   signal that the integrity protection was violated.

9.1.5.  Sending Subsequent Requests



   A client sending subsequent requests to the same server MUST send
   only the MESSAGE-INTEGRITY-SHA256 or the MESSAGE-INTEGRITY attribute
   that matches the attribute that was received in the response to the
   initial request.  Here, "same server" means same IP address and port
   number, not just the same URI or SRV lookup result.






Petit-Huguenin, et al.       Standards Track                   [Page 25]

RFC 8489                          STUN                     February 2020


9.2.  Long-Term Credential Mechanism



   The long-term credential mechanism relies on a long-term credential,
   in the form of a username and password that are shared between client
   and server.  The credential is considered long-term since it is
   assumed that it is provisioned for a user and remains in effect until
   the user is no longer a subscriber of the system or until it is
   changed.  This is basically a traditional "log-in" username and
   password given to users.

   Because these usernames and passwords are expected to be valid for
   extended periods of time, replay prevention is provided in the form
   of a digest challenge.  In this mechanism, the client initially sends
   a request, without offering any credentials or any integrity checks.
   The server rejects this request, providing the user a realm (used to
   guide the user or agent in selection of a username and password) and
   a nonce.  The nonce provides a limited replay protection.  It is a
   cookie, selected by the server and encoded in such a way as to
   indicate a duration of validity or client identity from which it is
   valid.  Only the server needs to know about the internal structure of
   the cookie.  The client retries the request, this time including its
   username and the realm and echoing the nonce provided by the server.
   The client also includes one of the message-integrity attributes
   defined in this document, which provides an HMAC over the entire
   request, including the nonce.  The server validates the nonce and
   checks the message integrity.  If they match, the request is
   authenticated.  If the nonce is no longer valid, it is considered
   "stale", and the server rejects the request, providing a new nonce.

   In subsequent requests to the same server, the client reuses the
   nonce, username, realm, and password it used previously.  In this
   way, subsequent requests are not rejected until the nonce becomes
   invalid by the server, in which case the rejection provides a new
   nonce to the client.

   Note that the long-term credential mechanism cannot be used to
   protect indications, since indications cannot be challenged.  Usages
   utilizing indications must either use a short-term credential or omit
   authentication and message integrity for them.

   To indicate that it supports this specification, a server MUST
   prepend the NONCE attribute value with the character string composed
   of "obMatJos2" concatenated with the (4-character) base64 [RFC4648]
   encoding of the 24-bit STUN Security Features as defined in
   Section 18.1.  The 24-bit Security Feature set is encoded as 3 bytes,
   with bit 0 as the most significant bit of the first byte and bit 23
   as the least significant bit of the third byte.  If no security
   features are used, then a byte array with all 24 bits set to zero



Petit-Huguenin, et al.       Standards Track                   [Page 26]

RFC 8489                          STUN                     February 2020


   MUST be encoded instead.  For the remainder of this document, the
   term "nonce cookie" will refer to the complete 13-character string
   prepended to the NONCE attribute value.

   Since the long-term credential mechanism is susceptible to offline
   dictionary attacks, deployments SHOULD utilize passwords that are
   difficult to guess.  In cases where the credentials are not entered
   by the user, but are rather placed on a client device during device
   provisioning, the password SHOULD have at least 128 bits of
   randomness.  In cases where the credentials are entered by the user,
   they should follow best current practices around password structure.

9.2.1.  Bid-Down Attack Prevention



   This document introduces two new security features that provide the
   ability to choose the algorithm used for password protection as well
   as the ability to use an anonymous username.  Both of these
   capabilities are optional in order to remain backwards compatible
   with previous versions of the STUN protocol.

   These new capabilities are subject to bid-down attacks whereby an
   attacker in the message path can remove these capabilities and force
   weaker security properties.  To prevent these kinds of attacks from
   going undetected, the nonce is enhanced with additional information.

   The value of the "nonce cookie" will vary based on the specific STUN
   Security Feature bits selected.  When this document makes reference
   to the "nonce cookie" in a section discussing a specific STUN
   Security Feature it is understood that the corresponding STUN
   Security Feature bit in the "nonce cookie" is set to 1.

   For example, when the PASSWORD-ALGORITHMS security feature (defined
   in Section 9.2.4) is used, the corresponding "Password algorithms"
   bit (defined in Section 18.1) is set to 1 in the "nonce cookie".

9.2.2.  HMAC Key



   For long-term credentials that do not use a different algorithm, as
   specified by the PASSWORD-ALGORITHM attribute, the key is 16 bytes:

                key = MD5(username ":" OpaqueString(realm)
                  ":" OpaqueString(password))

   Where MD5 is defined in [RFC1321] and [RFC6151], and the OpaqueString
   profile is defined in [RFC8265].  The encoding used is UTF-8
   [RFC3629].





Petit-Huguenin, et al.       Standards Track                   [Page 27]

RFC 8489                          STUN                     February 2020


   The 16-byte key is formed by taking the MD5 hash of the result of
   concatenating the following five fields: (1) the username, with any
   quotes and trailing nulls removed, as taken from the USERNAME
   attribute (in which case OpaqueString has already been applied); (2)
   a single colon; (3) the realm, with any quotes and trailing nulls
   removed and after processing using OpaqueString; (4) a single colon;
   and (5) the password, with any trailing nulls removed and after
   processing using OpaqueString.  For example, if the username is
   'user', the realm is 'realm', and the password is 'pass', then the
   16-byte HMAC key would be the result of performing an MD5 hash on the
   string 'user:realm:pass', the resulting hash being
   0x8493fbc53ba582fb4c044c456bdc40eb.

   The structure of the key when used with long-term credentials
   facilitates deployment in systems that also utilize SIP [RFC3261].
   Typically, SIP systems utilizing SIP's digest authentication
   mechanism do not actually store the password in the database.
   Rather, they store a value called "H(A1)", which is equal to the key
   defined above.  For example, this mechanism can be used with the
   authentication extensions defined in [RFC5090].

   When a PASSWORD-ALGORITHM is used, the key length and algorithm to
   use are described in Section 18.5.1.

9.2.3.  Forming a Request



   The first request from the client to the server (as identified by
   hostname if the DNS procedures of Section 8 are used and by IP
   address if not) is handled according to the rules in Section 9.2.3.1.
   When the client initiates a subsequent request once a previous
   request/response transaction has completed successfully, it follows
   the rules in Section 9.2.3.2.  Forming a request as a consequence of
   a 401 (Unauthenticated) or 438 (Stale Nonce) error response is
   covered in Section 9.2.5 and is not considered a "subsequent request"
   and thus does not utilize the rules described in Section 9.2.3.2.
   Each of these types of requests have a different mandatory
   attributes.

9.2.3.1.  First Request



   If the client has not completed a successful request/response
   transaction with the server, it MUST omit the USERNAME, USERHASH,
   MESSAGE-INTEGRITY, MESSAGE-INTEGRITY-SHA256, REALM, NONCE, PASSWORD-
   ALGORITHMS, and PASSWORD-ALGORITHM attributes.  In other words, the
   first request is sent as if there were no authentication or message
   integrity applied.





Petit-Huguenin, et al.       Standards Track                   [Page 28]

RFC 8489                          STUN                     February 2020


9.2.3.2.  Subsequent Requests



   Once a request/response transaction has completed, the client will
   have been presented a realm and nonce by the server and selected a
   username and password with which it authenticated.  The client SHOULD
   cache the username, password, realm, and nonce for subsequent
   communications with the server.  When the client sends a subsequent
   request, it MUST include either the USERNAME or USERHASH, REALM,
   NONCE, and PASSWORD-ALGORITHM attributes with these cached values.
   It MUST include a MESSAGE-INTEGRITY attribute or a MESSAGE-INTEGRITY-
   SHA256 attribute, computed as described in Sections 14.5 and 14.6
   using the cached password.  The choice between the two attributes
   depends on the attribute received in the response to the first
   request.

9.2.4.  Receiving a Request



   After the server has done the basic processing of a request, it
   performs the checks listed below in the order specified.  Note that
   it is RECOMMENDED that the REALM value be the domain name of the
   provider of the STUN server:

   o  If the message does not contain a MESSAGE-INTEGRITY or MESSAGE-
      INTEGRITY-SHA256 attribute, the server MUST generate an error
      response with an error code of 401 (Unauthenticated).  This
      response MUST include a REALM value.  The response MUST include a
      NONCE, selected by the server.  The server MUST NOT choose the
      same NONCE for two requests unless they have the same source IP
      address and port.  The server MAY support alternate password
      algorithms, in which case it can list them in preferential order
      in a PASSWORD-ALGORITHMS attribute.  If the server adds a
      PASSWORD-ALGORITHMS attribute, it MUST set the STUN Security
      Feature "Password algorithms" bit to 1.  The server MAY support
      anonymous username, in which case it MUST set the STUN Security
      Feature "Username anonymity" bit set to 1.  The response SHOULD
      NOT
contain a USERNAME, USERHASH, MESSAGE-INTEGRITY, or MESSAGE-
      INTEGRITY-SHA256 attribute.

      Note: Reusing a NONCE for different source IP addresses or ports
      was not explicitly forbidden in [RFC5389].

   o  If the message contains a MESSAGE-INTEGRITY or a MESSAGE-
      INTEGRITY-SHA256 attribute, but is missing either the USERNAME or
      USERHASH, REALM, or NONCE attribute, the server MUST generate an
      error response with an error code of 400 (Bad Request).  This
      response SHOULD NOT include a USERNAME, USERHASH, NONCE, or REALM





Petit-Huguenin, et al.       Standards Track                   [Page 29]

RFC 8489                          STUN                     February 2020


      attribute.  The response cannot contain a MESSAGE-INTEGRITY or
      MESSAGE-INTEGRITY-SHA256 attribute, as the attributes required to
      generate them are missing.

   o  If the NONCE attribute starts with the "nonce cookie" with the
      STUN Security Feature "Password algorithms" bit set to 1, the
      server performs these checks in the order specified:

      *  If the request contains neither the PASSWORD-ALGORITHMS nor the
         PASSWORD-ALGORITHM algorithm, then the request is processed as
         though PASSWORD-ALGORITHM were MD5.

      *  Otherwise, unless (1) PASSWORD-ALGORITHM and PASSWORD-
         ALGORITHMS are both present, (2) PASSWORD-ALGORITHMS matches
         the value sent in the response that sent this NONCE, and (3)
         PASSWORD-ALGORITHM matches one of the entries in PASSWORD-
         ALGORITHMS, the server MUST generate an error response with an
         error code of 400 (Bad Request).

   o  If the value of the USERNAME or USERHASH attribute is not valid,
      the server MUST generate an error response with an error code of
      401 (Unauthenticated).  This response MUST include a REALM value.
      The response MUST include a NONCE, selected by the server.  The
      response MUST include a PASSWORD-ALGORITHMS attribute.  The
      response SHOULD NOT contain a USERNAME or USERHASH attribute.  The
      response MAY include a MESSAGE-INTEGRITY or MESSAGE-INTEGRITY-
      SHA256 attribute, using the previous key to calculate it.

   o  If the MESSAGE-INTEGRITY-SHA256 attribute is present, compute the
      value for the message integrity as described in Section 14.6,
      using the password associated with the username.  Otherwise, using
      the same password, compute the value for the MESSAGE-INTEGRITY
      attribute as described in Section 14.5.  If the resulting value
      does not match the contents of the MESSAGE-INTEGRITY attribute or
      the MESSAGE-INTEGRITY-SHA256 attribute, the server MUST reject the
      request with an error response.  This response MUST use an error
      code of 401 (Unauthenticated).  It MUST include the REALM and
      NONCE attributes and SHOULD NOT include the USERNAME, USERHASH,
      MESSAGE-INTEGRITY, or MESSAGE-INTEGRITY-SHA256 attribute.

   o  If the NONCE is no longer valid, the server MUST generate an error
      response with an error code of 438 (Stale Nonce).  This response
      MUST include NONCE, REALM, and PASSWORD-ALGORITHMS attributes and
      SHOULD NOT include the USERNAME and USERHASH attributes.  The
      NONCE attribute value MUST be valid.  The response MAY include a
      MESSAGE-INTEGRITY or MESSAGE-INTEGRITY-SHA256 attribute, using the





Petit-Huguenin, et al.       Standards Track                   [Page 30]

RFC 8489                          STUN                     February 2020


      previous NONCE to calculate it.  Servers can revoke nonces in
      order to provide additional security.  See Section 5.4 of
      [RFC7616] for guidelines.

   If these checks pass, the server continues to process the request.
   Any response generated by the server MUST include the MESSAGE-
   INTEGRITY-SHA256 attribute, computed using the username and password
   utilized to authenticate the request, unless the request was
   processed as though PASSWORD-ALGORITHM was MD5 (because the request
   contained neither PASSWORD-ALGORITHMS nor PASSWORD-ALGORITHM).  In
   that case, the MESSAGE-INTEGRITY attribute MUST be used instead of
   the MESSAGE-INTEGRITY-SHA256 attribute, and the REALM, NONCE,
   USERNAME, and USERHASH attributes SHOULD NOT be included.

9.2.5.  Receiving a Response



   If the response is an error response with an error code of 401
   (Unauthenticated) or 438 (Stale Nonce), the client MUST test if the
   NONCE attribute value starts with the "nonce cookie".  If so and the
   "nonce cookie" has the STUN Security Feature "Password algorithms"
   bit set to 1 but no PASSWORD-ALGORITHMS attribute is present, then
   the client MUST NOT retry the request with a new transaction.

   If the response is an error response with an error code of 401
   (Unauthenticated), the client SHOULD retry the request with a new
   transaction.  This request MUST contain a USERNAME or a USERHASH,
   determined by the client as the appropriate username for the REALM
   from the error response.  If the "nonce cookie" is present and has
   the STUN Security Feature "Username anonymity" bit set to 1, then the
   USERHASH attribute MUST be used; else, the USERNAME attribute MUST be
   used.  The request MUST contain the REALM, copied from the error
   response.  The request MUST contain the NONCE, copied from the error
   response.  If the response contains a PASSWORD-ALGORITHMS attribute,
   the request MUST contain the PASSWORD-ALGORITHMS attribute with the
   same content.  If the response contains a PASSWORD-ALGORITHMS
   attribute, and this attribute contains at least one algorithm that is
   supported by the client, then the request MUST contain a PASSWORD-
   ALGORITHM attribute with the first algorithm supported on the list.
   If the response contains a PASSWORD-ALGORITHMS attribute, and this
   attribute does not contain any algorithm that is supported by the
   client, then the client MUST NOT retry the request with a new
   transaction.  The client MUST NOT perform this retry if it is not
   changing the USERNAME, USERHASH, REALM, or its associated password
   from the previous attempt.







Petit-Huguenin, et al.       Standards Track                   [Page 31]

RFC 8489                          STUN                     February 2020


   If the response is an error response with an error code of 438 (Stale
   Nonce), the client MUST retry the request, using the new NONCE
   attribute supplied in the 438 (Stale Nonce) response.  This retry
   MUST also include either the USERNAME or USERHASH, the REALM, and
   either the MESSAGE-INTEGRITY or MESSAGE-INTEGRITY-SHA256 attribute.

   For all other responses, if the NONCE attribute starts with the
   "nonce cookie" with the STUN Security Feature "Password algorithms"
   bit set to 1 but PASSWORD-ALGORITHMS is not present, the response
   MUST be ignored.

   If the response is an error response with an error code of 400 (Bad
   Request) and does not contain either the MESSAGE-INTEGRITY or
   MESSAGE-INTEGRITY-SHA256 attribute, then the response MUST be
   discarded, as if it were never received.  This means that
   retransmits, if applicable, will continue.

      Note: In this case, the 400 response will never reach the
      application, resulting in a timeout.

   The client looks for the MESSAGE-INTEGRITY or MESSAGE-INTEGRITY-
   SHA256 attribute in the response (either success or failure).  If
   present, the client computes the message integrity over the response
   as defined in Sections 14.5 or 14.6, using the same password it
   utilized for the request.  If the resulting value matches the
   contents of the MESSAGE-INTEGRITY or MESSAGE-INTEGRITY-SHA256
   attribute, the response is considered authenticated.  If the value
   does not match, or if both MESSAGE-INTEGRITY and MESSAGE-INTEGRITY-
   SHA256 are absent, the processing depends on the request being sent
   over a reliable or an unreliable transport.

   If the request was sent over an unreliable transport, the response
   MUST be discarded, as if it had never been received.  This means that
   retransmits, if applicable, will continue.  If all the responses
   received are discarded, then instead of signaling a timeout after
   ending the transaction, the layer MUST signal that the integrity
   protection was violated.

   If the request was sent over a reliable transport, the response MUST
   be discarded, and the layer MUST immediately end the transaction and
   signal that the integrity protection was violated.

   If the response contains a PASSWORD-ALGORITHMS attribute, all the
   subsequent requests MUST be authenticated using MESSAGE-INTEGRITY-
   SHA256 only.






Petit-Huguenin, et al.       Standards Track                   [Page 32]

RFC 8489                          STUN                     February 2020


10.  ALTERNATE-SERVER Mechanism



   This section describes a mechanism in STUN that allows a server to
   redirect a client to another server.  This extension is optional, and
   a usage must define if and when this extension is used.  The
   ALTERNATE-SERVER attribute carries an IP address.

   A server using this extension redirects a client to another server by
   replying to a request message with an error response message with an
   error code of 300 (Try Alternate).  The server MUST include at least
   one ALTERNATE-SERVER attribute in the error response, which MUST
   contain an IP address of the same address family as the source IP
   address of the request message.  The server SHOULD include an
   additional ALTERNATE-SERVER attribute, after the mandatory one, that
   contains an IP address of the address family other than the source IP
   address of the request message.  The error response message MAY be
   authenticated; however, there are use cases for ALTERNATE-SERVER
   where authentication of the response is not possible or practical.
   If the transaction uses TLS or DTLS, if the transaction is
   authenticated by a MESSAGE-INTEGRITY-SHA256 attribute, and if the
   server wants to redirect to a server that uses a different
   certificate, then it MUST include an ALTERNATE-DOMAIN attribute
   containing the name inside the subjectAltName of that certificate.
   This series of conditions on the MESSAGE-INTEGRITY-SHA256 attribute
   indicates that the transaction is authenticated and that the client
   implements this specification and therefore can process the
   ALTERNATE-DOMAIN attribute.

   A client using this extension handles a 300 (Try Alternate) error
   code as follows.  The client looks for an ALTERNATE-SERVER attribute
   in the error response.  If one is found, then the client considers
   the current transaction as failed and reattempts the request with the
   server specified in the attribute, using the same transport protocol
   used for the previous request.  That request, if authenticated, MUST
   utilize the same credentials that the client would have used in the
   request to the server that performed the redirection.  If the
   transport protocol uses TLS or DTLS, then the client looks for an
   ALTERNATE-DOMAIN attribute.  If the attribute is found, the domain
   MUST be used to validate the certificate using the recommendations in
   [RFC6125].  The certificate MUST contain an identifier of type DNS-ID
   or CN-ID (eventually with wildcards) but not of type SRV-ID or URI-
   ID.  If the attribute is not found, the same domain that was used for
   the original request MUST be used to validate the certificate.  If
   the client has been redirected to a server to which it has already
   sent this request within the last five minutes, it MUST ignore the
   redirection and consider the transaction to have failed.  This
   prevents infinite ping-ponging between servers in case of redirection
   loops.



Petit-Huguenin, et al.       Standards Track                   [Page 33]

RFC 8489                          STUN                     February 2020


11.  Backwards Compatibility with RFC 3489



   In addition to the backward compatibility already described in
   Section 12 of [RFC5389], DTLS MUST NOT be used with [RFC3489]
   (referred to as "classic STUN").  Any STUN request or indication
   without the magic cookie (see Section 6 of [RFC5389]) over DTLS MUST
   be considered invalid: all requests MUST generate a 500 (Server
   Error) error response, and indications MUST be ignored.

12.  Basic Server Behavior



   This section defines the behavior of a basic, stand-alone STUN
   server.

   Historically, "classic STUN" [RFC3489] only defined the behavior of a
   server that was providing clients with server reflexive transport
   addresses by receiving and replying to STUN Binding requests.
   [RFC5389] redefined the protocol as an extensible framework, and the
   server functionality became the sole STUN Usage defined in that
   document.  This STUN Usage is also known as "Basic STUN Server".

   The STUN server MUST support the Binding method.  It SHOULD NOT
   utilize the short-term or long-term credential mechanism.  This is
   because the work involved in authenticating the request is more than
   the work in simply processing it.  It SHOULD NOT utilize the
   ALTERNATE-SERVER mechanism for the same reason.  It MUST support UDP
   and TCP.  It MAY support STUN over TCP/TLS or STUN over UDP/DTLS;
   however, DTLS and TLS provide minimal security benefits in this basic
   mode of operation.  It does not require a keep-alive mechanism
   because a TCP or TLS-over-TCP connection is closed after the end of
   the Binding transaction.  It MAY utilize the FINGERPRINT mechanism
   but MUST NOT require it.  Since the stand-alone server only runs
   STUN, FINGERPRINT provides no benefit.  Requiring it would break
   compatibility with RFC 3489, and such compatibility is desirable in a
   stand-alone server.  Stand-alone STUN servers SHOULD support
   backwards compatibility with clients using [RFC3489], as described in
   Section 11.

   It is RECOMMENDED that administrators of STUN servers provide DNS
   entries for those servers as described in Section 8.  If both A and
   AAAA resource records are returned, then the client can
   simultaneously send STUN Binding requests to the IPv4 and IPv6
   addresses (as specified in [RFC8305]), as the Binding request is
   idempotent.  Note that the MAPPED-ADDRESS or XOR-MAPPED-ADDRESS
   attributes that are returned will not necessarily match the address
   family of the server address used.





Petit-Huguenin, et al.       Standards Track                   [Page 34]

RFC 8489                          STUN                     February 2020


   A basic STUN server is not a solution for NAT traversal by itself.
   However, it can be utilized as part of a solution through STUN
   Usages.  This is discussed further in Section 13.

13.  STUN Usages



   STUN by itself is not a solution to the NAT traversal problem.
   Rather, STUN defines a tool that can be used inside a larger
   solution.  The term "STUN Usage" is used for any solution that uses
   STUN as a component.

   A STUN Usage defines how STUN is actually utilized -- when to send
   requests, what to do with the responses, and which optional
   procedures defined here (or in an extension to STUN) are to be used.
   A usage also defines:

   o  Which STUN methods are used.

   o  What transports are used.  If DTLS-over-UDP is used, then
      implementing the denial-of-service countermeasure described in
      Section 4.2.1 of [RFC6347] is mandatory.

   o  What authentication and message-integrity mechanisms are used.

   o  The considerations around manual vs. automatic key derivation for
      the integrity mechanism, as discussed in [RFC4107].

   o  What mechanisms are used to distinguish STUN messages from other
      messages.  When STUN is run over TCP or TLS-over-TCP, a framing
      mechanism may be required.

   o  How a STUN client determines the IP address and port of the STUN
      server.

   o  How simultaneous use of IPv4 and IPv6 addresses (Happy Eyeballs
      [RFC8305]) works with non-idempotent transactions when both
      address families are found for the STUN server.

   o  Whether backwards compatibility to RFC 3489 is required.

   o  What optional attributes defined here (such as FINGERPRINT and
      ALTERNATE-SERVER) or in other extensions are required.

   o  If MESSAGE-INTEGRITY-SHA256 truncation is permitted, and the
      limits permitted for truncation.

   o  The keep-alive mechanism if STUN is run over TCP or TLS-over-TCP.




Petit-Huguenin, et al.       Standards Track                   [Page 35]

RFC 8489                          STUN                     February 2020


   o  If anycast addresses can be used for the server in case 1) TCP or
      TLS-over-TCP or 2) authentication is used.

   In addition, any STUN Usage must consider the security implications
   of using STUN in that usage.  A number of attacks against STUN are
   known (see the Security Considerations section in this document), and
   any usage must consider how these attacks can be thwarted or
   mitigated.

   Finally, a usage must consider whether its usage of STUN is an
   example of the Unilateral Self-Address Fixing approach to NAT
   traversal and, if so, address the questions raised in RFC 3424
   [RFC3424].

14.  STUN Attributes



   After the STUN header are zero or more attributes.  Each attribute
   MUST be TLV encoded, with a 16-bit type, 16-bit length, and value.
   Each STUN attribute MUST end on a 32-bit boundary.  As mentioned
   above, all fields in an attribute are transmitted most significant
   bit first.

      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
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |         Type                  |            Length             |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |                         Value (variable)                ....
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

                    Figure 4: Format of STUN Attributes

   The value in the Length field MUST contain the length of the Value
   part of the attribute, prior to padding, measured in bytes.  Since
   STUN aligns attributes on 32-bit boundaries, attributes whose content
   is not a multiple of 4 bytes are padded with 1, 2, or 3 bytes of
   padding so that its value contains a multiple of 4 bytes.  The
   padding bits MUST be set to zero on sending and MUST be ignored by
   the receiver.

   Any attribute type MAY appear more than once in a STUN message.
   Unless specified otherwise, the order of appearance is significant:
   only the first occurrence needs to be processed by a receiver, and
   any duplicates MAY be ignored by a receiver.

   To allow future revisions of this specification to add new attributes
   if needed, the attribute space is divided into two ranges.
   Attributes with type values between 0x0000 and 0x7FFF are



Petit-Huguenin, et al.       Standards Track                   [Page 36]

RFC 8489                          STUN                     February 2020


   comprehension-required attributes, which means that the STUN agent
   cannot successfully process the message unless it understands the
   attribute.  Attributes with type values between 0x8000 and 0xFFFF are
   comprehension-optional attributes, which means that those attributes
   can be ignored by the STUN agent if it does not understand them.

   The set of STUN attribute types is maintained by IANA.  The initial
   set defined by this specification is found in Section 18.3.

   The rest of this section describes the format of the various
   attributes defined in this specification.

14.1.  MAPPED-ADDRESS



   The MAPPED-ADDRESS attribute indicates a reflexive transport address
   of the client.  It consists of an 8-bit address family and a 16-bit
   port, followed by a fixed-length value representing the IP address.
   If the address family is IPv4, the address MUST be 32 bits.  If the
   address family is IPv6, the address MUST be 128 bits.  All fields
   must be in network byte order.

   The format of the MAPPED-ADDRESS attribute is:

      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
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |0 0 0 0 0 0 0 0|    Family     |           Port                |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |                                                               |
     |                 Address (32 bits or 128 bits)                 |
     |                                                               |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

               Figure 5: Format of MAPPED-ADDRESS Attribute

   The address family can take on the following values:

   0x01:IPv4
   0x02:IPv6

   The first 8 bits of the MAPPED-ADDRESS MUST be set to 0 and MUST be
   ignored by receivers.  These bits are present for aligning parameters
   on natural 32-bit boundaries.

   This attribute is used only by servers for achieving backwards
   compatibility with [RFC3489] clients.





Petit-Huguenin, et al.       Standards Track                   [Page 37]

RFC 8489                          STUN                     February 2020


14.2.  XOR-MAPPED-ADDRESS



   The XOR-MAPPED-ADDRESS attribute is identical to the MAPPED-ADDRESS
   attribute, except that the reflexive transport address is obfuscated
   through the XOR function.

   The format of the XOR-MAPPED-ADDRESS is:

      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
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |0 0 0 0 0 0 0 0|    Family     |         X-Port                |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |                X-Address (Variable)
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

             Figure 6: Format of XOR-MAPPED-ADDRESS Attribute

   The Family field represents the IP address family and is encoded
   identically to the Family field in MAPPED-ADDRESS.

   X-Port is computed by XOR'ing the mapped port with the most
   significant 16 bits of the magic cookie.  If the IP address family is
   IPv4, X-Address is computed by XOR'ing the mapped IP address with the
   magic cookie.  If the IP address family is IPv6, X-Address is
   computed by XOR'ing the mapped IP address with the concatenation of
   the magic cookie and the 96-bit transaction ID.  In all cases, the
   XOR operation works on its inputs in network byte order (that is, the
   order they will be encoded in the message).

   The rules for encoding and processing the first 8 bits of the
   attribute's value, the rules for handling multiple occurrences of the
   attribute, and the rules for processing address families are the same
   as for MAPPED-ADDRESS.

   Note: XOR-MAPPED-ADDRESS and MAPPED-ADDRESS differ only in their
   encoding of the transport address.  The former encodes the transport
   address by XOR'ing it with the magic cookie.  The latter encodes it
   directly in binary.  RFC 3489 originally specified only MAPPED-
   ADDRESS.  However, deployment experience found that some NATs rewrite
   the 32-bit binary payloads containing the NAT's public IP address,
   such as STUN's MAPPED-ADDRESS attribute, in the well-meaning but
   misguided attempt to provide a generic Application Layer Gateway
   (ALG) function.  Such behavior interferes with the operation of STUN
   and also causes failure of STUN's message-integrity checking.






Petit-Huguenin, et al.       Standards Track                   [Page 38]

RFC 8489                          STUN                     February 2020


14.3.  USERNAME



   The USERNAME attribute is used for message integrity.  It identifies
   the username and password combination used in the message-integrity
   check.

   The value of USERNAME is a variable-length value containing the
   authentication username.  It MUST contain a UTF-8-encoded [RFC3629]
   sequence of fewer than 509 bytes and MUST have been processed using
   the OpaqueString profile [RFC8265].  A compliant implementation MUST
   be able to parse a UTF-8-encoded sequence of 763 or fewer octets to
   be compatible with [RFC5389].

      Note: [RFC5389] mistakenly referenced the definition of UTF-8 in
      [RFC2279].  [RFC2279] assumed up to 6 octets per characters
      encoded.  [RFC2279] was replaced by [RFC3629], which allows only 4
      octets per character encoded, consistent with changes made in
      Unicode 2.0 and ISO/IEC 10646.

      Note: This specification uses the OpaqueString profile instead of
      the UsernameCasePreserved profile for username string processing
      in order to improve compatibility with deployed password stores.
      Many password databases used for HTTP and SIP Digest
      authentication store the MD5 hash of username:realm:password
      instead of storing a plain text password.  In [RFC3489], STUN
      authentication was designed to be compatible with these existing
      databases to the extent possible, which like SIP and HTTP
      performed no pre-processing of usernames and passwords other than
      prohibiting non-space ASCII control characters.  The next revision
      of the STUN specification, [RFC5389], used the SASLprep [RFC4013]
      stringprep [RFC3454] profile to pre-process usernames and
      passwords.  SASLprep uses Unicode Normalization Form KC
      (Compatibility Decomposition, followed by Canonical Composition)
      [UAX15] and prohibits various control, space, and non-text,
      deprecated, or inappropriate codepoints.  The PRECIS framework
      [RFC8264] obsoletes stringprep.  PRECIS handling of usernames and
      passwords [RFC8265] uses Unicode Normalization Form C (Canonical
      Decomposition, followed by Canonical Composition).  While there
      are specific cases where different username strings under HTTP
      Digest could be mapped to a single STUN username processed with
      OpaqueString, these cases are extremely unlikely and easy to
      detect and correct.  With a UsernameCasePreserved profile, it
      would be more likely that valid usernames under HTTP Digest would
      not match their processed forms (specifically usernames containing
      bidirectional text and compatibility forms).  Operators are free
      to further restrict the allowed codepoints in usernames to avoid
      problematic characters.




Petit-Huguenin, et al.       Standards Track                   [Page 39]

RFC 8489                          STUN                     February 2020


14.4.  USERHASH



   The USERHASH attribute is used as a replacement for the USERNAME
   attribute when username anonymity is supported.

   The value of USERHASH has a fixed length of 32 bytes.  The username
   MUST have been processed using the OpaqueString profile [RFC8265],
   and the realm MUST have been processed using the OpaqueString profile
   [RFC8265] before hashing.

   The following is the operation that the client will perform to hash
   the username:

   userhash = SHA-256(OpaqueString(username) ":" OpaqueString(realm))

14.5.  MESSAGE-INTEGRITY



   The MESSAGE-INTEGRITY attribute contains an HMAC-SHA1 [RFC2104] of
   the STUN message.  The MESSAGE-INTEGRITY attribute can be present in
   any STUN message type.  Since it uses the SHA-1 hash, the HMAC will
   be 20 bytes.

   The key for the HMAC depends on which credential mechanism is in use.
   Section 9.1.1 defines the key for the short-term credential
   mechanism, and Section 9.2.2 defines the key for the long-term
   credential mechanism.  Other credential mechanisms MUST define the
   key that is used for the HMAC.

   The text used as input to HMAC is the STUN message, up to and
   including the attribute preceding the MESSAGE-INTEGRITY attribute.
   The Length field of the STUN message header is adjusted to point to
   the end of the MESSAGE-INTEGRITY attribute.  The value of the
   MESSAGE-INTEGRITY attribute is set to a dummy value.

   Once the computation is performed, the value of the MESSAGE-INTEGRITY
   attribute is filled in, and the value of the length in the STUN
   header is set to its correct value -- the length of the entire
   message.  Similarly, when validating the MESSAGE-INTEGRITY, the
   Length field in the STUN header must be adjusted to point to the end
   of the MESSAGE-INTEGRITY attribute prior to calculating the HMAC over
   the STUN message, up to and including the attribute preceding the
   MESSAGE-INTEGRITY attribute.  Such adjustment is necessary when
   attributes, such as FINGERPRINT and MESSAGE-INTEGRITY-SHA256, appear
   after MESSAGE-INTEGRITY.  See also [RFC5769] for examples of such
   calculations.






Petit-Huguenin, et al.       Standards Track                   [Page 40]

RFC 8489                          STUN                     February 2020


14.6.  MESSAGE-INTEGRITY-SHA256



   The MESSAGE-INTEGRITY-SHA256 attribute contains an HMAC-SHA256
   [RFC2104] of the STUN message.  The MESSAGE-INTEGRITY-SHA256
   attribute can be present in any STUN message type.  The MESSAGE-
   INTEGRITY-SHA256 attribute contains an initial portion of the HMAC-
   SHA-256 [RFC2104] of the STUN message.  The value will be at most 32
   bytes, but it MUST be at least 16 bytes and MUST be a multiple of 4
   bytes.  The value must be the full 32 bytes unless the STUN Usage
   explicitly specifies that truncation is allowed.  STUN Usages may
   specify a minimum length longer than 16 bytes.

   The key for the HMAC depends on which credential mechanism is in use.
   Section 9.1.1 defines the key for the short-term credential
   mechanism, and Section 9.2.2 defines the key for the long-term
   credential mechanism.  Other credential mechanism MUST define the key
   that is used for the HMAC.

   The text used as input to HMAC is the STUN message, up to and
   including the attribute preceding the MESSAGE-INTEGRITY-SHA256
   attribute.  The Length field of the STUN message header is adjusted
   to point to the end of the MESSAGE-INTEGRITY-SHA256 attribute.  The
   value of the MESSAGE-INTEGRITY-SHA256 attribute is set to a dummy
   value.

   Once the computation is performed, the value of the MESSAGE-
   INTEGRITY-SHA256 attribute is filled in, and the value of the length
   in the STUN header is set to its correct value -- the length of the
   entire message.  Similarly, when validating the MESSAGE-INTEGRITY-
   SHA256, the Length field in the STUN header must be adjusted to point
   to the end of the MESSAGE-INTEGRITY-SHA256 attribute prior to
   calculating the HMAC over the STUN message, up to and including the
   attribute preceding the MESSAGE-INTEGRITY-SHA256 attribute.  Such
   adjustment is necessary when attributes, such as FINGERPRINT, appear
   after MESSAGE-INTEGRITY-SHA256.  See also Appendix B.1 for examples
   of such calculations.

14.7.  FINGERPRINT



   The FINGERPRINT attribute MAY be present in all STUN messages.

   The value of the attribute is computed as the CRC-32 of the STUN
   message up to (but excluding) the FINGERPRINT attribute itself,
   XOR'ed with the 32-bit value 0x5354554e.  (The XOR operation ensures
   that the FINGERPRINT test will not report a false positive on a
   packet containing a CRC-32 generated by an application protocol.)
   The 32-bit CRC is the one defined in ITU V.42 [ITU.V42.2002], which




Petit-Huguenin, et al.       Standards Track                   [Page 41]

RFC 8489                          STUN                     February 2020


   has a generator polynomial of x^32 + x^26 + x^23 + x^22 + x^16 + x^12
   + x^11 + x^10 + x^8 + x^7 + x^5 + x^4 + x^2 + x + 1.  See the sample
   code for the CRC-32 in Section 8 of [RFC1952].

   When present, the FINGERPRINT attribute MUST be the last attribute in
   the message and thus will appear after MESSAGE-INTEGRITY and MESSAGE-
   INTEGRITY-SHA256.

   The FINGERPRINT attribute can aid in distinguishing STUN packets from
   packets of other protocols.  See Section 7.

   As with MESSAGE-INTEGRITY and MESSAGE-INTEGRITY-SHA256, the CRC used
   in the FINGERPRINT attribute covers the Length field from the STUN
   message header.  Therefore, prior to computation of the CRC, this
   value must be correct and include the CRC attribute as part of the
   message length.  When using the FINGERPRINT attribute in a message,
   the attribute is first placed into the message with a dummy value;
   then, the CRC is computed, and the value of the attribute is updated.
   If the MESSAGE-INTEGRITY or MESSAGE-INTEGRITY-SHA256 attribute is
   also present, then it must be present with the correct message-
   integrity value before the CRC is computed, since the CRC is done
   over the value of the MESSAGE-INTEGRITY and MESSAGE-INTEGRITY-SHA256
   attributes as well.

14.8.  ERROR-CODE



   The ERROR-CODE attribute is used in error response messages.  It
   contains a numeric error code value in the range of 300 to 699 plus a
   textual reason phrase encoded in UTF-8 [RFC3629]; it is also
   consistent in its code assignments and semantics with SIP [RFC3261]
   and HTTP [RFC7231].  The reason phrase is meant for diagnostic
   purposes and can be anything appropriate for the error code.
   Recommended reason phrases for the defined error codes are included
   in the IANA registry for error codes.  The reason phrase MUST be a
   UTF-8-encoded [RFC3629] sequence of fewer than 128 characters (which
   can be as long as 509 bytes when encoding them or 763 bytes when
   decoding them).

      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, should be 0         |Class|     Number    |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |      Reason Phrase (variable)                                ..
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

                 Figure 7: Format of ERROR-CODE Attribute




Petit-Huguenin, et al.       Standards Track                   [Page 42]

RFC 8489                          STUN                     February 2020


   To facilitate processing, the class of the error code (the hundreds
   digit) is encoded separately from the rest of the code, as shown in
   Figure 7.

   The Reserved bits SHOULD be 0 and are for alignment on 32-bit
   boundaries.  Receivers MUST ignore these bits.  The Class represents
   the hundreds digit of the error code.  The value MUST be between 3
   and 6.  The Number represents the binary encoding of the error code
   modulo 100, and its value MUST be between 0 and 99.

   The following error codes, along with their recommended reason
   phrases, are defined:

   300  Try Alternate: The client should contact an alternate server for
        this request.  This error response MUST only be sent if the
        request included either a USERNAME or USERHASH attribute and a
        valid MESSAGE-INTEGRITY or MESSAGE-INTEGRITY-SHA256 attribute;
        otherwise, it MUST NOT be sent and error code 400 (Bad Request)
        is suggested.  This error response MUST be protected with the
        MESSAGE-INTEGRITY or MESSAGE-INTEGRITY-SHA256 attribute, and
        receivers MUST validate the MESSAGE-INTEGRITY or MESSAGE-
        INTEGRITY-SHA256 of this response before redirecting themselves
        to an alternate server.

        Note: Failure to generate and validate message integrity for a
        300 response allows an on-path attacker to falsify a 300
        response thus causing subsequent STUN messages to be sent to a
        victim.

   400  Bad Request: The request was malformed.  The client SHOULD NOT
        retry the request without modification from the previous
        attempt.  The server may not be able to generate a valid
        MESSAGE-INTEGRITY or MESSAGE-INTEGRITY-SHA256 for this error, so
        the client MUST NOT expect a valid MESSAGE-INTEGRITY or MESSAGE-
        INTEGRITY-SHA256 attribute on this response.

   401  Unauthenticated: The request did not contain the correct
        credentials to proceed.  The client should retry the request
        with proper credentials.



   420  Unknown Attribute: The server received a STUN packet containing
        a comprehension-required attribute that it did not understand.
        The server MUST put this unknown attribute in the UNKNOWN-
        ATTRIBUTE attribute of its error response.

   438  Stale Nonce: The NONCE used by the client was no longer valid.
        The client should retry, using the NONCE provided in the
        response.





Petit-Huguenin, et al.       Standards Track                   [Page 43]

RFC 8489                          STUN                     February 2020


   500  Server Error: The server has suffered a temporary error.  The
        client should try again.



14.9.  REALM



   The REALM attribute may be present in requests and responses.  It
   contains text that meets the grammar for "realm-value" as described
   in [RFC3261] but without the double quotes and their surrounding
   whitespace.  That is, it is an unquoted realm-value (and is therefore
   a sequence of qdtext or quoted-pair).  It MUST be a UTF-8-encoded
   [RFC3629] sequence of fewer than 128 characters (which can be as long
   as 509 bytes when encoding them and as long as 763 bytes when
   decoding them) and MUST have been processed using the OpaqueString
   profile [RFC8265].

   Presence of the REALM attribute in a request indicates that long-term
   credentials are being used for authentication.  Presence in certain
   error responses indicates that the server wishes the client to use a
   long-term credential in that realm for authentication.

14.10.  NONCE



   The NONCE attribute may be present in requests and responses.  It
   contains a sequence of qdtext or quoted-pair, which are defined in
   [RFC3261].  Note that this means that the NONCE attribute will not
   contain the actual surrounding quote characters.  The NONCE attribute
   MUST be fewer than 128 characters (which can be as long as 509 bytes
   when encoding them and a long as 763 bytes when decoding them).  See
   Section 5.4 of [RFC7616] for guidance on selection of nonce values in
   a server.

14.11.  PASSWORD-ALGORITHMS



   The PASSWORD-ALGORITHMS attribute may be present in requests and
   responses.  It contains the list of algorithms that the server can
   use to derive the long-term password.

   The set of known algorithms is maintained by IANA.  The initial set
   defined by this specification is found in Section 18.5.

   The attribute contains a list of algorithm numbers and variable
   length parameters.  The algorithm number is a 16-bit value as defined
   in Section 18.5.  The parameters start with the length (prior to
   padding) of the parameters as a 16-bit value, followed by the
   parameters that are specific to each algorithm.  The parameters are
   padded to a 32-bit boundary, in the same manner as an attribute.





Petit-Huguenin, et al.       Standards Track                   [Page 44]

RFC 8489                          STUN                     February 2020


      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
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |         Algorithm 1           | Algorithm 1 Parameters Length |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |                    Algorithm 1 Parameters (variable)
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |         Algorithm 2           | Algorithm 2 Parameters Length |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |                    Algorithm 2 Parameters (variable)
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |                                                             ...

             Figure 8: Format of PASSWORD-ALGORITHMS Attribute

14.12.  PASSWORD-ALGORITHM



   The PASSWORD-ALGORITHM attribute is present only in requests.  It
   contains the algorithm that the server must use to derive a key from
   the long-term password.

   The set of known algorithms is maintained by IANA.  The initial set
   defined by this specification is found in Section 18.5.

   The attribute contains an algorithm number and variable length
   parameters.  The algorithm number is a 16-bit value as defined in
   Section 18.5.  The parameters starts with the length (prior to
   padding) of the parameters as a 16-bit value, followed by the
   parameters that are specific to the algorithm.  The parameters are
   padded to a 32-bit boundary, in the same manner as an attribute.
   Similarly, the padding bits MUST be set to zero on sending and MUST
   be ignored by the receiver.

      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
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |          Algorithm           |  Algorithm Parameters Length   |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |                    Algorithm Parameters (variable)
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

             Figure 9: Format of PASSWORD-ALGORITHM Attribute

14.13.  UNKNOWN-ATTRIBUTES



   The UNKNOWN-ATTRIBUTES attribute is present only in an error response
   when the response code in the ERROR-CODE attribute is 420 (Unknown
   Attribute).



Petit-Huguenin, et al.       Standards Track                   [Page 45]

RFC 8489                          STUN                     February 2020


   The attribute contains a list of 16-bit values, each of which
   represents an attribute type that was not understood by the server.

      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
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |      Attribute 1 Type         |       Attribute 2 Type        |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |      Attribute 3 Type         |       Attribute 4 Type    ...
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

             Figure 10: Format of UNKNOWN-ATTRIBUTES Attribute

      Note: In [RFC3489], this field was padded to 32 by duplicating the
      last attribute.  In this version of the specification, the normal
      padding rules for attributes are used instead.

14.14.  SOFTWARE



   The SOFTWARE attribute contains a textual description of the software
   being used by the agent sending the message.  It is used by clients
   and servers.  Its value SHOULD include manufacturer and version
   number.  The attribute has no impact on operation of the protocol and
   serves only as a tool for diagnostic and debugging purposes.  The
   value of SOFTWARE is variable length.  It MUST be a UTF-8-encoded
   [RFC3629] sequence of fewer than 128 characters (which can be as long
   as 509 when encoding them and as long as 763 bytes when decoding
   them).

14.15.  ALTERNATE-SERVER



   The alternate server represents an alternate transport address
   identifying a different STUN server that the STUN client should try.

   It is encoded in the same way as MAPPED-ADDRESS and thus refers to a
   single server by IP address.

14.16.  ALTERNATE-DOMAIN



   The alternate domain represents the domain name that is used to
   verify the IP address in the ALTERNATE-SERVER attribute when the
   transport protocol uses TLS or DTLS.

   The value of ALTERNATE-DOMAIN is variable length.  It MUST be a valid
   DNS name [RFC1123] (including A-labels [RFC5890]) of 255 or fewer
   ASCII characters.





Petit-Huguenin, et al.       Standards Track                   [Page 46]

RFC 8489                          STUN                     February 2020


15.  Operational Considerations



   STUN MAY be used with anycast addresses, but only with UDP and in
   STUN Usages where authentication is not used.

16.  Security Considerations



   Implementations and deployments of a STUN Usage using TLS or DTLS
   MUST follow the recommendations in [BCP195].

   Implementations and deployments of a STUN Usage using the long-term
   credential mechanism (Section 9.2) MUST follow the recommendations in
   Section 5 of [RFC7616].

16.1.  Attacks against the Protocol



16.1.1.  Outside Attacks



   An attacker can try to modify STUN messages in transit, in order to
   cause a failure in STUN operation.  These attacks are detected for
   both requests and responses through the message-integrity mechanism,
   using either a short-term or long-term credential.  Of course, once
   detected, the manipulated packets will be dropped, causing the STUN
   transaction to effectively fail.  This attack is possible only by an
   on-path attacker.

   An attacker that can observe, but not modify, STUN messages in-
   transit (for example, an attacker present on a shared access medium,
   such as Wi-Fi) can see a STUN request and then immediately send a
   STUN response, typically an error response, in order to disrupt STUN
   processing.  This attack is also prevented for messages that utilize
   MESSAGE-INTEGRITY.  However, some error responses, those related to
   authentication in particular, cannot be protected by MESSAGE-
   INTEGRITY.  When STUN itself is run over a secure transport protocol
   (e.g., TLS), these attacks are completely mitigated.

   Depending on the STUN Usage, these attacks may be of minimal
   consequence and thus do not require message integrity to mitigate.
   For example, when STUN is used to a basic STUN server to discover a
   server reflexive candidate for usage with ICE, authentication and
   message integrity are not required since these attacks are detected
   during the connectivity check phase.  The connectivity checks
   themselves, however, require protection for proper operation of ICE
   overall.  As described in Section 13, STUN Usages describe when
   authentication and message integrity are needed.






Petit-Huguenin, et al.       Standards Track                   [Page 47]

RFC 8489                          STUN                     February 2020


   Since STUN uses the HMAC of a shared secret for authentication and
   integrity protection, it is subject to offline dictionary attacks.
   When authentication is utilized, it SHOULD be with a strong password
   that is not readily subject to offline dictionary attacks.
   Protection of the channel itself, using TLS or DTLS, mitigates these
   attacks.

   STUN supports both MESSAGE-INTEGRITY and MESSAGE-INTEGRITY-SHA256,
   which makes STUN subject to bid-down attacks by an on-path attacker.
   An attacker could strip the MESSAGE-INTEGRITY-SHA256 attribute,
   leaving only the MESSAGE-INTEGRITY attribute and thus exploiting a
   potential vulnerability.  Protection of the channel itself, using TLS
   or DTLS, mitigates these attacks.  Timely removal of the support of
   MESSAGE-INTEGRITY in a future version of STUN is necessary.

   Note: The use of SHA-256 for password hashing does not meet modern
   standards, which are aimed at slowing down exhaustive password
   searches by providing a relatively slow minimum time to compute the
   hash.  Although better algorithms such as Argon2 [Argon2] are
   available, SHA-256 was chosen for consistency with [RFC7616].

16.1.2.  Inside Attacks



   A rogue client may try to launch a DoS attack against a server by
   sending it a large number of STUN requests.  Fortunately, STUN
   requests can be processed statelessly by a server, making such
   attacks hard to launch effectively.

   A rogue client may use a STUN server as a reflector, sending it
   requests with a falsified source IP address and port.  In such a
   case, the response would be delivered to that source IP and port.
   There is no amplification of the number of packets with this attack
   (the STUN server sends one packet for each packet sent by the
   client), though there is a small increase in the amount of data,
   since STUN responses are typically larger than requests.  This attack
   is mitigated by ingress source address filtering.

   Revealing the specific software version of the agent through the
   SOFTWARE attribute might allow them to become more vulnerable to
   attacks against software that is known to contain security holes.
   Implementers SHOULD make usage of the SOFTWARE attribute a
   configurable option.

16.1.3.  Bid-Down Attacks



   This document adds the possibility of selecting different algorithms
   to protect the confidentiality of the passwords stored on the server
   side when using the long-term credential mechanism while still



Petit-Huguenin, et al.       Standards Track                   [Page 48]

RFC 8489                          STUN                     February 2020


   ensuring compatibility with MD5, which was the algorithm used in
   [RFC5389].  This selection works by having the server send to the
   client the list of algorithms supported in a PASSWORD-ALGORITHMS
   attribute and having the client send back a PASSWORD-ALGORITHM
   attribute containing the algorithm selected.

   Because the PASSWORD-ALGORITHMS attribute has to be sent in an
   unauthenticated response, an on-path attacker wanting to exploit an
   eventual vulnerability in MD5 can just strip the PASSWORD-ALGORITHMS
   attribute from the unprotected response, thus making the server
   subsequently act as if the client was implementing the version of
   this protocol defined in [RFC5389].

   To protect against this attack and other similar bid-down attacks,
   the nonce is enriched with a set of security bits that indicates
   which security features are in use.  In the case of the selection of
   the password algorithm, the matching bit is set in the nonce returned
   by the server in the same response that contains the PASSWORD-
   ALGORITHMS attribute.  Because the nonce used in subsequent
   authenticated transactions is verified by the server to be identical
   to what was originally sent, it cannot be modified by an on-path
   attacker.  Additionally, the client is mandated to copy the received
   PASSWORD-ALGORITHMS attribute in the next authenticated transaction
   to that server.

   An on-path attack that removes the PASSWORD-ALGORITHMS will be
   detected because the client will not be able to send it back to the
   server in the next authenticated transaction.  The client will detect
   that attack because the security bit is set but the matching
   attribute is missing; this will end the session.  A client using an
   older version of this protocol will not send the PASSWORD-ALGORITHMS
   back but can only use MD5 anyway, so the attack is inconsequential.

   The on-path attack may also try to remove the security bit together
   with the PASSWORD-ALGORITHMS attribute, but the server will discover
   that when the next authenticated transaction contains an invalid
   nonce.

   An on-path attack that removes some algorithms from the PASSWORD-
   ALGORITHMS attribute will be equally defeated because that attribute
   will be different from the original one when the server verifies it
   in the subsequent authenticated transaction.

   Note that the bid-down protection mechanism introduced in this
   document is inherently limited by the fact that it is not possible to
   detect an attack until the server receives the second request after
   the 401 (Unauthenticated) response.




Petit-Huguenin, et al.       Standards Track                   [Page 49]

RFC 8489                          STUN                     February 2020


   SHA-256 was chosen as the new default for password hashing for its
   compatibility with [RFC7616], but because SHA-256 (like MD5) is a
   comparatively fast algorithm, it does little to deter brute-force
   attacks.  Specifically, this means that if the user has a weak
   password, an attacker that captures a single exchange can use a
   brute-force attack to learn the user's password and then potentially
   impersonate the user to the server and to other servers where the
   same password was used.  Note that such an attacker can impersonate
   the user to the server itself without any brute-force attack.

   A stronger (which is to say, slower) algorithm, like Argon2 [Argon2],
   would help both of these cases; however, in the first case, it would
   only help after the database entry for this user is updated to
   exclusively use that stronger mechanism.

   The bid-down defenses in this protocol prevent an attacker from
   forcing the client and server to complete a handshake using weaker
   algorithms than they jointly support, but only if the weakest joint
   algorithm is strong enough that it cannot be compromised by a brute-
   force attack.  However, this does not defend against many attacks on
   those algorithms; specifically, an on-path attacker might perform a
   bid-down attack on a client that supports both Argon2 [Argon2] and
   SHA-256 for password hashing and use that to collect a MESSAGE-
   INTEGRITY-SHA256 value that it can then use for an offline brute-
   force attack.  This would be detected when the server receives the
   second request, but that does not prevent the attacker from obtaining
   the MESSAGE-INTEGRITY-SHA256 value.

   Similarly, an attack against the USERHASH mechanism will not succeed
   in establishing a session as the server will detect that the feature
   was discarded on path, but the client would still have been convinced
   to send its username in the clear in the USERNAME attribute, thus
   disclosing it to the attacker.

   Finally, when the bid-down protection mechanism is employed for a
   future upgrade of the HMAC algorithm used to protect messages, it
   will offer only a limited protection if the current HMAC algorithm is
   already compromised.

16.2.  Attacks Affecting the Usage



   This section lists attacks that might be launched against a usage of
   STUN.  Each STUN Usage must consider whether these attacks are
   applicable to it and, if so, discuss countermeasures.

   Most of the attacks in this section revolve around an attacker
   modifying the reflexive address learned by a STUN client through a
   Binding request/response transaction.  Since the usage of the



Petit-Huguenin, et al.       Standards Track                   [Page 50]

RFC 8489                          STUN                     February 2020


   reflexive address is a function of the usage, the applicability and
   remediation of these attacks are usage-specific.  In common
   situations, modification of the reflexive address by an on-path
   attacker is easy to do.  Consider, for example, the common situation
   where STUN is run directly over UDP.  In this case, an on-path
   attacker can modify the source IP address of the Binding request
   before it arrives at the STUN server.  The STUN server will then
   return this IP address in the XOR-MAPPED-ADDRESS attribute to the
   client and send the response back to that (falsified) IP address and
   port.  If the attacker can also intercept this response, it can
   direct it back towards the client.  Protecting against this attack by
   using a message-integrity check is impossible, since a message-
   integrity value cannot cover the source IP address and the
   intervening NAT must be able to modify this value.  Instead, one
   solution to prevent the attacks listed below is for the client to
   verify the reflexive address learned, as is done in ICE [RFC8445].

   Other usages may use other means to prevent these attacks.

16.2.1.  Attack I: Distributed DoS (DDoS) against a Target



   In this attack, the attacker provides one or more clients with the
   same faked reflexive address that points to the intended target.
   This will trick the STUN clients into thinking that their reflexive
   addresses are equal to that of the target.  If the clients hand out
   that reflexive address in order to receive traffic on it (for
   example, in SIP messages), the traffic will instead be sent to the
   target.  This attack can provide substantial amplification,
   especially when used with clients that are using STUN to enable
   multimedia applications.  However, it can only be launched against
   targets for which packets from the STUN server to the target pass
   through the attacker, limiting the cases in which it is possible.

16.2.2.  Attack II: Silencing a Client



   In this attack, the attacker provides a STUN client with a faked
   reflexive address.  The reflexive address it provides is a transport
   address that routes to nowhere.  As a result, the client won't
   receive any of the packets it expects to receive when it hands out
   the reflexive address.  This exploitation is not very interesting for
   the attacker.  It impacts a single client, which is frequently not
   the desired target.  Moreover, any attacker that can mount the attack
   could also deny service to the client by other means, such as
   preventing the client from receiving any response from the STUN
   server, or even a DHCP server.  As with the attack described in
   Section 16.2.1, this attack is only possible when the attacker is on
   path for packets sent from the STUN server towards this unused IP
   address.



Petit-Huguenin, et al.       Standards Track                   [Page 51]

RFC 8489                          STUN                     February 2020


16.2.3.  Attack III: Assuming the Identity of a Client



   This attack is similar to attack II.  However, the faked reflexive
   address points to the attacker itself.  This allows the attacker to
   receive traffic that was destined for the client.

16.2.4.  Attack IV: Eavesdropping



   In this attack, the attacker forces the client to use a reflexive
   address that routes to itself.  It then forwards any packets it
   receives to the client.  This attack allows the attacker to observe
   all packets sent to the client.  However, in order to launch the
   attack, the attacker must have already been able to observe packets
   from the client to the STUN server.  In most cases (such as when the
   attack is launched from an access network), this means that the
   attacker could already observe packets sent to the client.  This
   attack is, as a result, only useful for observing traffic by
   attackers on the path from the client to the STUN server, but not
   generally on the path of packets being routed towards the client.

   Note that this attack can be trivially launched by the STUN server
   itself, so users of STUN servers should have the same level of trust
   in the users of STUN servers as any other node that can insert itself
   into the communication flow.

16.3.  Hash Agility Plan



   This specification uses HMAC-SHA256 for computation of the message
   integrity, sometimes in combination with HMAC-SHA1.  If, at a later
   time, HMAC-SHA256 is found to be compromised, the following remedy
   should be applied:

   o  Both a new message-integrity attribute and a new STUN Security
      Feature bit will be allocated in a Standards Track document.  The
      new message-integrity attribute will have its value computed using
      a new hash.  The STUN Security Feature bit will be used to
      simultaneously 1) signal to a STUN client using the long-term
      credential mechanism that this server supports this new hash
      algorithm and 2) prevent bid-down attacks on the new message-
      integrity attribute.

   o  STUN clients and servers using the short-term credential mechanism
      will need to update the external mechanism that they use to signal
      what message-integrity attributes are in use.

   The bid-down protection mechanism described in this document is new
   and thus cannot currently protect against a bid-down attack that
   lowers the strength of the hash algorithm to HMAC-SHA1.  This is why,



Petit-Huguenin, et al.       Standards Track                   [Page 52]

RFC 8489                          STUN                     February 2020


   after a transition period, a new document updating this one will
   assign a new STUN Security Feature bit for deprecating HMAC-SHA1.
   When used, this bit will signal that HMAC-SHA1 is deprecated and
   should no longer be used.

   Similarly, if HMAC-SHA256 is found to be compromised, a new userhash
   attribute and a new STUN Security Feature bit will be allocated in a
   Standards Track document.  The new userhash attribute will have its
   value computed using a new hash.  The STUN Security Feature bit will
   be used to simultaneously 1) signal to a STUN client using the long-
   term credential mechanism that this server supports this new hash
   algorithm for the userhash attribute and 2) prevent bid-down attacks
   on the new userhash attribute.

17.  IAB Considerations



   The IAB has studied the problem of Unilateral Self-Address Fixing
   (UNSAF), which is the general process by which a client attempts to
   determine its address in another realm on the other side of a NAT
   through a collaborative protocol reflection mechanism [RFC3424].
   STUN can be used to perform this function using a Binding request/
   response transaction if one agent is behind a NAT and the other is on
   the public side of the NAT.

   The IAB has suggested that protocols developed for this purpose
   document a specific set of considerations.  Because some STUN Usages
   provide UNSAF functions (such as ICE [RFC8445]) and others do not
   (such as SIP Outbound [RFC5626]), answers to these considerations
   need to be addressed by the usages themselves.

18.  IANA Considerations



18.1.  STUN Security Features Registry



   A STUN Security Feature set defines 24 bits as flags.

   IANA has created a new registry containing the STUN Security Features
   that are protected by the bid-down attack prevention mechanism
   described in Section 9.2.1.

   The initial STUN Security Features are:

   Bit 0: Password algorithms
   Bit 1: Username anonymity
   Bit 2-23: Unassigned






Petit-Huguenin, et al.       Standards Track                   [Page 53]

RFC 8489                          STUN                     February 2020


   Bits are assigned starting from the most significant side of the bit
   set, so Bit 0 is the leftmost bit and Bit 23 is the rightmost bit.

   New Security Features are assigned by Standards Action [RFC8126].

18.2.  STUN Methods Registry



   A STUN method is a hex number in the range 0x000-0x0FF.  The encoding
   of a STUN method into a STUN message is described in Section 5.

   STUN methods in the range 0x000-0x07F are assigned by IETF Review
   [RFC8126].  STUN methods in the range 0x080-0x0FF are assigned by
   Expert Review [RFC8126].  The responsibility of the expert is to
   verify that the selected codepoint(s) is not in use and that the
   request is not for an abnormally large number of codepoints.
   Technical review of the extension itself is outside the scope of the
   designated expert responsibility.

   IANA has updated the name for method 0x002 as described below as well
   as updated the reference from RFC 5389 to RFC 8489 for the following
   STUN methods:

   0x000: Reserved
   0x001: Binding
   0x002: Reserved; was SharedSecret prior to [RFC5389]

18.3.  STUN Attributes Registry



   A STUN attribute type is a hex number in the range 0x0000-0xFFFF.
   STUN attribute types in the range 0x0000-0x7FFF are considered
   comprehension-required; STUN attribute types in the range
   0x8000-0xFFFF are considered comprehension-optional.  A STUN agent
   handles unknown comprehension-required and comprehension-optional
   attributes differently.

   STUN attribute types in the first half of the comprehension-required
   range (0x0000-0x3FFF) and in the first half of the comprehension-
   optional range (0x8000-0xBFFF) are assigned by IETF Review [RFC8126].
   STUN attribute types in the second half of the comprehension-required
   range (0x4000-0x7FFF) and in the second half of the comprehension-
   optional range (0xC000-0xFFFF) are assigned by Expert Review
   [RFC8126].  The responsibility of the expert is to verify that the
   selected codepoint(s) are not in use and that the request is not for
   an abnormally large number of codepoints.  Technical review of the
   extension itself is outside the scope of the designated expert
   responsibility.





Petit-Huguenin, et al.       Standards Track                   [Page 54]

RFC 8489                          STUN                     February 2020


18.3.1.  Updated Attributes



   IANA has updated the names for attributes 0x0002, 0x0004, 0x0005,
   0x0007, and 0x000B as well as updated the reference from RFC 5389 to
   RFC 8489 for each the following STUN methods.

   In addition, [RFC5389] introduced a mistake in the name of attribute
   0x0003; [RFC5389] called it CHANGE-ADDRESS when it was actually
   previously called CHANGE-REQUEST.  Thus, IANA has updated the
   description for 0x0003 to read "Reserved; was CHANGE-REQUEST prior to
   [RFC5389]".

   Comprehension-required range (0x0000-0x7FFF):
   0x0000: Reserved
   0x0001: MAPPED-ADDRESS
   0x0002: Reserved; was RESPONSE-ADDRESS prior to [RFC5389]
   0x0003: Reserved; was CHANGE-REQUEST prior to [RFC5389]
   0x0004: Reserved; was SOURCE-ADDRESS prior to [RFC5389]
   0x0005: Reserved; was CHANGED-ADDRESS prior to [RFC5389]
   0x0006: USERNAME
   0x0007: Reserved; was PASSWORD prior to [RFC5389]
   0x0008: MESSAGE-INTEGRITY
   0x0009: ERROR-CODE
   0x000A: UNKNOWN-ATTRIBUTES
   0x000B: Reserved; was REFLECTED-FROM prior to [RFC5389]
   0x0014: REALM
   0x0015: NONCE
   0x0020: XOR-MAPPED-ADDRESS

   Comprehension-optional range (0x8000-0xFFFF)
   0x8022: SOFTWARE
   0x8023: ALTERNATE-SERVER
   0x8028: FINGERPRINT

18.3.2.  New Attributes



   IANA has added the following attribute to the "STUN Attributes"
   registry:

   Comprehension-required range (0x0000-0x7FFF):
   0x001C: MESSAGE-INTEGRITY-SHA256
   0x001D: PASSWORD-ALGORITHM
   0x001E: USERHASH

   Comprehension-optional range (0x8000-0xFFFF)
   0x8002: PASSWORD-ALGORITHMS
   0x8003: ALTERNATE-DOMAIN




Petit-Huguenin, et al.       Standards Track                   [Page 55]

RFC 8489                          STUN                     February 2020


18.4.  STUN Error Codes Registry



   A STUN error code is a number in the range 0-699.  STUN error codes
   are accompanied by a textual reason phrase in UTF-8 [RFC3629] that is
   intended only for human consumption and can be anything appropriate;
   this document proposes only suggested values.

   STUN error codes are consistent in codepoint assignments and
   semantics with SIP [RFC3261] and HTTP [RFC7231].

   New STUN error codes are assigned based on IETF Review [RFC8126].
   The specification must carefully consider how clients that do not
   understand this error code will process it before granting the
   request.  See the rules in Section 6.3.4.

   IANA has updated the reference from RFC 5389 to RFC 8489 for the
   error codes defined in Section 14.8.

   IANA has changed the name of the 401 error code from "Unauthorized"
   to "Unauthenticated".

18.5.  STUN Password Algorithms Registry



   IANA has created a new registry titled "STUN Password Algorithms".

   A password algorithm is a hex number in the range 0x0000-0xFFFF.

   The initial contents of the "Password Algorithm" registry are as
   follows:

   0x0000: Reserved
   0x0001: MD5
   0x0002: SHA-256
   0x0003-0xFFFF: Unassigned

   Password algorithms in the first half of the range (0x0000-0x7FFF)
   are assigned by IETF Review [RFC8126].  Password algorithms in the
   second half of the range (0x8000-0xFFFF) are assigned by Expert
   Review [RFC8126].












Petit-Huguenin, et al.       Standards Track                   [Page 56]

RFC 8489                          STUN                     February 2020


18.5.1.  Password Algorithms



18.5.1.1.  MD5



   This password algorithm is taken from [RFC1321].

   The key length is 16 bytes, and the parameters value is empty.

      Note: This algorithm MUST only be used for compatibility with
      legacy systems.

                key = MD5(username ":" OpaqueString(realm)
                  ":" OpaqueString(password))

18.5.1.2.  SHA-256



   This password algorithm is taken from [RFC7616].

   The key length is 32 bytes, and the parameters value is empty.

              key = SHA-256(username ":" OpaqueString(realm)
                ":" OpaqueString(password))

18.6.  STUN UDP and TCP Port Numbers



   IANA has updated the reference from RFC 5389 to RFC 8489 for the
   following ports in the "Service Name and Transport Protocol Port
   Number Registry".

   stun   3478/tcp   Session Traversal Utilities for NAT (STUN) port
   stun   3478/udp   Session Traversal Utilities for NAT (STUN) port
   stuns  5349/tcp   Session Traversal Utilities for NAT (STUN) port

19.  Changes since RFC 5389



   This specification obsoletes [RFC5389].  This specification differs
   from RFC 5389 in the following ways:

   o  Added support for DTLS-over-UDP [RFC6347].

   o  Made clear that the RTO is considered stale if there are no
      transactions with the server.

   o  Aligned the RTO calculation with [RFC6298].

   o  Updated the ciphersuites for TLS.

   o  Added support for STUN URI [RFC7064].



Petit-Huguenin, et al.       Standards Track                   [Page 57]

RFC 8489                          STUN                     February 2020


   o  Added support for SHA256 message integrity.

   o  Updated the Preparation, Enforcement, and Comparison of
      Internationalized Strings (PRECIS) support to [RFC8265].

   o  Added protocol and registry to choose the password encryption
      algorithm.

   o  Added support for anonymous username.

   o  Added protocol and registry for preventing bid-down attacks.

   o  Specified that sharing a NONCE is no longer permitted.

   o  Added the possibility of using a domain name in the alternate
      server mechanism.

   o  Added more C snippets.

   o  Added test vector.

20.  References



20.1.  Normative References



   [ITU.V42.2002]
              International Telecommunication Union, "Error-correcting
              procedures for DCEs using asynchronous-to-synchronous
              conversion", ITU-T Recommendation V.42, March 2002.

   [KARN87]   Karn, P. and C. Partridge, "Improving Round-Trip Time
              Estimates in Reliable Transport Protocols", SIGCOMM '87,
              Proceedings of the ACM workshop on Frontiers in computer
              communications technology, Pages 2-7,
              DOI 10.1145/55483.55484, August 1987.

   [RFC0791]  Postel, J., "Internet Protocol", STD 5, RFC 791,
              DOI 10.17487/RFC0791, September 1981,
              <https://www.rfc-editor.org/info/rfc791>.

   [RFC1122]  Braden, R., Ed., "Requirements for Internet Hosts -
              Communication Layers", STD 3, RFC 1122,
              DOI 10.17487/RFC1122, October 1989,
              <https://www.rfc-editor.org/info/rfc1122>.







Petit-Huguenin, et al.       Standards Track                   [Page 58]

RFC 8489                          STUN                     February 2020


   [RFC1123]  Braden, R., Ed., "Requirements for Internet Hosts -
              Application and Support", STD 3, RFC 1123,
              DOI 10.17487/RFC1123, October 1989,
              <https://www.rfc-editor.org/info/rfc1123>.

   [RFC1321]  Rivest, R., "The MD5 Message-Digest Algorithm", RFC 1321,
              DOI 10.17487/RFC1321, April 1992,
              <https://www.rfc-editor.org/info/rfc1321>.

   [RFC2104]  Krawczyk, H., Bellare, M., and R. Canetti, "HMAC: Keyed-
              Hashing for Message Authentication", RFC 2104,
              DOI 10.17487/RFC2104, February 1997,
              <https://www.rfc-editor.org/info/rfc2104>.

   [RFC2119]  Bradner, S., "Key words for use in RFCs to Indicate
              Requirement Levels", BCP 14, RFC 2119,
              DOI 10.17487/RFC2119, March 1997,
              <https://www.rfc-editor.org/info/rfc2119>.

   [RFC2782]  Gulbrandsen, A., Vixie, P., and L. Esibov, "A DNS RR for
              specifying the location of services (DNS SRV)", RFC 2782,
              DOI 10.17487/RFC2782, February 2000,
              <https://www.rfc-editor.org/info/rfc2782>.

   [RFC3629]  Yergeau, F., "UTF-8, a transformation format of ISO
              10646", STD 63, RFC 3629, DOI 10.17487/RFC3629, November
              2003, <https://www.rfc-editor.org/info/rfc3629>.

   [RFC4648]  Josefsson, S., "The Base16, Base32, and Base64 Data
              Encodings", RFC 4648, DOI 10.17487/RFC4648, October 2006,
              <https://www.rfc-editor.org/info/rfc4648>.

   [RFC5890]  Klensin, J., "Internationalized Domain Names for
              Applications (IDNA): Definitions and Document Framework",
              RFC 5890, DOI 10.17487/RFC5890, August 2010,
              <https://www.rfc-editor.org/info/rfc5890>.

   [RFC6125]  Saint-Andre, P. and J. Hodges, "Representation and
              Verification of Domain-Based Application Service Identity
              within Internet Public Key Infrastructure Using X.509
              (PKIX) Certificates in the Context of Transport Layer
              Security (TLS)", RFC 6125, DOI 10.17487/RFC6125, March
              2011, <https://www.rfc-editor.org/info/rfc6125>.

   [RFC6151]  Turner, S. and L. Chen, "Updated Security Considerations
              for the MD5 Message-Digest and the HMAC-MD5 Algorithms",
              RFC 6151, DOI 10.17487/RFC6151, March 2011,
              <https://www.rfc-editor.org/info/rfc6151>.



Petit-Huguenin, et al.       Standards Track                   [Page 59]

RFC 8489                          STUN                     February 2020


   [RFC6298]  Paxson, V., Allman, M., Chu, J., and M. Sargent,
              "Computing TCP's Retransmission Timer", RFC 6298,
              DOI 10.17487/RFC6298, June 2011,
              <https://www.rfc-editor.org/info/rfc6298>.

   [RFC6347]  Rescorla, E. and N. Modadugu, "Datagram Transport Layer
              Security Version 1.2", RFC 6347, DOI 10.17487/RFC6347,
              January 2012, <https://www.rfc-editor.org/info/rfc6347>.

   [RFC7064]  Nandakumar, S., Salgueiro, G., Jones, P., and M. Petit-
              Huguenin, "URI Scheme for the Session Traversal Utilities
              for NAT (STUN) Protocol", RFC 7064, DOI 10.17487/RFC7064,
              November 2013, <https://www.rfc-editor.org/info/rfc7064>.

   [RFC7350]  Petit-Huguenin, M. and G. Salgueiro, "Datagram Transport
              Layer Security (DTLS) as Transport for Session Traversal
              Utilities for NAT (STUN)", RFC 7350, DOI 10.17487/RFC7350,
              August 2014, <https://www.rfc-editor.org/info/rfc7350>.

   [RFC7616]  Shekh-Yusef, R., Ed., Ahrens, D., and S. Bremer, "HTTP
              Digest Access Authentication", RFC 7616,
              DOI 10.17487/RFC7616, September 2015,
              <https://www.rfc-editor.org/info/rfc7616>.

   [RFC8174]  Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC
              2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174,
              May 2017, <https://www.rfc-editor.org/info/rfc8174>.

   [RFC8200]  Deering, S. and R. Hinden, "Internet Protocol, Version 6
              (IPv6) Specification", STD 86, RFC 8200,
              DOI 10.17487/RFC8200, July 2017,
              <https://www.rfc-editor.org/info/rfc8200>.

   [RFC8265]  Saint-Andre, P. and A. Melnikov, "Preparation,
              Enforcement, and Comparison of Internationalized Strings
              Representing Usernames and Passwords", RFC 8265,
              DOI 10.17487/RFC8265, October 2017,
              <https://www.rfc-editor.org/info/rfc8265>.

   [RFC8305]  Schinazi, D. and T. Pauly, "Happy Eyeballs Version 2:
              Better Connectivity Using Concurrency", RFC 8305,
              DOI 10.17487/RFC8305, December 2017,
              <https://www.rfc-editor.org/info/rfc8305>.








Petit-Huguenin, et al.       Standards Track                   [Page 60]

RFC 8489                          STUN                     February 2020


20.2.  Informative References



   [Argon2]   Biryukov, A., Dinu, D., Khovratovich, D., and S.
              Josefsson, "The memory-hard Argon2 password hash and
              proof-of-work function", Work in Progress, draft-irtf-
              cfrg-argon2-09, November 2019.

   [BCP195]   Sheffer, Y., Holz, R., and P. Saint-Andre,
              "Recommendations for Secure Use of Transport Layer
              Security (TLS) and Datagram Transport Layer Security
              (DTLS)", BCP 195, RFC 7525, May 2015,
              <https://www.rfc-editor.org/info/bcp195>.

   [RFC1952]  Deutsch, P., "GZIP file format specification version 4.3",
              RFC 1952, DOI 10.17487/RFC1952, May 1996,
              <https://www.rfc-editor.org/info/rfc1952>.

   [RFC2279]  Yergeau, F., "UTF-8, a transformation format of ISO
              10646", RFC 2279, DOI 10.17487/RFC2279, January 1998,
              <https://www.rfc-editor.org/info/rfc2279>.

   [RFC3261]  Rosenberg, J., Schulzrinne, H., Camarillo, G., Johnston,
              A., Peterson, J., Sparks, R., Handley, M., and E.
              Schooler, "SIP: Session Initiation Protocol", RFC 3261,
              DOI 10.17487/RFC3261, June 2002,
              <https://www.rfc-editor.org/info/rfc3261>.

   [RFC3424]  Daigle, L., Ed. and IAB, "IAB Considerations for
              UNilateral Self-Address Fixing (UNSAF) Across Network
              Address Translation", RFC 3424, DOI 10.17487/RFC3424,
              November 2002, <https://www.rfc-editor.org/info/rfc3424>.

   [RFC3454]  Hoffman, P. and M. Blanchet, "Preparation of
              Internationalized Strings ("stringprep")", RFC 3454,
              DOI 10.17487/RFC3454, December 2002,
              <https://www.rfc-editor.org/info/rfc3454>.

   [RFC3489]  Rosenberg, J., Weinberger, J., Huitema, C., and R. Mahy,
              "STUN - Simple Traversal of User Datagram Protocol (UDP)
              Through Network Address Translators (NATs)", RFC 3489,
              DOI 10.17487/RFC3489, March 2003,
              <https://www.rfc-editor.org/info/rfc3489>.

   [RFC4013]  Zeilenga, K., "SASLprep: Stringprep Profile for User Names
              and Passwords", RFC 4013, DOI 10.17487/RFC4013, February
              2005, <https://www.rfc-editor.org/info/rfc4013>.





Petit-Huguenin, et al.       Standards Track                   [Page 61]

RFC 8489                          STUN                     February 2020


   [RFC4107]  Bellovin, S. and R. Housley, "Guidelines for Cryptographic
              Key Management", BCP 107, RFC 4107, DOI 10.17487/RFC4107,
              June 2005, <https://www.rfc-editor.org/info/rfc4107>.

   [RFC5090]  Sterman, B., Sadolevsky, D., Schwartz, D., Williams, D.,
              and W. Beck, "RADIUS Extension for Digest Authentication",
              RFC 5090, DOI 10.17487/RFC5090, February 2008,
              <https://www.rfc-editor.org/info/rfc5090>.

   [RFC5389]  Rosenberg, J., Mahy, R., Matthews, P., and D. Wing,
              "Session Traversal Utilities for NAT (STUN)", RFC 5389,
              DOI 10.17487/RFC5389, October 2008,
              <https://www.rfc-editor.org/info/rfc5389>.

   [RFC5626]  Jennings, C., Ed., Mahy, R., Ed., and F. Audet, Ed.,
              "Managing Client-Initiated Connections in the Session
              Initiation Protocol (SIP)", RFC 5626,
              DOI 10.17487/RFC5626, October 2009,
              <https://www.rfc-editor.org/info/rfc5626>.

   [RFC5766]  Mahy, R., Matthews, P., and J. Rosenberg, "Traversal Using
              Relays around NAT (TURN): Relay Extensions to Session
              Traversal Utilities for NAT (STUN)", RFC 5766,
              DOI 10.17487/RFC5766, April 2010,
              <https://www.rfc-editor.org/info/rfc5766>.

   [RFC5769]  Denis-Courmont, R., "Test Vectors for Session Traversal
              Utilities for NAT (STUN)", RFC 5769, DOI 10.17487/RFC5769,
              April 2010, <https://www.rfc-editor.org/info/rfc5769>.

   [RFC5780]  MacDonald, D. and B. Lowekamp, "NAT Behavior Discovery
              Using Session Traversal Utilities for NAT (STUN)",
              RFC 5780, DOI 10.17487/RFC5780, May 2010,
              <https://www.rfc-editor.org/info/rfc5780>.

   [RFC6544]  Rosenberg, J., Keranen, A., Lowekamp, B., and A. Roach,
              "TCP Candidates with Interactive Connectivity
              Establishment (ICE)", RFC 6544, DOI 10.17487/RFC6544,
              March 2012, <https://www.rfc-editor.org/info/rfc6544>.

   [RFC7231]  Fielding, R., Ed. and J. Reschke, Ed., "Hypertext Transfer
              Protocol (HTTP/1.1): Semantics and Content", RFC 7231,
              DOI 10.17487/RFC7231, June 2014,
              <https://www.rfc-editor.org/info/rfc7231>.







Petit-Huguenin, et al.       Standards Track                   [Page 62]

RFC 8489                          STUN                     February 2020


   [RFC8126]  Cotton, M., Leiba, B., and T. Narten, "Guidelines for
              Writing an IANA Considerations Section in RFCs", BCP 26,
              RFC 8126, DOI 10.17487/RFC8126, June 2017,
              <https://www.rfc-editor.org/info/rfc8126>.

   [RFC8264]  Saint-Andre, P. and M. Blanchet, "PRECIS Framework:
              Preparation, Enforcement, and Comparison of
              Internationalized Strings in Application Protocols",
              RFC 8264, DOI 10.17487/RFC8264, October 2017,
              <https://www.rfc-editor.org/info/rfc8264>.

   [RFC8445]  Keranen, A., Holmberg, C., and J. Rosenberg, "Interactive
              Connectivity Establishment (ICE): A Protocol for Network
              Address Translator (NAT) Traversal", RFC 8445,
              DOI 10.17487/RFC8445, July 2018,
              <https://www.rfc-editor.org/info/rfc8445>.

   [RFC8446]  Rescorla, E., "The Transport Layer Security (TLS) Protocol
              Version 1.3", RFC 8446, DOI 10.17487/RFC8446, August 2018,
              <https://www.rfc-editor.org/info/rfc8446>.

   [STUN-PMTUD]
              Petit-Huguenin, M., Salgueiro, G., and F. Garrido,
              "Packetization Layer Path MTU Discovery (PLMTUD) For UDP
              Transports Using Session Traversal Utilities for NAT
              (STUN)", Work in Progress, draft-ietf-tram-stun-pmtud-15,
              December 2019.

   [UAX15]    Unicode Standard Annex #15, "Unicode Normalization Forms",
              edited by Mark Davis and Ken Whistler.  An integral part
              of The Unicode Standard,
              <http://unicode.org/reports/tr15/>.



















Petit-Huguenin, et al.       Standards Track                   [Page 63]

RFC 8489                          STUN                     February 2020


Appendix A.  C Snippet to Determine STUN Message Types



   Given a 16-bit STUN message type value in host byte order in msg_type
   parameter, below are C macros to determine the STUN message types:

   <CODE BEGINS>
   #define IS_REQUEST(msg_type)       (((msg_type) & 0x0110) == 0x0000)
   #define IS_INDICATION(msg_type)    (((msg_type) & 0x0110) == 0x0010)
   #define IS_SUCCESS_RESP(msg_type)  (((msg_type) & 0x0110) == 0x0100)
   #define IS_ERR_RESP(msg_type)      (((msg_type) & 0x0110) == 0x0110)
   <CODE ENDS>

   A function to convert method and class into a message type:

   <CODE BEGINS>
   int type(int method, int cls) {
     return (method & 0x1F80) << 2 | (method & 0x0070) << 1
       | (method & 0x000F) | (cls & 0x0002) << 7
       | (cls & 0x0001) << 4;
     }
   <CODE ENDS>

   A function to extract the method from the message type:

   <CODE BEGINS>
   int method(int type) {
     return (type & 0x3E00) >> 2 | (type & 0x00E0) >> 1
       | (type & 0x000F);
     }
   <CODE ENDS>

   A function to extract the class from the message type:

   <CODE BEGINS>
   int cls(int type) {
     return (type & 0x0100) >> 7 | (type & 0x0010) >> 4;
     }
   <CODE ENDS>













Petit-Huguenin, et al.       Standards Track                   [Page 64]

RFC 8489                          STUN                     February 2020


Appendix B.  Test Vectors



   This section augments the list of test vectors defined in [RFC5769]
   with MESSAGE-INTEGRITY-SHA256.  All the formats and definitions
   listed in Section 2 of [RFC5769] apply here.

B.1.  Sample Request with Long-Term Authentication with MESSAGE-
      INTEGRITY-SHA256 and USERHASH



   This request uses the following parameters:

   Username: "<U+30DE><U+30C8><U+30EA><U+30C3><U+30AF><U+30B9>" (without
   quotes) unaffected by OpaqueString [RFC8265] processing

   Password: "The<U+00AD>M<U+00AA>tr<U+2168>" and "TheMatrIX" (without
   quotes) respectively before and after OpaqueString [RFC8265]
   processing

   Nonce: "obMatJos2AAACf//499k954d6OL34oL9FSTvy64sA" (without quotes)

   Realm: "example.org" (without quotes)

        00 01 00 9c      Request type and message length
        21 12 a4 42      Magic cookie
        78 ad 34 33   }
        c6 ad 72 c0   }  Transaction ID
        29 da 41 2e   }
        00 1e 00 20      USERHASH attribute header
        4a 3c f3 8f   }
        ef 69 92 bd   }
        a9 52 c6 78   }
        04 17 da 0f   }  Userhash value (32 bytes)
        24 81 94 15   }
        56 9e 60 b2   }
        05 c4 6e 41   }
        40 7f 17 04   }
        00 15 00 29      NONCE attribute header
        6f 62 4d 61   }
        74 4a 6f 73   }
        32 41 41 41   }
        43 66 2f 2f   }
        34 39 39 6b   }  Nonce value and padding (3 bytes)
        39 35 34 64   }
        36 4f 4c 33   }
        34 6f 4c 39   }
        46 53 54 76   }
        79 36 34 73   }
        41 00 00 00   }



Petit-Huguenin, et al.       Standards Track                   [Page 65]

RFC 8489                          STUN                     February 2020


        00 14 00 0b      REALM attribute header
        65 78 61 6d   }
        70 6c 65 2e   }  Realm value (11 bytes) and padding (1 byte)
        6f 72 67 00   }
        00 1c 00 20      MESSAGE-INTEGRITY-SHA256 attribute header
        e4 68 6c 8f   }
        0e de b5 90   }
        13 e0 70 90   }
        01 0a 93 ef   }  HMAC-SHA256 value
        cc bc cc 54   }
        4c 0a 45 d9   }
        f8 30 aa 6d   }
        6f 73 5a 01   }

Acknowledgements



   Thanks to Michael Tuexen, Tirumaleswar Reddy, Oleg Moskalenko, Simon
   Perreault, Benjamin Schwartz, Rifaat Shekh-Yusef, Alan Johnston,
   Jonathan Lennox, Brandon Williams, Olle Johansson, Martin Thomson,
   Mihaly Meszaros, Tolga Asveren, Noriyuki Torii, Spencer Dawkins, Dale
   Worley, Matthew Miller, Peter Saint-Andre, Julien Elie, Mirja
   Kuehlewind, Eric Rescorla, Ben Campbell, Adam Roach, Alexey Melnikov,
   and Benjamin Kaduk for the comments, suggestions, and questions that
   helped improve this document.

   The Acknowledgements section of RFC 5389 appeared as follows:

   The authors would like to thank Cedric Aoun, Pete Cordell, Cullen
   Jennings, Bob Penfield, Xavier Marjou, Magnus Westerlund, Miguel
   Garcia, Bruce Lowekamp, and Chris Sullivan for their comments, and
   Baruch Sterman and Alan Hawrylyshen for initial implementations.
   Thanks for Leslie Daigle, Allison Mankin, Eric Rescorla, and Henning
   Schulzrinne for IESG and IAB input on this work.

Contributors

   Christian Huitema and Joel Weinberger were original coauthors of
   RFC 3489.













Petit-Huguenin, et al.       Standards Track                   [Page 66]

RFC 8489                          STUN                     February 2020


Authors' Addresses



   Marc Petit-Huguenin
   Impedance Mismatch

   Email: marc@petit-huguenin.org


   Gonzalo Salgueiro
   Cisco
   7200-12 Kit Creek Road
   Research Triangle Park, NC  27709
   United States of America

   Email: gsalguei@cisco.com


   Jonathan Rosenberg
   Five9
   Edison, NJ
   United States of America

   Email: jdrosen@jdrosen.net
   URI:   http://www.jdrosen.net


   Dan Wing
   Citrix Systems, Inc.
   United States of America

   Email: dwing-ietf@fuggles.com


   Rohan Mahy
   Unaffiliated

   Email: rohan.ietf@gmail.com


   Philip Matthews
   Nokia
   600 March Road
   Ottawa, Ontario  K2K 2T6
   Canada

   Phone: 613-784-3139
   Email: philip_matthews@magma.ca




Petit-Huguenin, et al.       Standards Track                   [Page 67]