Internet Engineering Task Force (IETF)                             C. Dô
Request for Comments: 8967                                W. Kolodziejak
Obsoletes: 7298                                            J. Chroboczek
Category: Standards Track              IRIF, University of Paris-Diderot
ISSN: 2070-1721                                             January 2021


          MAC Authentication for the Babel Routing Protocol

Abstract

  This document describes a cryptographic authentication mechanism for
  the Babel routing protocol that has provisions for replay avoidance.
  This document obsoletes RFC 7298.

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/rfc8967.

Copyright Notice

  Copyright (c) 2021 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
    1.1.  Applicability
    1.2.  Assumptions and Security Properties
    1.3.  Specification of Requirements
  2.  Conceptual Overview of the Protocol
  3.  Data Structures
    3.1.  The Interface Table
    3.2.  The Neighbour Table
  4.  Protocol Operation
    4.1.  MAC Computation
    4.2.  Packet Transmission
    4.3.  Packet Reception
    4.4.  Expiring Per-Neighbour State
  5.  Incremental Deployment and Key Rotation
  6.  Packet Format
    6.1.  MAC TLV
    6.2.  PC TLV
    6.3.  Challenge Request TLV
    6.4.  Challenge Reply TLV
  7.  Security Considerations
  8.  IANA Considerations
  9.  References
    9.1.  Normative References
    9.2.  Informational References
  Acknowledgments
  Authors' Addresses

1.  Introduction

  By default, the Babel routing protocol [RFC8966] trusts the
  information contained in every UDP datagram that it receives on the
  Babel port.  An attacker can redirect traffic to itself or to a
  different node in the network, causing a variety of potential issues.
  In particular, an attacker might:

  *  spoof a Babel packet and redirect traffic by announcing a route
     with a smaller metric, a larger sequence number, or a longer
     prefix;

  *  spoof a malformed packet, which could cause an insufficiently
     robust implementation to crash or interfere with the rest of the
     network;

  *  replay a previously captured Babel packet, which could cause
     traffic to be redirected or otherwise interfere with the network.

  Protecting a Babel network is challenging due to the fact that the
  Babel protocol uses both unicast and multicast communication.  One
  possible approach, used notably by the Babel over Datagram Transport
  Layer Security (DTLS) protocol [RFC8968], is to use unicast
  communication for all semantically significant communication, and
  then use a standard unicast security protocol to protect the Babel
  traffic.  In this document, we take the opposite approach: we define
  a cryptographic extension to the Babel protocol that is able to
  protect both unicast and multicast traffic and thus requires very few
  changes to the core protocol.  This document obsoletes [RFC7298].

1.1.  Applicability

  The protocol defined in this document assumes that all interfaces on
  a given link are equally trusted and share a small set of symmetric
  keys (usually just one, and two during key rotation).  The protocol
  is inapplicable in situations where asymmetric keying is required,
  where the trust relationship is partial, or where large numbers of
  trusted keys are provisioned on a single link at the same time.

  This protocol supports incremental deployment (where an insecure
  Babel network is made secure with no service interruption), and it
  supports graceful key rotation (where the set of keys is changed with
  no service interruption).

  This protocol does not require synchronised clocks, it does not
  require persistently monotonic clocks, and it does not require
  persistent storage except for what might be required for storing
  cryptographic keys.

1.2.  Assumptions and Security Properties

  The correctness of the protocol relies on the following assumptions:

  *  that the Message Authentication Code (MAC) being used is
     invulnerable to forgery, i.e., that an attacker is unable to
     generate a packet with a correct MAC without access to the secret
     key;

  *  that a node never generates the same index or nonce twice over the
     lifetime of a key.

  The first assumption is a property of the MAC being used.  The second
  assumption can be met either by using a robust random number
  generator [RFC4086] and sufficiently large indices and nonces, by
  using a reliable hardware clock, or by rekeying often enough that
  collisions are unlikely.

  If the assumptions above are met, the protocol described in this
  document has the following properties:

  *  it is invulnerable to spoofing: any Babel packet accepted as
     authentic is the exact copy of a packet originally sent by an
     authorised node;

  *  locally to a single node, it is invulnerable to replay: if a node
     has previously accepted a given packet, then it will never again
     accept a copy of this packet or an earlier packet from the same
     sender;

  *  among different nodes, it is only vulnerable to immediate replay:
     if a node A has accepted an authentic packet from C, then a node B
     will only accept a copy of that packet if B has accepted an older
     packet from C, and B has received no later packet from C.

  While this protocol makes efforts to mitigate the effects of a denial
  of service attack, it does not fully protect against such attacks.

1.3.  Specification of Requirements

  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.

2.  Conceptual Overview of the Protocol

  When a node B sends out a Babel packet through an interface that is
  configured for MAC cryptographic protection, it computes one or more
  MACs (one per key) that it appends to the packet.  When a node A
  receives a packet over an interface that requires MAC cryptographic
  protection, it independently computes a set of MACs and compares them
  to the MACs appended to the packet; if there is no match, the packet
  is discarded.

  In order to protect against replay, B maintains a per-interface
  32-bit integer known as the "packet counter" (PC).  Whenever B sends
  a packet through the interface, it embeds the current value of the PC
  within the region of the packet that is protected by the MACs and
  increases the PC by at least one.  When A receives the packet, it
  compares the value of the PC with the one contained in the previous
  packet received from B, and unless it is strictly greater, the packet
  is discarded.

  By itself, the PC mechanism is not sufficient to protect against
  replay.  Consider a peer A that has no information about a peer B
  (e.g., because it has recently rebooted).  Suppose that A receives a
  packet ostensibly from B carrying a given PC; since A has no
  information about B, it has no way to determine whether the packet is
  freshly generated or a replay of a previously sent packet.

  In this situation, peer A discards the packet and challenges B to
  prove that it knows the MAC key.  It sends a "Challenge Request", a
  TLV containing a unique nonce, a value that has never been used
  before and will never be used again.  Peer B replies to the Challenge
  Request with a "Challenge Reply", a TLV containing a copy of the
  nonce chosen by A, in a packet protected by MAC and containing the
  new value of B's PC.  Since the nonce has never been used before, B's
  reply proves B's knowledge of the MAC key and the freshness of the
  PC.

  By itself, this mechanism is safe against replay if B never resets
  its PC.  In practice, however, this is difficult to ensure, as
  persistent storage is prone to failure, and hardware clocks, even
  when available, are occasionally reset.  Suppose that B resets its PC
  to an earlier value and sends a packet with a previously used PC n.
  Peer A challenges B, B successfully responds to the challenge, and A
  accepts the PC equal to n + 1.  At this point, an attacker C may send
  a replayed packet with PC equal to n + 2, which will be accepted by
  A.

  Another mechanism is needed to protect against this attack.  In this
  protocol, every PC is tagged with an "index", an arbitrary string of
  octets.  Whenever B resets its PC, or whenever B doesn't know whether
  its PC has been reset, it picks an index that it has never used
  before (either by drawing it randomly or by using a reliable hardware
  clock) and starts sending PCs with that index.  Whenever A detects
  that B has changed its index, it challenges B again.

  With this additional mechanism, this protocol is invulnerable to
  replay attacks (see Section 1.2).

3.  Data Structures

  Every Babel node maintains a set of conceptual data structures
  described in Section 3.2 of [RFC8966].  This protocol extends these
  data structures as follows.

3.1.  The Interface Table

  Every Babel node maintains an interface table, as described in
  Section 3.2.3 of [RFC8966].  Implementations of this protocol MUST
  allow each interface to be provisioned with a set of one or more MAC
  keys and the associated MAC algorithms (see Section 4.1 for suggested
  algorithms and Section 7 for suggested methods for key generation).
  In order to allow incremental deployment of this protocol (see
  Section 5), implementations SHOULD allow an interface to be
  configured in a mode in which it participates in the MAC
  authentication protocol but accepts packets that are not
  authenticated.

  This protocol extends each table entry associated with an interface
  on which MAC authentication has been configured with two new pieces
  of data:

  *  a set of one or more MAC keys, each associated with a given MAC
     algorithm;

  *  a pair (Index, PC), where Index is an arbitrary string of 0 to 32
     octets, and PC is a 32-bit (4-octet) integer.

  We say that an index is fresh when it has never been used before with
  any of the keys currently configured on the interface.  The Index
  field is initialised to a fresh index, for example, by drawing a
  random string of sufficient length (see Section 7 for suggested
  sizes), and the PC is initialised to an arbitrary value (typically
  0).

3.2.  The Neighbour Table

  Every Babel node maintains a neighbour table, as described in
  Section 3.2.4 of [RFC8966].  This protocol extends each entry in this
  table with two new pieces of data:

  *  a pair (Index, PC), where Index is a string of 0 to 32 octets, and
     PC is a 32-bit (4-octet) integer;

  *  a Nonce, which is an arbitrary string of 0 to 192 octets, and an
     associated challenge expiry timer.

  The Index and PC are initially undefined, and they are managed as
  described in Section 4.3.  The Nonce and challenge expiry timer are
  initially undefined, and they are used as described in
  Section 4.3.1.1.

4.  Protocol Operation

4.1.  MAC Computation

  A Babel node computes the MAC of a Babel packet as follows.

  First, the node builds a pseudo-header that will participate in MAC
  computation but will not be sent.  If the packet is carried over
  IPv6, the pseudo-header has the following format:

   0                   1                   2                   3
   0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  |                                                               |
  +                                                               +
  |                                                               |
  +                          Src address                          +
  |                                                               |
  +                                                               +
  |                                                               |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  |           Src port            |                               |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+                               +
  |                                                               |
  +                                                               +
  |                         Dest address                          |
  +                                                               +
  |                                                               |
  +                               +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  |                               |           Dest port           |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

  If the packet is carried over IPv4, the pseudo-header has the
  following format:

   0                   1                   2                   3
   0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  |                          Src address                          |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  |           Src port            |        Dest address           |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  |                               |           Dest port           |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

  Fields:

  Src address   The source IP address of the packet.

  Src port      The source UDP port number of the packet.

  Dest address  The destination IP address of the packet.

  Src port      The destination UDP port number of the packet.

  The node takes the concatenation of the pseudo-header and the Babel
  packet including the packet header but excluding the packet trailer
  (from octet 0 inclusive up to (Body Length + 4) exclusive) and
  computes a MAC with one of the implemented algorithms.  Every
  implementation MUST implement HMAC-SHA256 as defined in [RFC6234] and
  Section 2 of [RFC2104], SHOULD implement keyed BLAKE2s [RFC7693] with
  128-bit (16-octet) digests, and MAY implement other MAC algorithms.

4.2.  Packet Transmission

  A Babel node might delay actually sending TLVs by a small amount, in
  order to aggregate multiple TLVs in a single packet up to the
  interface MTU (Section 4 of [RFC8966]).  For an interface on which
  MAC protection is configured, the TLV aggregation logic MUST take
  into account the overhead due to PC TLVs (one in each packet) and MAC
  TLVs (one per configured key).

  Before sending a packet, the following actions are performed:

  *  a PC TLV containing the PC and Index associated with the outgoing
     interface MUST be appended to the packet body;

     -  the PC MUST be incremented by a strictly positive amount
        (typically just 1);

     -  if the PC overflows, a fresh index MUST be generated (as
        defined in Section 3.1);

     a node MUST NOT include multiple PC TLVs in a single packet;

  *  for each key configured on the interface, a MAC is computed as
     specified in Section 4.1 and stored in a MAC TLV that MUST be
     appended to the packet trailer (see Section 4.2 of [RFC8966]).

4.3.  Packet Reception

  When a packet is received on an interface that is configured for MAC
  protection, the following steps are performed before the packet is
  passed to normal processing:

  *  First, the receiver checks whether the trailer of the received
     packet carries at least one MAC TLV; if not, the packet MUST be
     immediately dropped and processing stops.  Then, for each key
     configured on the receiving interface, the receiver computes the
     MAC of the packet.  It then compares every generated MAC against
     every MAC included in the packet; if there is at least one match,
     the packet passes the MAC test; if there is none, the packet MUST
     be silently dropped and processing stops at this point.  In order
     to avoid memory exhaustion attacks, an entry in the neighbour
     table MUST NOT be created before the MAC test has passed
     successfully.  The MAC of the packet MUST NOT be computed for each
     MAC TLV contained in the packet, but only once for each configured
     key.

  *  If an entry for the sender does not exist in the neighbour table,
     it MAY be created at this point (or, alternatively, its creation
     can be delayed until a challenge needs to be sent, see below).

  *  The packet body is then parsed a first time.  During this
     "preparse" phase, the packet body is traversed and all TLVs are
     ignored except PC, Challenge Request, and Challenge Reply TLVs.
     When a PC TLV is encountered, the enclosed PC and Index are saved
     for later processing.  If multiple PCs are found (which should not
     happen, see Section 4.2), only the first one is processed, the
     remaining ones MUST be silently ignored.  If a Challenge Request
     is encountered, a Challenge Reply MUST be scheduled, as described
     in Section 4.3.1.2.  If a Challenge Reply is encountered, it is
     tested for validity as described in Section 4.3.1.3, and a note is
     made of the result of the test.

  *  The preparse phase above yields two pieces of data: the PC and
     Index from the first PC TLV, and a bit indicating whether the
     packet contains a successful Challenge Reply.  If the packet does
     not contain a PC TLV, the packet MUST be dropped, and processing
     stops at this point.  If the packet contains a successful
     Challenge Reply, then the PC and Index contained in the PC TLV
     MUST be stored in the neighbour table entry corresponding to the
     sender (which already exists in this case), and the packet is
     accepted.

  *  Otherwise, if there is no entry in the neighbour table
     corresponding to the sender, or if such an entry exists but
     contains no Index, or if the Index it contains is different from
     the Index contained in the PC TLV, then a challenge MUST be sent
     as described in Section 4.3.1.1, the packet MUST be dropped, and
     processing stops at this stage.

  *  At this stage, the packet contains no successful Challenge Reply,
     and the Index contained in the PC TLV is equal to the Index in the
     neighbour table entry corresponding to the sender.  The receiver
     compares the received PC with the PC contained in the neighbour
     table; if the received PC is smaller or equal than the PC
     contained in the neighbour table, the packet MUST be dropped and
     processing stops (no challenge is sent in this case, since the
     mismatch might be caused by harmless packet reordering on the
     link).  Otherwise, the PC contained in the neighbour table entry
     is set to the received PC, and the packet is accepted.

  In the algorithm described above, Challenge Requests are processed
  and challenges are sent before the (Index, PC) pair is verified
  against the neighbour table.  This simplifies the implementation
  somewhat (the node may simply schedule outgoing requests as it walks
  the packet during the preparse phase) but relies on the rate limiting
  described in Section 4.3.1.1 to avoid sending too many challenges in
  response to replayed packets.  As an optimisation, a node MAY ignore
  all Challenge Requests contained in a packet except the last one, and
  it MAY ignore a Challenge Request in the case where it is contained
  in a packet with an Index that matches the one in the neighbour table
  and a PC that is smaller or equal to the one contained in the
  neighbour table.  Since it is still possible to replay a packet with
  an obsolete Index, the rate limiting described in Section 4.3.1.1 is
  required even if this optimisation is implemented.

  The same is true of Challenge Replies.  However, since validating a
  Challenge Reply has minimal additional cost (it is just a bitwise
  comparison of two strings of octets), a similar optimisation for
  Challenge Replies is not worthwhile.

  After the packet has been accepted, it is processed as normal, except
  that any PC, Challenge Request, and Challenge Reply TLVs that it
  contains are silently ignored.

4.3.1.  Challenge Requests and Replies

  During the preparse stage, the receiver might encounter a mismatched
  Index, to which it will react by scheduling a Challenge Request.  It
  might encounter a Challenge Request TLV, to which it will reply with
  a Challenge Reply TLV.  Finally, it might encounter a Challenge Reply
  TLV, which it will attempt to match with a previously sent Challenge
  Request TLV in order to update the neighbour table entry
  corresponding to the sender of the packet.

4.3.1.1.  Sending Challenges

  When it encounters a mismatched Index during the preparse phase, a
  node picks a nonce that it has never used with any of the keys
  currently configured on the relevant interface, for example, by
  drawing a sufficiently large random string of bytes or by consulting
  a strictly monotonic hardware clock.  It MUST then store the nonce in
  the entry of the neighbour table associated to the neighbour (the
  entry might need to be created at this stage), initialise the
  neighbour's challenge expiry timer to 30 seconds, and send a
  Challenge Request TLV to the unicast address corresponding to the
  neighbour.

  A node MAY aggregate a Challenge Request with other TLVs; in other
  words, if it has already buffered TLVs to be sent to the unicast
  address of the neighbour, it MAY send the buffered TLVs in the same
  packet as the Challenge Request.  However, it MUST arrange for the
  Challenge Request to be sent in a timely manner, as any packets
  received from that neighbour will be silently ignored until the
  challenge completes.

  A node MUST impose a rate limitation to the challenges it sends; the
  limit SHOULD default to one Challenge Request every 300 ms and MAY be
  configurable.  This rate limiting serves two purposes.  First, since
  a challenge may be sent in response to a packet replayed by an
  attacker, it limits the number of challenges that an attacker can
  cause a node to send.  Second, it limits the number of challenges
  sent when there are multiple packets in flight from a single
  neighbour.

4.3.1.2.  Replying to Challenges

  When it encounters a Challenge Request during the preparse phase, a
  node constructs a Challenge Reply TLV by copying the Nonce from the
  Challenge Request into the Challenge Reply.  It MUST then send the
  Challenge Reply to the unicast address from which the Challenge
  Request was sent.  A challenge sent to a multicast address MUST be
  silently ignored.

  A node MAY aggregate a Challenge Reply with other TLVs; in other
  words, if it has already buffered TLVs to be sent to the unicast
  address of the sender of the Challenge Request, it MAY send the
  buffered TLVs in the same packet as the Challenge Reply.  However, it
  MUST arrange for the Challenge Reply to be sent in a timely manner
  (within a few seconds) and SHOULD NOT send any other packets over the
  same interface before sending the Challenge Reply, as those would be
  dropped by the challenger.

  Since a Challenge Reply might be caused by a replayed Challenge
  Request, a node MUST impose a rate limitation to the Challenge
  Replies it sends; the limit SHOULD default to one Challenge Reply for
  each peer every 300 ms and MAY be configurable.

4.3.1.3.  Receiving Challenge Replies

  When it encounters a Challenge Reply during the preparse phase, a
  node consults the neighbour table entry corresponding to the
  neighbour that sent the Challenge Reply.  If no challenge is in
  progress, i.e., if there is no Nonce stored in the neighbour table
  entry or the challenge timer has expired, the Challenge Reply MUST be
  silently ignored, and the challenge has failed.

  Otherwise, the node compares the Nonce contained in the Challenge
  Reply with the Nonce contained in the neighbour table entry.  If the
  two are equal (they have the same length and content), then the
  challenge has succeeded and the nonce stored in the neighbour table
  for this neighbour SHOULD be discarded; otherwise, the challenge has
  failed (and the nonce is not discarded).

4.4.  Expiring Per-Neighbour State

  The per-neighbour (Index, PC) pair is maintained in the neighbour
  table, and is normally discarded when the neighbour table entry
  expires.  Implementations MUST ensure that an (Index, PC) pair is
  discarded within a finite time since the last time a packet has been
  accepted.  In particular, unsuccessful challenges MUST NOT prevent an
  (Index, PC) pair from being discarded for unbounded periods of time.

  A possible implementation strategy for implementations that use a
  Hello history (Appendix A of [RFC8966]) is to discard the (Index, PC)
  pair whenever the Hello history becomes empty.  Another
  implementation strategy is to use a timer that is reset whenever a
  packet is accepted and to discard the (Index, PC) pair whenever the
  timer expires.  If the latter strategy is used, the timer SHOULD
  default to a value of 5 minutes and MAY be configurable.

5.  Incremental Deployment and Key Rotation

  In order to perform incremental deployment, the nodes in the network
  are first configured in a mode where packets are sent with
  authentication but not checked on reception.  Once all the nodes in
  the network are configured to send authenticated packets, nodes are
  reconfigured to reject unauthenticated packets.

  In order to perform key rotation, the new key is added to all the
  nodes.  Once this is done, both the old and the new key are sent in
  all packets, and packets are accepted if they are properly signed by
  either of the keys.  At that point, the old key is removed.

  In order to support the procedures described above, implementations
  of this protocol SHOULD support an interface configuration in which
  packets are sent authenticated but received packets are accepted
  without verification, and they SHOULD allow changing the set of keys
  associated with an interface without a restart.

6.  Packet Format

6.1.  MAC TLV

   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 = 16   |    Length     |     MAC...
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-

  Fields:

  Type      Set to 16 to indicate a MAC TLV.

  Length    The length of the body, in octets, exclusive of the Type
            and Length fields.  The length depends on the MAC algorithm
            being used.

  MAC       The body contains the MAC of the packet, computed as
            described in Section 4.1.

  This TLV is allowed in the packet trailer (see Section 4.2 of
  [RFC8966]) and MUST be ignored if it is found in the packet body.

6.2.  PC TLV

   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 = 17   |    Length     |             PC                |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  |                               |            Index...
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-

  Fields:

  Type      Set to 17 to indicate a PC TLV.

  Length    The length of the body, in octets, exclusive of the Type
            and Length fields.

  PC        The Packet Counter (PC), a 32-bit (4-octet) unsigned
            integer that is increased with every packet sent over this
            interface.  A fresh index (as defined in Section 3.1) MUST
            be generated whenever the PC overflows.

  Index     The sender's Index, an opaque string of 0 to 32 octets.

  Indices are limited to a size of 32 octets: a node MUST NOT send a
  TLV with an index of size strictly larger than 32 octets, and a node
  MAY ignore a PC TLV with an index of length strictly larger than 32
  octets.  Indices of length 0 are valid: if a node has reliable stable
  storage and the packet counter never overflows, then only one index
  is necessary, and the value of length 0 is the canonical choice.

6.3.  Challenge Request TLV

   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 = 18   |    Length     |     Nonce...
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-

  Fields:

  Type      Set to 18 to indicate a Challenge Request TLV.

  Length    The length of the body, in octets, exclusive of the Type
            and Length fields.

  Nonce     The nonce uniquely identifying the challenge, an opaque
            string of 0 to 192 octets.

  Nonces are limited to a size of 192 octets: a node MUST NOT send a
  Challenge Request TLV with a nonce of size strictly larger than 192
  octets, and a node MAY ignore a nonce that is of size strictly larger
  than 192 octets.  Nonces of length 0 are valid: if a node has
  reliable stable storage, then it may use a sequential counter for
  generating nonces that get encoded in the minimum number of octets
  required; the value 0 is then encoded as the string of length 0.

6.4.  Challenge Reply TLV

   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 = 19   |    Length     |     Nonce...
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-

  Fields:

  Type      Set to 19 to indicate a Challenge Reply TLV.

  Length    The length of the body, in octets, exclusive of the Type
            and Length fields.

  Nonce     A copy of the nonce contained in the corresponding
            Challenge Request.

7.  Security Considerations

  This document defines a mechanism that provides basic security
  properties for the Babel routing protocol.  The scope of this
  protocol is strictly limited: it only provides authentication (we
  assume that routing information is not confidential), it only
  supports symmetric keying, and it only allows for the use of a small
  number of symmetric keys on every link.  Deployments that need more
  features, e.g., confidentiality or asymmetric keying, should use a
  more feature-rich security mechanism such as the one described in
  [RFC8968].

  This mechanism relies on two assumptions, as described in
  Section 1.2.  First, it assumes that the MAC being used is
  invulnerable to forgery (Section 1.1 of [RFC6039]); at the time of
  writing, HMAC-SHA256, which is mandatory to implement (Section 4.1),
  is believed to be safe against practical attacks.

  Second, it assumes that indices and nonces are generated uniquely
  over the lifetime of a key used for MAC computation (more precisely,
  indices must be unique for a given (key, source) pair, and nonces
  must be unique for a given (key, source, destination) triple).  This
  property can be satisfied either by using a cryptographically secure
  random number generator to generate indices and nonces that contain
  enough entropy (64-bit values are believed to be large enough for all
  practical applications) or by using a reliably monotonic hardware
  clock.  If uniqueness cannot be guaranteed (e.g., because a hardware
  clock has been reset), then rekeying is necessary.

  The expiry mechanism mandated in Section 4.4 is required to prevent
  an attacker from delaying an authentic packet by an unbounded amount
  of time.  If an attacker is able to delay the delivery of a packet
  (e.g., because it is located at a Layer 2 switch), then the packet
  will be accepted as long as the corresponding (Index, PC) pair is
  present at the receiver.  If the attacker is able to cause the
  (Index, PC) pair to persist for arbitrary amounts of time (e.g., by
  repeatedly causing failed challenges), then it is able to delay the
  packet by arbitrary amounts of time, even after the sender has left
  the network, which could allow it to redirect or blackhole traffic to
  destinations previously advertised by the sender.

  This protocol exposes large numbers of packets and their MACs to an
  attacker that is able to capture packets; it is therefore vulnerable
  to brute-force attacks.  Keys must be chosen in a manner that makes
  them difficult to guess.  Ideally, they should have a length of 32
  octets (both for HMAC-SHA256 and BLAKE2s), and be chosen randomly.
  If, for some reason, it is necessary to derive keys from a human-
  readable passphrase, it is recommended to use a key derivation
  function that hampers dictionary attacks, such as PBKDF2 [RFC8018],
  bcrypt [BCRYPT], or scrypt [RFC7914].  In that case, only the derived
  keys should be communicated to the routers; the original passphrase
  itself should be kept on the host used to perform the key generation
  (e.g., an administrator's secure laptop computer).

  While it is probably not possible to be immune against denial of
  service (DoS) attacks in general, this protocol includes a number of
  mechanisms designed to mitigate such attacks.  In particular,
  reception of a packet with no correct MAC creates no local Babel
  state (Section 4.3).  Reception of a replayed packet with correct
  MAC, on the other hand, causes a challenge to be sent; this is
  mitigated somewhat by requiring that challenges be rate limited
  (Section 4.3.1.1).

  Receiving a replayed packet with an obsolete index causes an entry to
  be created in the neighbour table, which, at first sight, makes the
  protocol susceptible to resource exhaustion attacks (similarly to the
  familiar "TCP SYN Flooding" attack [RFC4987]).  However, the MAC
  computation includes the sender address (Section 4.1), and thus the
  amount of storage that an attacker can force a node to consume is
  limited by the number of distinct source addresses used with a single
  MAC key (see also Section 4 of [RFC8966], which mandates that the
  source address is a link-local IPv6 address or a local IPv4 address).

  In order to make this kind of resource exhaustion attacks less
  effective, implementations may use a separate table of uncompleted
  challenges that is separate from the neighbour table used by the core
  protocol (the data structures described in Section 3.2 of [RFC8966]
  are conceptual, and any data structure that yields the same result
  may be used).  Implementers might also consider using the fact that
  the nonces included in Challenge Requests and Replies can be fairly
  large (up to 192 octets), which should in principle allow encoding
  the per-challenge state as a secure "cookie" within the nonce itself;
  note, however, that any such scheme will need to prevent cookie
  replay.

8.  IANA Considerations

  IANA has allocated the following values in the Babel TLV Types
  registry:

                +======+===================+===========+
                | Type | Name              | Reference |
                +======+===================+===========+
                | 16   | MAC               | RFC 8967  |
                +------+-------------------+-----------+
                | 17   | PC                | RFC 8967  |
                +------+-------------------+-----------+
                | 18   | Challenge Request | RFC 8967  |
                +------+-------------------+-----------+
                | 19   | Challenge Reply   | RFC 8967  |
                +------+-------------------+-----------+

                                Table 1

9.  References

9.1.  Normative References

  [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>.

  [RFC6234]  Eastlake 3rd, D. and T. Hansen, "US Secure Hash Algorithms
             (SHA and SHA-based HMAC and HKDF)", RFC 6234,
             DOI 10.17487/RFC6234, May 2011,
             <https://www.rfc-editor.org/info/rfc6234>.

  [RFC7693]  Saarinen, M-J., Ed. and J-P. Aumasson, "The BLAKE2
             Cryptographic Hash and Message Authentication Code (MAC)",
             RFC 7693, DOI 10.17487/RFC7693, November 2015,
             <https://www.rfc-editor.org/info/rfc7693>.

  [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>.

  [RFC8966]  Chroboczek, J. and D. Schinazi, "The Babel Routing
             Protocol", RFC 8966, DOI 10.17487/RFC8966, January 2021,
             <https://www.rfc-editor.org/info/rfc8966>.

9.2.  Informational References

  [BCRYPT]   Niels, P. and D. Mazières, "A Future-Adaptable Password
             Scheme", Proceedings of the FREENIX Track: 1999 USENIX
             Annual Technical Conference, June 1999.

  [RFC4086]  Eastlake 3rd, D., Schiller, J., and S. Crocker,
             "Randomness Requirements for Security", BCP 106, RFC 4086,
             DOI 10.17487/RFC4086, June 2005,
             <https://www.rfc-editor.org/info/rfc4086>.

  [RFC4987]  Eddy, W., "TCP SYN Flooding Attacks and Common
             Mitigations", RFC 4987, DOI 10.17487/RFC4987, August 2007,
             <https://www.rfc-editor.org/info/rfc4987>.

  [RFC6039]  Manral, V., Bhatia, M., Jaeggli, J., and R. White, "Issues
             with Existing Cryptographic Protection Methods for Routing
             Protocols", RFC 6039, DOI 10.17487/RFC6039, October 2010,
             <https://www.rfc-editor.org/info/rfc6039>.

  [RFC7298]  Ovsienko, D., "Babel Hashed Message Authentication Code
             (HMAC) Cryptographic Authentication", RFC 7298,
             DOI 10.17487/RFC7298, July 2014,
             <https://www.rfc-editor.org/info/rfc7298>.

  [RFC7914]  Percival, C. and S. Josefsson, "The scrypt Password-Based
             Key Derivation Function", RFC 7914, DOI 10.17487/RFC7914,
             August 2016, <https://www.rfc-editor.org/info/rfc7914>.

  [RFC8018]  Moriarty, K., Ed., Kaliski, B., and A. Rusch, "PKCS #5:
             Password-Based Cryptography Specification Version 2.1",
             RFC 8018, DOI 10.17487/RFC8018, January 2017,
             <https://www.rfc-editor.org/info/rfc8018>.

  [RFC8968]  Décimo, A., Schinazi, D., and J. Chroboczek, "Babel
             Routing Protocol over Datagram Transport Layer Security",
             RFC 8968, DOI 10.17487/RFC8968, January 2021,
             <https://www.rfc-editor.org/info/rfc8968>.

Acknowledgments

  The protocol described in this document is based on the original HMAC
  protocol defined by Denis Ovsienko [RFC7298].  The use of a pseudo-
  header was suggested by David Schinazi.  The use of an index to avoid
  replay was suggested by Markus Stenberg.  The authors are also
  indebted to Antonin Décimo, Donald Eastlake, Toke Høiland-Jørgensen,
  Florian Horn, Benjamin Kaduk, Dave Taht, and Martin Vigoureux.

Authors' Addresses

  Clara Dô
  IRIF, University of Paris-Diderot
  75205 Paris CEDEX 13
  France

  Email: [email protected]


  Weronika Kolodziejak
  IRIF, University of Paris-Diderot
  75205 Paris CEDEX 13
  France

  Email: [email protected]


  Juliusz Chroboczek
  IRIF, University of Paris-Diderot
  Case 7014
  75205 Paris CEDEX 13
  France

  Email: [email protected]