Internet Engineering Task Force (IETF)                   P. Thubert, Ed.
Request for Comments: 8138                                         Cisco
Category: Standards Track                                     C. Bormann
ISSN: 2070-1721                                           Uni Bremen TZI
                                                              L. Toutain
                                                    IMT-TELECOM Bretagne
                                                          IMT Atlantique
                                                               R. Cragie
                                                                     ARM
                                                              March
                                                              April 2017

  IPv6 over Low-Power Wireless Personal Area Network (6LoWPAN) Routing
                                 Header

Abstract

   This specification introduces a new IPv6 over Low-Power Wireless
   Personal Area Network (6LoWPAN) dispatch type for use in 6LoWPAN
   route-over topologies, which initially covers the needs of Routing
   Protocol for Low-Power and Lossy Networks (RPL) data packet
   compression (RFC 6550).  Using this dispatch type, this specification
   defines a method to compress the RPL Option (RFC 6553) information
   and Routing Header type 3 (RFC 6554), an efficient IP-in-IP
   technique, and is extensible for more applications.

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
   http://www.rfc-editor.org/info/rfc8138.

Copyright Notice

   Copyright (c) 2017 IETF Trust and the persons identified as the
   document authors.  All rights reserved.

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

Table of Contents

   1.  Introduction  . . . . . . . . . . . . . . . . . . . . . . . .   3
   2.  Terminology . . . . . . . . . . . . . . . . . . . . . . . . .   6
   3.  Using the Page Dispatch . . . . . . . . . . . . . . . . . . .   6
     3.1.  New Routing Header Dispatch (6LoRH) . . . . . . . . . . .   6
     3.2.  Placement of 6LoRH Headers  . . . . . . . . . . . . . . .   7
       3.2.1.  Relative to Non-6LoRH Headers . . . . . . . . . . . .   7
       3.2.2.  Relative to Other 6LoRH Headers . . . . . . . . . . .   7
   4.  6LoWPAN Routing Header General Format . . . . . . . . . . . .   8
     4.1.  Elective Format . . . . . . . . . . . . . . . . . . . . .   9
     4.2.  Critical Format . . . . . . . . . . . . . . . . . . . . .   9
     4.3.  Compressing Addresses . . . . . . . . . . . . . . . . . .  10
       4.3.1.  Coalescence . . . . . . . . . . . . . . . . . . . . .  10
       4.3.2.  DODAG Root Address Determination  . . . . . . . . . .  11
   5.  The SRH-6LoRH Header  . . . . . . . . . . . . . . . . . . . .  12
     5.1.  Encoding  . . . . . . . . . . . . . . . . . . . . . . . .  12
     5.2.  SRH-6LoRH General Operation . . . . . . . . . . . . . . .  13
       5.2.1.  Uncompressed SRH Operation  . . . . . . . . . . . . .  13
       5.2.2.  6LoRH-Compressed SRH Operation  . . . . . . . . . . .  14
       5.2.3.  Inner LOWPAN_IPHC Compression . . . . . . . . . . . .  14
     5.3.  The Design Point of Popping Entries . . . . . . . . . . .  15
     5.4.  Compression Reference for SRH-6LoRH Header Entries  . . .  16
     5.5.  Popping Headers . . . . . . . . . . . . . . . . . . . . .  17
     5.6.  Forwarding  . . . . . . . . . . . . . . . . . . . . . . .  18
   6.  The RPL Packet Information 6LoRH  . . . . . . (RPI-6LoRH)  . . . . . . . .  18
     6.1.  Compressing the RPLInstanceID . . . . . . . . . . . . . .  20
     6.2.  Compressing the SenderRank  . . . . . . . . . . . . . . .  20
     6.3.  The Overall RPI-6LoRH Encoding  . . . . . . . . . . . . .  20
   7.  The IP-in-IP 6LoRH Header . . . . . . . . . . . . . . . . . .  23
   8.  Management Considerations . . . . . . . . . . . . . . . . . .  24
   9.  Security Considerations . . . . . . . . . . . . . . . . . . .  26
   10. IANA Considerations . . . . . . . . . . . . . . . . . . . . .  26
     10.1.  Reserving Space in 6LoWPAN Dispatch Page 1 . . . . . . .  26
     10.2.  New Critical 6LoWPAN Routing Header Type Registry  . . .  27
     10.3.  New Elective 6LoWPAN Routing Header Type Registry  . . .  27
   11. References  . . . . . . . . . . . . . . . . . . . . . . . . .  27
     11.1.  Normative References . . . . . . . . . . . . . . . . . .  27
     11.2.  Informative References . . . . . . . . . . . . . . . . .  28
   Appendix A.  Examples . . . . . . . . . . . . . . . . . . . . . .  29
     A.1.  Examples Compressing the RPI  . . . . . . . . . . . . . .  29
     A.2.  Example of a Downward Packet in Non-Storing Mode  . . . .  31
     A.3.  Example of SRH-6LoRH Life Cycle . . . . . . . . . . . . .  33
   Acknowledgements  . . . . . . . . . . . . . . . . . . . . . . . .  35
   Authors' Addresses  . . . . . . . . . . . . . . . . . . . . . . .  35

1.  Introduction

   The design of Low-Power and Lossy Networks (LLNs) is generally
   focused on saving energy, a very constrained resource in most cases.
   The other constraints, such as the memory capacity and the duty
   cycling of the LLN devices, derive from that primary concern.  Energy
   is often available from primary batteries that are expected to last
   for years, or it is scavenged from the environment in very limited
   quantities.  Any protocol that is intended for use in LLNs must be
   designed with the primary concern of saving energy as a strict
   requirement.

   Controlling the amount of data transmission is one possible venue to
   save energy.  In a number of LLN standards, the frame size is limited
   to much smaller values than the guaranteed IPv6 Maximum Transmission
   Unit (MTU) of 1280 bytes.  In particular, an LLN that relies on the
   classical Physical Layer (PHY) of IEEE 802.15.4 [IEEE.802.15.4] is
   limited to 127 bytes per frame.  The need to compress IPv6 packets
   over IEEE 802.15.4 led to the writing of "Compression Format for IPv6
   Datagrams over IEEE 802.15.4-Based Networks" [RFC6282].

   Innovative route-over techniques have been and still are being
   developed for routing inside an LLN.  Generally, such techniques
   require additional information in the packet to provide loop
   prevention and to indicate information such as flow identification,
   source routing information, etc.

   For reasons such as security and the capability to send ICMPv6 errors
   (see "Internet Control Message Protocol (ICMPv6) for the Internet
   Protocol Version 6 (IPv6) Specification" [RFC4443]) back to the
   source, an original packet must not be tampered with, and any
   information that must be inserted in or removed from an IPv6 packet
   must be placed in an extra IP-in-IP encapsulation.

   This is the case when the additional routing information is inserted
   by a router on the path of a packet, for instance, the root of a
   mesh, as opposed to the source node, with the Non-Storing mode of the
   "RPL: IPv6 Routing Protocol for Low-Power and Lossy Networks"
   [RFC6550].

   This is also the case when some routing information must be removed
   from a packet that flows outside the LLN.

   "When to use RFC 6553, RFC 6554 and IPv6-in-IPv6" [RPL-INFO] details
   different cases where IPv6 headers defined in the RPL Option for
   Carrying RPL Information in Data-Plane Datagrams [RFC6553], Header
   for Source Routes with RPL [RFC6554], and IPv6-in-IPv6 encapsulation,
   are inserted or removed from packets in LLN environments operating
   RPL.

   When using RFC 6282 [RFC6282], the outer IP header of an IP-in-IP
   encapsulation may be compressed down to 2 octets in stateless
   compression and down to 3 octets in stateful compression when context
   information must be added.

      0                                       1
      0   1   2   3   4   5   6   7   8   9   0   1   2   3   4   5
    +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+
    | 0 | 1 | 1 |  TF   |NH | HLIM  |CID|SAC|  SAM  | M |DAC|  DAM  |
    +---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+

              Figure 1: LOWPAN_IPHC Base Encoding (RFC 6282)

   The stateless compression of an IPv6 address can only happen if the
   IPv6 address can de deduced from the Media Access Control (MAC)
   addresses, meaning that the IP endpoint is also the MAC-layer
   endpoint.  This is usually not the case in a RPL network, which is
   generally a multi-hop route-over (i.e., operated at Layer 3) network.
   A better compression, which does not involve variable compressions
   depending on the hop in the mesh, can be achieved based on the fact
   that the outer encapsulation is usually between the source (or
   destination) of the inner packet and the root.  Also, the inner IP
   header can only be compressed by RFC 6282 [RFC6282] if all the fields
   preceding it are also compressed.  This specification makes the inner
   IP header the first header to be compressed by RFC 6282 [RFC6282],
   and it keeps the inner packet encoded the same way whether or not it
   is encapsulated, thus preserving existing implementations.

   As an example, RPL [RFC6550] is designed to optimize the routing
   operations in constrained LLNs.  As part of this optimization, RPL
   requires the addition of RPL Packet Information (RPI) in every
   packet, as defined in Section 11.2 of RFC 6550 [RFC6550].

   "The Routing Protocol for Low-Power and Lossy Networks (RPL) Option
   for Carrying RPL Information in Data-Plane Datagrams" [RFC6553]
   specification indicates how the RPI can be placed in a RPL Option
   (RPL-OPT) that is placed in an IPv6 Hop-by-Hop header.

   This representation demands a total of 8 bytes, while, in most cases,
   the actual RPI payload requires only 19 bits.  Since the Hop-by-Hop
   header must not flow outside of the RPL domain, it must be inserted
   in packets entering the domain and be removed from packets that leave
   the domain.  In both cases, this operation implies an IP-in-IP
   encapsulation.

   Additionally, in the case of the Non-Storing Mode of Operation (MOP),
   RPL requires a Source Routing Header (SRH) in all packets that are
   routed down a RPL graph.  For that purpose, "An IPv6 Routing Header
   for Source Routes with the Routing Protocol for Low-Power and Lossy
   Networks (RPL)" [RFC6554] defines the type 3 Routing Header for IPv6
   (RH3).

          ------+---------                           ^
                |          Internet                  |
                |                                    | Native IPv6
             +-----+                                 |
             |     | Border Router (RPL Root)      + | +
             |     |                               | | |
             +-----+                               | | | tunneled
                |                                  | | | using
          o    o   o    o                          | | | IPv6-in-
      o o   o  o   o  o  o o   o                   | | | IPv6 and
     o  o o  o o    o   o   o  o  o                | | | RPL SRH
     o   o    o  o     o  o    o  o  o             | | |
    o  o   o  o   o         o   o o                | | |
    o          o             o     o               + v +
                      LLN

              Figure 2: IP-in-IP Encapsulation within the LLN

   With Non-Storing RPL, even if the source is a node in the same LLN,
   the packet must first reach up the graph to the root so that the root
   can insert the SRH to go down the graph.  In any fashion, whether the
   packet was originated in a node in the LLN or outside the LLN, and
   regardless of whether or not the packet stays within the LLN, as long
   as the source of the packet is not the root itself, the source-
   routing operation also implies an IP-in-IP encapsulation at the root
   in order to insert the SRH.

   "An Architecture for IPv6 over the TSCH mode of IEEE 802.15.4"
   [IPv6-ARCH] specifies the operation of IPv6 over the mode of
   operation described in "Using IEEE 802.15.4e Time-Slotted Channel
   Hopping (TSCH) in the Internet of Things (IoT): Problem Statement"
   [RFC7554].  The architecture requires the use of both RPL and the 6lo
   adaptation layer over IEEE 802.15.4.  Because it inherits the
   constraints on frame size from the MAC layer, 6TiSCH cannot afford to
   allocate 8 bytes per packet on the RPI, hence the requirement for
   6LoWPAN header compression of the RPI.

   An extensible compression technique is required that simplifies IP-
   in-IP encapsulation when it is needed and optimally compresses
   existing routing artifacts found in RPL LLNs.

   This specification extends the 6lo adaptation layer framework
   ([RFC4944] [RFC6282]) so as to carry routing information for route-
   over networks based on RPL.  It includes the formats necessary for
   RPL and is extensible for additional formats.

2.  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 RFC
   2119 [RFC2119].

   This document uses the terms from, and is consistent with, "Terms
   Used in Routing for Low-Power and Lossy Networks" [RFC7102] and RPL
   [RFC6550].

   The terms "route-over" and "mesh-under" are defined in RFC 6775
   [RFC6775].

   Other terms in use in LLNs are found in "Terminology for Constrained-
   Node Networks" [RFC7228].

   The term "byte" is used in its now customary sense as a synonym for
   "octet".

3.  Using the Page Dispatch

   The "IPv6 over Low-Power Wireless Personal Area Network (6LoWPAN)
   Paging Dispatch" [RFC8025] specification extends the 6lo adaptation
   layer framework ([RFC4944] [RFC6282]) by introducing a concept of
   "context" in the 6LoWPAN parser, a context being identified by a Page
   number.  The specification defines 16 Pages.

   This document operates within Page 1, which is indicated by a
   dispatch value of binary 11110001.

3.1.  New Routing Header Dispatch (6LoRH)

   This specification introduces a new 6LoWPAN Routing Header (6LoRH) to
   carry IPv6 routing information.  The 6LoRH may contain source routing
   information such as a compressed form of SRH, as well as other sorts
   of routing information such as the RPI and IP-in-IP encapsulation.

   The 6LoRH is expressed in a 6loWPAN packet as a Type-Length-Value
   (TLV) field, which is extensible for future use.

   It is expected that a router that does not recognize the 6LoRH
   general format detailed in Section 4 will drop the packet when a
   6LoRH is present.

   This specification uses the bit pattern 10xxxxxx in Page 1 for the
   new 6LoRH Dispatch.  Section 4 describes how RPL artifacts in data
   packets can be compressed as 6LoRH headers.

3.2.  Placement of 6LoRH Headers

3.2.1.  Relative to Non-6LoRH Headers

   In a zone of a packet where Page 1 is active (that is, once the Page
   1 Paging Dispatch is parsed, and until another Paging Dispatch is
   parsed as described in the 6LoWPAN Paging Dispatch specification
   [RFC8025]), the parsing of the packet MUST follow this specification
   if the 6LoRH Bit Pattern (see Section 3.1) is found.

   With this specification, the 6LoRH Dispatch is only defined in
   context is active.

   Because a 6LoRH header requires a Page 1 context, it MUST always be
   placed after any Fragmentation Header and/or Mesh Header as defined
   in RFC 4944 [RFC4944].

   A 6LoRH header MUST always be placed before the LOWPAN_IPHC as
   defined in RFC 6282 [RFC6282].  It is designed in such a fashion that
   placing or removing a header that is encoded with 6LoRH does not
   modify the part of the packet that is encoded with LOWPAN_IPHC,
   whether or not there is an IP-in-IP encapsulation.  For instance, the
   final destination of the packet is always the one in the LOWPAN_IPHC,
   whether or not there is a Routing Header.

3.2.2.  Relative to Other 6LoRH Headers

   The "Internet Protocol, Version 6 (IPv6) Specification" [RFC2460]
   defines chains of headers that are introduced by an IPv6 header and
   terminated by either another IPv6 header (IP-in-IP) or an Upper-Layer
   Protocol (ULP) header.  When an outer header is stripped from the
   packet, the whole chain goes with it.  When one or more headers are
   inserted by an intermediate router, that router normally chains the
   headers and encapsulates the result in IP-in-IP.

   With this specification, the chains of headers MUST be compressed in
   the same order as they appear in the uncompressed form of the packet.
   This means that if there is more than one nested IP-in-IP
   encapsulation, the first IP-in-IP encapsulation, with all its chain
   of headers, is encoded first in the compressed form.

   In the compressed form of a packet that has a Source Route or a Hop-
   by-Hop (HbH) Options Header [RFC2460] after the inner IPv6 header
   (e.g., if there is no IP-in-IP encapsulation), these headers are
   placed in the 6LoRH form before the 6LOWPAN_IPHC that represents the
   IPv6 header (see Section 3.2.1).  If this packet gets encapsulated
   and some other SRH or HbH Options Headers are added as part of the
   encapsulation, placing the 6LoRH headers next to one another may
   present an ambiguity on which header belongs to which chain in the
   uncompressed form.

   In order to disambiguate the headers that follow the inner IPv6
   header in the uncompressed form from the headers that follow the
   outer IP-in-IP header, it is REQUIRED that the compressed IP-in-IP
   header is placed last in the encoded chain.  This means that the
   6LoRH headers that are found after the last compressed IP-in-IP
   header are to be inserted after the IPv6 header that is encoded with
   the 6LOWPAN_IPHC when decompressing the packet.

   With regard to the relative placement of the SRH and the RPI in the
   compressed form, it is a design point for this specification that the
   SRH entries are consumed as the packet progresses down the LLN (see
   Section 5.3).  In order to make this operation simpler in the
   compressed form, it is REQUIRED that in the compressed form, the
   addresses along the source route path are encoded in the order of the
   path, and that the compressed SRH are placed before the compressed
   RPI.

4.  6LoWPAN Routing Header General Format

   The 6LoRH uses the Dispatch Value Bit Pattern of 10xxxxxx in Page 1.

   The Dispatch Value Bit Pattern is split in two forms of 6LoRH:

      Elective (6LoRHE), which may skipped if not understood

      Critical (6LoRHC), which may not be ignored

   For each form, a Type field is used to encode the type of 6LoRH.

   Note that there is a different registry for the Type field of each
   form of 6LoRH.

   This means that a value for the Type that is defined for one form of
   6LoRH may be redefined in the future for the other form.

4.1.  Elective Format

   The 6LoRHE uses the Dispatch Value Bit Pattern of 101xxxxx.  A 6LoRHE
   may be ignored and skipped in parsing.  If it is ignored, the 6LoRHE
   is forwarded with no change inside the LLN.

       0                   1
       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-       ...        -+
      |1|0|1| Length  |      Type     |                    |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-       ...        -+
                                       <--    Length    -->

                 Figure 3: Elective 6LoWPAN Routing Header

   Length:  Length of the 6LoRHE expressed in bytes, excluding the first
         2 bytes.  This enables a node to skip a 6LoRHE header that it
         does not support and/or cannot parse, for instance, if the Type
         is not recognized.

   Type: Type of the 6LoRHE

4.2.  Critical Format

   The 6LoRHC uses the Dispatch Value Bit Pattern of 100xxxxx.

   A node that does not support the 6LoRHC Type MUST silently discard
   the packet.

   Note: A situation where a node receives a message with a Critical
   6LoWPAN Routing Header that it does not understand should not occur
   and is an administrative error, see Section 8.

     0                   1
     0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-              ...               -+
    |1|0|0|   TSE   |      Type     |                                  |
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-              ...               -+
                                     <-- Length implied by Type/TSE -->

                 Figure 4: Critical 6LoWPAN Routing Header

   Type-Specific Extension (TSE):  The meaning depends on the Type,
         which must be known in all of the nodes.  The interpretation of
         the TSE depends on the Type field that follows.  For instance,
         it may be used to transport control bits, the number of
         elements in an array, or the length of the remainder of the
         6LoRHC expressed in a unit other than bytes.

   Type: Type of the 6LoRHC

4.3.  Compressing Addresses

   The general technique used in this document to compress an address is
   first to determine a reference that has a long prefix match with this
   address and then elide that matching piece.  In order to reconstruct
   the compressed address, the receiving node will perform the process
   of coalescence described in Section 4.3.1.

   One possible reference is the root of the RPL Destination-Oriented
   Directed Acyclic Graph (DODAG) that is being traversed.  It is used
   by 6LoRH as the reference to compress an outer IP header in case of
   an IP-in-IP encapsulation.  If the root is the source of the packet,
   this technique allows one to fully elide the source address in the
   compressed form of the IP header.  If the root is not the
   encapsulator, then the Encapsulator Address may still be compressed
   using the root as a reference.  How the address of the root is
   determined is discussed in Section 4.3.2.

   Once the address of the source of the packet is determined, it
   becomes the reference for the compression of the addresses that are
   located in compressed SRH headers that are present inside the IP-in-
   IP encapsulation in the uncompressed form.

4.3.1.  Coalescence

   An IPv6 compressed address is coalesced with a reference address by
   overriding the N rightmost bytes of the reference address with the
   compressed address, where N is the length of the compressed address,
   as indicated by the Type of the SRH-6LoRH header in Figure 7.

   The reference address MAY be a compressed address as well, in which
   case, it MUST be compressed in a form that is of an equal or greater
   length than the address that is being coalesced.

   A compressed address is expanded by coalescing it with a reference
   address.  In the particular case of a Type 4 SRH-6LoRH, the address
   is expressed in full and the coalescence is a complete override as
   illustrated in Figure 5.

   RRRRRRRRRRRRRRRRRRR  A reference address, which may be
                        compressed or not

               CCCCCCC  A compressed address, which may be
                        shorter or the same as the reference

   RRRRRRRRRRRRCCCCCCC  A coalesced address, which may be the
                        same compression as the reference

                      Figure 5: Coalescing Addresses

4.3.2.  DODAG Root Address Determination

   Stateful address compression requires that some state is installed in
   the devices to store the compression information that is elided from
   the packet.  That state is stored in an abstract context table, and
   some form of index is found in the packet to obtain the compression
   information from the context table.

   With RFC 6282 [RFC6282], the state is provided to the stack by the
   6LoWPAN Neighbor Discovery Protocol (NDP) [RFC6775].  NDP exchanges
   the context through the 6LoWPAN Context Option in Router
   Advertisement (RA) messages.  In the compressed form of the packet,
   the context can be signaled in a Context Identifier Extension.

   With this specification, the compression information is provided to
   the stack by RPL, and RPL exchanges it through the DODAGID field in
   the DAG Information Object (DIO) messages, as described in more
   detail below.  In the compressed form of the packet, the context can
   be signaled in by the RPLInstanceID in the RPI.

   With RPL [RFC6550], the address of the DODAG root is known from the
   DODAGID field of the DIO messages.  For a Global Instance, the
   RPLInstanceID that is present in the RPI is enough information to
   identify the DODAG that this node participates with and its
   associated root.  But, for a Local Instance, the address of the root
   MUST be explicit, either in some device configuration or signaled in
   the packet, as the source or the destination address, respectively.

   When implicit, the address of the DODAG root MUST be determined as
   follows:

      If the whole network is a single DODAG, then the root can be well-
      known and does not need to be signaled in the packets.  But, since
      RPL does not expose that property, it can only be known by a
      configuration applied to all nodes.

      Else, the router that encapsulates the packet and compresses it
      with this specification MUST also place an RPI in the packet as
      prescribed by RPL to enable the identification of the DODAG.  The
      RPI must be present even in the case when the router also places
      an SRH header in the packet.

   It is expected that the RPL implementation maintains an abstract
   context table, indexed by Global RPLInstanceID, that provides the
   address of the root of the DODAG that this node participates in for
   that particular RPL Instance.

5.  The SRH-6LoRH Header

5.1.  Encoding

   A Source Routing Header 6LoRH (SRH-6LoRH) header provides a compressed form
   for the SRH, as defined in RFC 6554 [RFC6554], for use by RPL
   routers.

   One or more SRH-6LoRH header(s) MAY be placed in a 6LoWPAN packet.

   If a non-RPL router receives a packet with an SRH-6LoRH header, there
   was a routing or a configuration error (see Section 8).

   The desired reaction for the non-RPL router is to drop the packet packet, as
   opposed to skip skipping the header and forward forwarding the packet.

   The Dispatch Value Bit Pattern for the SRH-6LoRH header indicates it
   is Critical.  Routers that understand the 6LoRH general format
   detailed in Section 4 cannot ignore a 6LoRH header of this type and
   will drop the packet if it is unknown to them.

       0                   1
       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-    -+-    -+ ... +-    -+
      |1|0|0|  Size   |6LoRH Type 0..4| Hop1 | Hop2 |     | HopN |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-    -+-    -+ ... +-    -+

                Where N = Size + 1

                          Figure 6: The SRH-6LoRH

   The 6LoRH Type of an SRH-6LoRH header indicates the compression level
   used for that header.

   The fields following the 6LoRH Type are compressed addresses
   indicating the consecutive hops and are ordered from the first to the
   last hop.

   All the addresses in a given SRH-6LoRH header MUST be compressed in
   an identical fashion, so the Length of the compressed form is the
   same for all.

   In order to get different degrees of compression, multiple
   consecutive SRH-6LoRH headers MUST be used.

   Type 0 means that the address is compressed down to one byte, whereas
   Type 4 means that the address is provided in full in the SRH-6LoRH
   with no compression.  The complete list of Types of SRH-6LoRH and the
   corresponding compression level are provided in Figure 7:

     +-----------+----------------------+
     |   6LoRH   | Length of compressed |
     |   Type    | IPv6 address (bytes) |
     +-----------+----------------------+
     |    0      |       1              |
     |    1      |       2              |
     |    2      |       4              |
     |    3      |       8              |
     |    4      |      16              |
     +-----------+----------------------+

                       Figure 7: The SRH-6LoRH Types

   In the case of an SRH-6LoRH header, the TSE field is used as a Size,
   which encodes the number of hops minus 1; so a Size of 0 means one
   hop, and the maximum that can be encoded is 32 hops.  (If more than
   32 hops need to be expressed, a sequence of SRH-6LoRH elements can be
   employed.)  The result is that the Length, in bytes, of an SRH-6LoRH
   header is:

   2 + Length_of_compressed_IPv6_address * (Size + 1)

5.2.  SRH-6LoRH General Operation

5.2.1.  Uncompressed SRH Operation

   In the non-compressed uncompressed form, when the root generates or forwards a
   packet in Non-Storing mode, it needs to include a Source Routing
   Header [RFC6554] to signal a strict source route path to a final
   destination down the DODAG.

   All the hops along the path, except the first one, are encoded in
   order in the SRH.  The last entry in the SRH is the final
   destination; the destination in the IPv6 header is the first hop
   along the source route path.  The intermediate hops perform a swap
   and the Segments Left field indicates the active entry in the Routing
   Header [RFC2460].

   The current destination of the packet, which is the termination of
   the current segment, is indicated at all times by the destination
   address of the IPv6 header.

5.2.2.  6LoRH-Compressed SRH Operation

   The handling of the SRH-6LoRH is different: there is no swap, and a
   forwarding router that corresponds to the first entry in the first
   SRH-6LoRH, upon reception of a packet, effectively consumes that
   entry when forwarding.  This means that the size of a compressed
   source-routed packet decreases as the packet progresses along its
   path and that the routing information is lost along the way.  This
   also means that an SRH encoded with 6LoRH is not recoverable and
   cannot be protected.

   When compressed with this specification, all the remaining hops MUST
   be encoded in order in one or more consecutive SRH-6LoRH headers.
   Whether or not there is an SRH-6LoRH header present, the address of
   the final destination is indicated in the LOWPAN_IPHC at all times
   along the path.  Examples of this are provided in Appendix A.

   The current destination (termination of the current segment) for a
   compressed source-routed packet is indicated in the first entry of
   the first SRH-6LoRH.  In strict source routing, that entry MUST match
   an address of the router that receives the packet.

   The last entry in the last SRH-6LoRH is the last router on the way to
   the final destination in the LLN.  This router can be the final
   destination if it is found desirable to carry a whole IP-in-IP
   encapsulation all the way.  Else, it is the RPL parent of the final
   destination, or a router acting at 6LoWPAN Router (6LR) [RFC6775] for
   the destination host, and it is advertising the host as an external
   route to RPL.

   If the SRH-6LoRH header is contained in an IP-in-IP encapsulation,
   the last router removes the whole chain of headers.  Otherwise, it
   removes the SRH-6LoRH header only.

5.2.3.  Inner LOWPAN_IPHC Compression

   6LoWPAN ND [RFC6282] is designed to support more than one IPv6
   address per node and per Interface Identifier (IID); an IID is
   typically derived from a MAC address to optimize the LOWPAN_IPHC
   compression.

   Link-local addresses are compressed with stateless address
   compression (S/DAC=0).  The other addresses are derived from
   different prefixes, and they can be compressed with stateful address
   compression based on a context (S/DAC=1).

   But, stateless compression is only defined for the specific link-
   local prefix as opposed to the prefix in an encapsulating header.
   And with stateful compression, the compression reference is found in
   a context, as opposed to an encapsulating header.

   The result is that, in the case of an IP-in-IP encapsulation, it is
   possible to compress an inner source (respective destination) IP
   address in a LOWPAN_IPHC based on the encapsulating IP header only if
   stateful (context-based) compression is used.  The compression will
   operate only if the IID in the source (respective destination) IP
   address in the outer and inner headers match, which usually means
   that they refer to the same node.  This is encoded as S/DAC = 1 and
   S/AM=11.  It must be noted that the outer destination address that is
   used to compress the inner destination address is the last entry in
   the last SRH-6LoRH header.

5.3.  The Design Point of Popping Entries

   In order to save energy and to optimize the chances of transmission
   success on lossy media, it is a design point for this specification
   that the entries in the SRH that have been used are removed from the
   packet.  This creates a discrepancy from the art of IPv6, where
   Routing Headers are mutable but recoverable.

   With this specification, the packet can be expanded at any hop into a
   valid IPv6 packet, including an SRH, and compressed back.  But the
   packet, as decompressed along the way, will not carry all the
   consumed addresses that packet would have if it had been forwarded in
   the uncompressed form.

   It is noted that:

      The value of keeping the whole RH in an IPv6 header is for the
      receiver to reverse it to use the symmetrical path on the way
      back.

      It is generally not a good idea to reverse a Routing Header.  The
      RH may have been used to stay away from the shortest path for some
      reason that is only valid on the way in (segment routing).

      There is no use in reversing an RH in the present RPL
      specifications.

      Point-to-Point (P2P) RPL reverses a path that was learned
      reactively as a part of the protocol operation, which is probably
      a cleaner way than a reversed echo on the data path.

      Reversing a header is discouraged (by RFC 2460 [RFC2460]) for
      Redirected Header Option (RHO) unless it is authenticated, which
      requires an Authentication Header (AH).  There is no definition of
      an AH operation for SRH, and there is no indication that the need
      exists in LLNs.

      AH does not protect the RH on the way.  AH is a validation at the
      receiver with the sole value of enabling the receiver to reverse
      it.

      A RPL domain is usually protected by L2 security, which secures
      both RPL itself and the RH in the packets at every hop.  This is a
      better security than that provided by AH.

   In summary, the benefit of saving energy and lowering the chances of
   loss by sending smaller frames over the LLN are seen as overwhelming
   compared to the value of possibly reversing the header.

5.4.  Compression Reference for SRH-6LoRH Header Entries

   In order to optimize the compression of IP addresses present in the
   SRH headers, this specification requires that the 6LoWPAN layer
   identifies an address that is used as a reference for the
   compression.

   With this specification, the Compression Reference for the first
   address found in an SRH header is the source of the IPv6 packet, and
   then the reference for each subsequent entry is the address of its
   predecessor once it is uncompressed.

   With RPL [RFC6550], an SRH header may only be present in Non-Storing
   mode, and it may only be placed in the packet by the root of the
   DODAG, which must be the source of the resulting IPv6 packet
   [RFC2460].  In this case, the address used as Compression Reference
   is the address of the root.

   The Compression Reference MUST be determined as follows:

      The reference address may be obtained by configuration.  The
      configuration may indicate either the address in full or the
      identifier of a 6LoWPAN Context that carries the address
      [RFC6775], for instance, one of the 16 Context Identifiers used in
      LOWPAN_IPHC [RFC6282].

      Else, if there is no IP-in-IP encapsulation, the source address in
      the IPv6 header that is compressed with LOWPAN_IPHC is the
      reference for the compression.

      Else, if the IP-in-IP compression specified in this document is
      used and the Encapsulator Address is provided, then the
      Encapsulator Address is the reference.

      Else, meaning that the IP-in-IP compression specified in this
      document is used and the encapsulator is implicitly the root, the
      address of the root is the reference.

5.5.  Popping Headers

   Upon reception, the router checks whether the address in the first
   entry of the first SRH-6LoRH is one of its own addresses.  If that is
   the case, the router MUST consume that entry before forwarding, which
   is an action of popping from a stack, where the stack is effectively
   the sequence of entries in consecutive SRH-6LoRH headers.

   Popping an entry of an SRH-6LoRH header is a recursive action
   performed as follows:

      If the Size of the current SRH-6LoRH header is 1 or more, indicating more
      (indicating that there are at least 2 entries in the header, header), the
      router removes the first entry and decrements the Size (by 1).

      Else (meaning by 1.

      If the Size of the current SRH-6LoRH header is 0 (indicating that this
      there is the last only 1 entry in the SRH-6LoRH
      header), if header) and there is no next subsequent
      SRH-6LoRH header after this, then the current SRH-6LoRH is removed.

      Else, if

      If the Size of the current SRH-6LoRH header is 0 and there is a next
      subsequent SRH-6LoRH of a and the Type with a larger or of the subsequent SRH-6LoRH is
      equal value, meaning a to or greater than the Type of the current SRH-6LoRH header
      (indicating the same or lesser compression yielding the same or
      larger compressed forms, forms), then the current SRH-6LoRH is removed.

      Else,

      If the first entry Size of the next current SRH-6LoRH header is popped from 0 and there is a
      subsequent SRH-6LoRH and the
      next Type of the subsequent SRH-6LoRH is
      less the Type of the current SRH-6LoRH header, the first entry of
      the subsequent SRH-6LoRH is removed and coalesced with the first
      entry of this SRH-
      6LoRH. the current SRH-6LoRH.

      At the end of the process, if there is are no more SRH-6LoRH in the
      packet, then the processing node is the last router along the
      source route path.

   An example of this operation is provided in Appendix A.3.

5.6.  Forwarding

   When receiving a packet with an SRH-6LoRH, a router determines the
   IPv6 address of the current segment endpoint.

   If strict source routing is enforced and this router is not the
   segment endpoint for the packet, then this router MUST drop the
   packet.

   If this router is the current segment endpoint, then the router pops
   its address as described in Section 5.5 and continues processing the
   packet.

   If there is still an SRH-6LoRH, then the router determines the new
   segment endpoint and routes the packet towards that endpoint.

   Otherwise, the router uses the destination in the inner IP header to
   forward or accept the packet.

   The segment endpoint of a packet MUST be determined as follows:

      The router first determines the Compression Reference as discussed
      in Section 4.3.1.

      The router then coalesces the Compression Reference with the first
      entry of the first SRH-6LoRH header as discussed in Section 5.4.
      If the type of the SRH-6LoRH header is type Type 4, then the coalescence is a full
      override.

   Since the Compression Reference is an uncompressed address, the
   coalesced IPv6 address is also expressed in the full 128 bits.

6.  The RPL Packet Information 6LoRH (RPI-6LoRH)

   Section 11.2 of the RPL document [RFC6550] specifies the RPL Packet
   Information (RPI) as a set of fields that are placed by RPL routers
   in IP packets to identify the RPL Instance, detect anomalies, and
   trigger corrective actions.

   In particular, the SenderRank, which is the scalar metric computed by
   a specialized Objective Function such as described in RFC 6552
   [RFC6552], indicates the Rank of the sender and is modified at each
   hop.  The SenderRank field is used to validate that the packet
   progresses in the expected direction, either upwards or downwards,
   along the DODAG.

   RPL defines the "The Routing Protocol for Low-Power and Lossy
   Networks (RPL) Option for Carrying RPL Information in Data-Plane
   Datagrams" [RFC6553] to transport the RPI, which is carried in an
   IPv6 Hop-by-Hop Options Header [RFC2460], typically consuming 8 bytes
   per packet.

   With RFC 6553 [RFC6553], the RPL Option is encoded as 6 octets, which
   must be placed in a Hop-by-Hop header that consumes two additional
   octets for a total of 8 octets.  To limit the header's range to just
   the RPL domain, the Hop-by-Hop header must be added to (or removed
   from) packets that cross the border of the RPL domain.

   The 8-byte overhead is detrimental to LLN operation, particularly
   with regard to bandwidth and battery constraints.  These bytes may
   cause a containing frame to grow above maximum frame size, leading to
   Layer 2 or 6LoWPAN [RFC4944] fragmentation, which in turn leads to
   even more energy expenditure and issues discussed in "LLN Fragment
   Forwarding and Recovery" [FORWARD-FRAG].

   An additional overhead comes from the need, in certain cases, to add
   an IP-in-IP encapsulation to carry the Hop-by-Hop header.  This is
   needed when the router that inserts the Hop-by-Hop header is not the
   source of the packet so that an error can be returned to the router.
   This is also the case when a packet originated by a RPL node must be
   stripped from the Hop-by-Hop header to be routed outside the RPL
   domain.

   For that reason, this specification defines an IP-in-IP-6LoRH header
   in Section 7, but it must be noted that removal of a 6LoRH header
   does not require manipulation of the packet in the LOWPAN_IPHC, and
   thus, if the source address in the LOWPAN_IPHC is the node that
   inserted the IP-in-IP-6LoRH header, then this situation alone does
   not mandate an IP-in-IP-6LoRH header.

   Note: It was found that some implementations omit the RPI for packets
   going down the RPL graph in Non-Storing mode, even though RPL
   indicates that the RPI should be placed in the packet.  With this
   specification, the RPI is important to indicate the RPLInstanceID, so
   the RPI should not be omitted.

   As a result, a RPL packet may bear only an RPI-6LoRH header and no
   IP-in-IP-6LoRH header.  In that case, the source and destination of
   the packet are specified by the LOWPAN_IPHC.

   As with RFC 6553 [RFC6553], the fields in the RPI include an 'O', an
   'R', and an 'F' bit, an 8-bit RPLInstanceID (with some internal
   structure), and a 16-bit SenderRank.

   The remainder of this section defines the RPI-6LoRH header, which is
   a Critical 6LoWPAN Routing Header that is designed to transport the
   RPI in 6LoWPAN LLNs.

6.1.  Compressing the RPLInstanceID

   RPL Instances are discussed in Section 5 of the RPL specification
   [RFC6550].  A number of simple use cases do not require more than one
   RPL Instance, and in such cases, the RPL Instance is expected to be
   the Global Instance 0.  A global RPLInstanceID is encoded in a
   RPLInstanceID field as follows:

       0 1 2 3 4 5 6 7
      +-+-+-+-+-+-+-+-+
      |0|     ID      |  Global RPLInstanceID in 0..127
      +-+-+-+-+-+-+-+-+

         Figure 8: RPLInstanceID Field Format for Global Instances

   For the particular case of the Global Instance 0, the RPLInstanceID
   field is all zeros.  This specification allows one the compressor to
   elide a RPLInstanceID field that is all zeros and defines an I flag
   that, when set, signals that the field is elided.

6.2.  Compressing the SenderRank

   The SenderRank is the result of the DAGRank operation on the Rank of
   the sender; here, the DAGRank operation is defined in Section 3.5.1
   of the RPL specification [RFC6550] as:

      DAGRank(rank) = floor(rank/MinHopRankIncrease)

   If MinHopRankIncrease is set to a multiple of 256, the least
   significant eight bits of the SenderRank will be all zeroes; by
   eliding those, the SenderRank can be compressed into a single byte.
   This idea is used in RFC 6550 [RFC6550], by defining
   DEFAULT_MIN_HOP_RANK_INCREASE as 256, and in RFC 6552 [RFC6552],
   which defaults MinHopRankIncrease to DEFAULT_MIN_HOP_RANK_INCREASE.

   This specification allows for the SenderRank to be encoded as either
   1 or 2 bytes and defines a K flag that, when set, signals that a
   single byte is used.

6.3.  The Overall RPI-6LoRH Encoding

   The RPI-6LoRH header provides a compressed form for the RPL RPI.
   Routers that need to forward a packet with a RPI-6LoRH header are
   expected to be RPL routers that support this specification.

   If a non-RPL router receives a packet with a RPI-6LoRH header, there
   was a routing or a configuration error (see Section 8).

   The desired reaction for the non-RPL router is to drop the packet as
   opposed to skip the header and forward the packet, which could end up
   forming loops by reinjecting the packet in the wrong RPL Instance.

   The Dispatch Value Bit Pattern for the SRH-6LoRH header indicates it
   is Critical.  Routers that understand the 6LoRH general format
   detailed in Section 4 cannot ignore a 6LoRH header of this type and
   will drop the packet if it is unknown to them.

   Since the RPI-6LoRH header is a Critical header, the TSE field does
   not need to be a length expressed in bytes.  In that case,  Here, the field is fully
   reused for control bits that encode the O, R, and F flags from the
   RPI, as well as the I and K flags that indicate the compression
   format.

   The Type for the RPI-6LoRH is Type 5.

   The RPI-6LoRH header is immediately followed by the RPLInstanceID
   field, unless that field is fully elided, and then the SenderRank,
   which is either compressed into one byte or fully in-lined as 2
   bytes.  The I and K flags in the RPI-6LoRH header indicate whether
   the RPLInstanceID is elided and/or the SenderRank is compressed.
   Depending on these bits, the Length of the RPI-6LoRH may vary as
   described hereafter.

       0                   1                   2
       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+  ...  -+-+-+
      |1|0|0|O|R|F|I|K| 6LoRH Type=5  |   Compressed fields  |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+  ...  -+-+-+

                  Figure 9: The Generic RPI-6LoRH Format

   O, R, and F bits:  The O, R, and F bits are defined in Section 11.2
         of RFC 6550 [RFC6550].

   I flag:  If it is set, the RPLInstanceID is elided and the
         RPLInstanceID is the Global RPLInstanceID 0.  If it is not set,
         the octet immediately following the type Type field contains the
         RPLInstanceID as specified in Section 5.1 of RFC 6550
         [RFC6550].

   K flag:  If it is set, the SenderRank is compressed into 1 octet,
         with the least significant octet elided.  If it is not set, the
         SenderRank is fully inlined as 2 octets.

   In Figure 10, the RPLInstanceID is the Global RPLInstanceID 0, and
   the MinHopRankIncrease is a multiple of 256, so the least significant
   byte is all zeros and can be elided:

       0                   1                   2
       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |1|0|0|O|R|F|1|1| 6LoRH Type=5  | SenderRank    |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
                I=1, K=1

                 Figure 10: The Most Compressed RPI-6LoRH

   In Figure 11, the RPLInstanceID is the Global RPLInstanceID 0, but
   both bytes of the SenderRank are significant so it cannot be
   compressed:

       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
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |1|0|0|O|R|F|1|0| 6LoRH Type=5  |        SenderRank             |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
                I=1, K=0

                   Figure 11: Eliding the RPLInstanceID

   In Figure 12, the RPLInstanceID is not the Global RPLInstanceID 0,
   and the MinHopRankIncrease is a multiple of 256:

       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
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |1|0|0|O|R|F|0|1| 6LoRH Type=5  | RPLInstanceID |  SenderRank   |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
                I=0, K=1

                     Figure 12: Compressing SenderRank

   In Figure 13, the RPLInstanceID is not the Global RPLInstanceID 0,
   and both bytes of the SenderRank are significant:

       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
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |1|0|0|O|R|F|0|0| 6LoRH Type=5  | RPLInstanceID |    Sender-...
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
        ...-Rank      |
      +-+-+-+-+-+-+-+-+
                I=0, K=0

             Figure 13: The Least Compressed Form of RPI-6LoRH

7.  The IP-in-IP 6LoRH Header

   The IP-in-IP 6LoRH (IP-in-IP-6LoRH) header is an Elective 6LoWPAN
   Routing Header that provides a compressed form for the encapsulating
   IPv6 Header in the case of an IP-in-IP encapsulation.

   An IP-in-IP encapsulation is used to insert a field such as a Routing
   Header or an RPI at a router that is not the source of the packet.
   In order to send an error back regarding the inserted field, the
   address of the router that performs the insertion must be provided.

   The encapsulation can also enable the last router prior to the
   Destination to remove a field such as the RPI, but this can be done
   in the compressed form by removing the RPI-6LoRH, so an IP-in-IP-
   6LoRH encapsulation is not required for that sole purpose.

   The Dispatch Value Bit Pattern for the SRH-6LoRH header indicates it
   is Elective.  This field is not Critical for routing since it does
   not indicate the destination of the packet, which is either encoded
   in an SRH-6LoRH header or in the inner IP header.  A 6LoRH header of
   this type can be skipped if not understood (per Section 4), and the
   6LoRH header indicates the Length in bytes.

     0                   1                   2
     0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-       ...      -+
    |1|0|1| Length  | 6LoRH Type 6  |  Hop Limit    | Encaps. Address  |
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-       ...      -+

                       Figure 14: The IP-in-IP-6LoRH

   The Length of an IP-in-IP-6LoRH header is expressed in bytes and MUST
   be at least 1, to indicate a Hop Limit (HL) that is decremented at
   each hop.  When the HL reaches 0, the packet is dropped per RFC 2460
   [RFC2460].

   If the Length of an IP-in-IP-6LoRH header is exactly 1, then the
   Encapsulator Address is elided, which means that the encapsulator is
   a well-known router, for instance, the root in a RPL graph.

   The most efficient compression of an IP-in-IP encapsulation that can
   be achieved with this specification is obtained when an endpoint of
   the packet is the root of the RPL DODAG associated to the RPL
   Instance that is used to forward the packet, and the root address is
   known implicitly as opposed to signaled explicitly in the data
   packets.

   If the Length of an IP-in-IP-6LoRH header is greater than 1, then an
   Encapsulator Address is placed in a compressed form after the Hop
   Limit field.  The value of the Length indicates which compression is
   performed on the Encapsulator Address.  For instance, a Length of 3
   indicates that the Encapsulator Address is compressed to 2 bytes.
   The reference for the compression is the address of the root of the
   DODAG.  The way the address of the root is determined is discussed in
   Section 4.3.2.

   With RPL, the destination address in the IP-in-IP header is
   implicitly the root in the RPL graph for packets going upwards; in
   Storing mode, it is the destination address in the LOWPAN_IPHC for
   packets going downwards.  In Non-Storing mode, there is no implicit
   value for packets going downwards.

   If the implicit value is correct, the destination IP address of the
   IP-in-IP encapsulation can be elided.  Else, the destination IP
   address of the IP-in-IP header is transported in an SRH-6LoRH header
   as the first entry of the first of these headers.

   If the final destination of the packet is a leaf that does not
   support this specification, then the chain of 6LoRH headers must be
   stripped by the RPL/6LR router to which the leaf is attached.  In
   that example, the destination IP address of the IP-in-IP header
   cannot be elided.

   In the special case where a 6LoRH header is used to route 6LoWPAN
   fragments, the destination address is not accessible in the
   LOWPAN_IPHC on all fragments and can be elided only for the first
   fragment and for packets going upwards.

8.  Management Considerations

   Though it is possible to decompress a packet at any hop, this
   specification is optimized to enable that a packet is forwarded in
   its compressed form all the way, and it makes sense to deploy
   homogeneous networks where all nodes, or no nodes at all, use the
   compression technique detailed therein.

   This specification aims at a simple implementation running in
   constrained nodes, so it does indeed expect a homogeneous network
   and, as a consequence, it does not provide a method to determine the
   level of support by the next hops at forwarding time.

   Should an extension to this specification provide such a method,
   forwarding nodes could compress or uncompress decompress the RPL artifacts
   appropriately and enable a backward compatibility between nodes that
   support this specification and nodes that do not.

   It results that this specification does not attempt to enable such
   backwards compatibility.  It does not require extraneous code to
   exchange and handle error messages to automatically correct mismatch
   situations either.

   When a packet is expected to carry a 6LoRH header but does not, the
   node that discovers the issue is expected to send an ICMPv6 error
   message to the root.  It should be sent at an adapted rate-limitation
   and with a Type type 4 (indicating a "Parameter Problem" error) Problem") and a Code code 0
   (indicating an "Erroneous header "Unrecognized Next Header field encountered" error). encountered").  The
   relevant portion of the received packet should be embedded and the
   offset therein where the 6LoRH header was expected should be pointed
   out.

   When a packet is received with a 6LoRH header that is not recognized,
   the node that discovers the issue is expected to send an ICMPv6 error
   message to the root.  It should be sent at an adapted rate-limitation
   and with a Type type 4 (indicating a "Parameter Problem" error) Problem") and a Code code 1
   (indicating an "Unrecognized Next Header type encountered" error). encountered").  The
   relevant portion of the received packet should be embedded and the
   offset therein where the 6LoRH header was expected should be pointed
   out.

   In both cases, the node SHOULD NOT place a 6LoRH header as defined in
   this specification in the resulting message, and the node should
   either omit the RPI or place it uncompressed after the IPv6 header.

   Additionally, in both cases, an alternate management method may be
   preferred in order to notify the network administrator that there is
   a configuration error.

   Keeping the network homogeneous is either a deployment issue, by
   deploying only devices with a same capability, or a management issue,
   by configuring all devices to either use or not use a certain level
   of this compression technique and its future additions.

   In particular, the situation where a node receives a message with a
   Critical 6LoWPAN Routing Header that it does not understand is an
   administrative error whereby the wrong device is placed in a network,
   or the device is misconfigured.

   When a mismatch situation is detected, it is expected that the device
   raises some management alert indicating the issue, e.g., that it has
   to drop a packet with a Critical 6LoRH.

9.  Security Considerations

   The security considerations of RFC 4944 [RFC4944], RFC 6282
   [RFC6282], and RFC 6553 [RFC6553] apply.

   Using a compressed format as opposed to the full in-line format is
   logically equivalent and is believed not to create an opening for a
   new threat when compared to RFC 6550 [RFC6550], RFC 6553 [RFC6553],
   and RFC 6554 [RFC6554], noting that, even though intermediate hops
   are removed from the SRH header as they are consumed, a node may
   still identify that the rest of the source-routed path includes a
   loop or not (see the "Security" section of RFC 6554).  It must be
   noted that if the attacker is not part of the loop, then there is
   always a node at the beginning of the loop that can detect it and
   remove it.

10.  IANA Considerations

   This specification reserves Dispatch Value Bit Patterns within the
   6LoWPAN Dispatch Page 1 as follows:

      10 1xxxxx: for Elective 6LoWPAN Routing Headers

      10 0xxxxx: for Critical 6LoWPAN Routing Headers

   Additionally, this document creates two IANA registries: one for the
   Critical 6LoWPAN Routing Header Type and one for the Elective 6LoWPAN
   Routing Header Type, each with 256 possible values, from 0 to 255, as
   described below.

   Future assignments are made by IANA using the "RFC Required"
   procedure [RFC5226].

10.2.  New Critical 6LoWPAN Routing Header Type Registry

   This document creates an IANA registry titled "Critical 6LoWPAN
   Routing Header Type" and assigns the following values:

      0-4: SRH-6LoRH [RFC8138]

      5: RPI-6LoRH [RFC8138]

10.3.  New Elective 6LoWPAN Routing Header Type Registry

   This document creates an IANA registry titled "Elective 6LoWPAN
   Routing Header Type" and assigns the following value:

      6: IP-in-IP-6LoRH [RFC8138]

11.  References

11.1.  Normative References

   [IEEE.802.15.4]
              IEEE, "IEEE std. 802.15.4, Part. 15.4: Wireless Medium
              Access Control (MAC) and Physical Layer (PHY)
              Specifications Standard for Low-Rate Wireless Personal Area Networks", 2015.
              IEEE 802.15.4-2015, DOI 10.1109/IEEESTD.2016.7460875,
              <http://ieeexplore.ieee.org/document/7460875/>.

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

   [RFC2460]  Deering, S. and R. Hinden, "Internet Protocol, Version 6
              (IPv6) Specification", RFC 2460, DOI 10.17487/RFC2460,
              December 1998, <http://www.rfc-editor.org/info/rfc2460>.

   [RFC4443]  Conta, A., Deering, S., and M. Gupta, Ed., "Internet
              Control Message Protocol (ICMPv6) for the Internet
              Protocol Version 6 (IPv6) Specification", RFC 4443,
              DOI 10.17487/RFC4443, March 2006,
              <http://www.rfc-editor.org/info/rfc4443>.

   [RFC4944]  Montenegro, G., Kushalnagar, N., Hui, J., and D. Culler,
              "Transmission of IPv6 Packets over IEEE 802.15.4
              Networks", RFC 4944, DOI 10.17487/RFC4944, September 2007,
              <http://www.rfc-editor.org/info/rfc4944>.

   [RFC5226]  Narten, T. and H. Alvestrand, "Guidelines for Writing an
              IANA Considerations Section in RFCs", BCP 26, RFC 5226,
              DOI 10.17487/RFC5226, May 2008,
              <http://www.rfc-editor.org/info/rfc5226>.

   [RFC6282]  Hui, J., Ed. and P. Thubert, "Compression Format for IPv6
              Datagrams over IEEE 802.15.4-Based Networks", RFC 6282,
              DOI 10.17487/RFC6282, September 2011,
              <http://www.rfc-editor.org/info/rfc6282>.

   [RFC6550]  Winter, T., Ed., Thubert, P., Ed., Brandt, A., Hui, J.,
              Kelsey, R., Levis, P., Pister, K., Struik, R., Vasseur,
              JP., and R. Alexander, "RPL: IPv6 Routing Protocol for
              Low-Power and Lossy Networks", RFC 6550,
              DOI 10.17487/RFC6550, March 2012,
              <http://www.rfc-editor.org/info/rfc6550>.

   [RFC6552]  Thubert, P., Ed., "Objective Function Zero for the Routing
              Protocol for Low-Power and Lossy Networks (RPL)",
              RFC 6552, DOI 10.17487/RFC6552, March 2012,
              <http://www.rfc-editor.org/info/rfc6552>.

   [RFC6553]  Hui, J. and JP. Vasseur, "The Routing Protocol for Low-
              Power and Lossy Networks (RPL) Option for Carrying RPL
              Information in Data-Plane Datagrams", RFC 6553,
              DOI 10.17487/RFC6553, March 2012,
              <http://www.rfc-editor.org/info/rfc6553>.

   [RFC6554]  Hui, J., Vasseur, JP., Culler, D., and V. Manral, "An IPv6
              Routing Header for Source Routes with the Routing Protocol
              for Low-Power and Lossy Networks (RPL)", RFC 6554,
              DOI 10.17487/RFC6554, March 2012,
              <http://www.rfc-editor.org/info/rfc6554>.

   [RFC8025]  Thubert, P., Ed. and R. Cragie, "IPv6 over Low-Power
              Wireless Personal Area Network (6LoWPAN) Paging Dispatch",
              RFC 8025, DOI 10.17487/RFC8025, November 2016,
              <http://www.rfc-editor.org/info/rfc8025>.

11.2.  Informative References

   [FORWARD-FRAG]
              Thubert, P., Ed. and J. Hui, "LLN Fragment Forwarding and
              Recovery", Work in Progress, draft-thubert-6lo-forwarding-
              fragments-04, January 2017.

   [IPv6-ARCH]
              Thubert, P., Ed., "An Architecture for IPv6 over the TSCH
              mode of IEEE 802.15.4", Work in Progress, draft-ietf-
              6tisch-architecture-11, January 2017.

   [RFC6775]  Shelby, Z., Ed., Chakrabarti, S., Nordmark, E., and C.
              Bormann, "Neighbor Discovery Optimization for IPv6 over
              Low-Power Wireless Personal Area Networks (6LoWPANs)",
              RFC 6775, DOI 10.17487/RFC6775, November 2012,
              <http://www.rfc-editor.org/info/rfc6775>.

   [RFC7102]  Vasseur, JP., "Terms Used in Routing for Low-Power and
              Lossy Networks", RFC 7102, DOI 10.17487/RFC7102, January
              2014, <http://www.rfc-editor.org/info/rfc7102>.

   [RFC7228]  Bormann, C., Ersue, M., and A. Keranen, "Terminology for
              Constrained-Node Networks", RFC 7228,
              DOI 10.17487/RFC7228, May 2014,
              <http://www.rfc-editor.org/info/rfc7228>.

   [RFC7554]  Watteyne, T., Ed., Palattella, M., and L. Grieco, "Using
              IEEE 802.15.4e Time-Slotted Channel Hopping (TSCH) in the
              Internet of Things (IoT): Problem Statement", RFC 7554,
              DOI 10.17487/RFC7554, May 2015,
              <http://www.rfc-editor.org/info/rfc7554>.

   [RPL-INFO]
              Robles, I., Richardson, M., and P. Thubert, "When to use
              RFC 6553, 6554 and IPv6-in-IPv6", Work in Progress, draft-
              ietf-roll-useofrplinfo-12, March 2017.

Appendix A.  Examples

A.1.  Examples Compressing the RPI

   The example in Figure 15 illustrates the 6LoRH compression of a
   classical packet in Storing mode in all directions, as well as in
   Non-Storing mode for a packet going up the DODAG following the
   default route to the root.  In this particular example, a
   fragmentation process takes place per RFC 4944 [RFC4944], and the
   fragment headers must be placed in Page 0 before switching to Page 1:

   +-  ...  -+-  ...  -+-+ ... -+- ... +-+-+ ... -+-+-+-+-+-+-+-+-+-+...
   |Frag type|Frag hdr |11110001|  RPI-  |IP-in-IP| LOWPAN_IPHC | ...
   |RFC 4944 |RFC 4944 | Page 1 | 6LoRH  | 6LoRH  |             |
   +-  ...  -+-  ...  -+-+ ... -+- ... +-+-+ ... -+-+-+-+-+-+-+-+-+-+...
                                                   <-  RFC 6282  ->
                                                    No RPL artifact

   +-  ...  -+-  ...  -+-+ ... -+-+ ... -+- ... +-+-+-+-+-+-+-+-+-+-+...
   |Frag type|Frag hdr |
   |RFC 4944 |RFC 4944 |  Payload (cont)
   +-  ...  -+-  ...  -+-+ ... -+-+ ... -+- ... +-+-+-+-+-+-+-+-+-+-+...

   +-  ...  -+-  ...  -+-+ ... -+-+ ... -+- ... +-+-+-+-+-+-+-+-+-+-+...
   |Frag type|Frag hdr |
   |RFC 4944 |RFC 4944 |  Payload (cont)
   +-  ...  -+-  ...  -+-+ ... -+-+ ... -+- ... +-+-+-+-+-+-+-+-+-+-+...

               Figure 15: Example Compressed Packet with RPI

   In Storing mode, if the packet stays within the RPL domain, then it
   is possible to save the IP-in-IP encapsulation, in which case, only
   the RPI is compressed with a 6LoRH, as illustrated in Figure 16 in
   the case of a non-fragmented ICMP packet:

   +- ...  -+-+- ... -+-+-+-+ ... -+-+-+-+ ... -+-+-+-+-+-+-+-+-+-+-+...
   |11110001| RPI-6LoRH |  NH = 0      | NH = 58  |  ICMP message ...
   |Page 1  |  type  Type 5   | 6LOWPAN_IPHC | (ICMP)   |  (no compression)
   +- ...  -+-+- ... -+-+-+-+ ... -+-+-+-+ ... -+-+-+-+-+-+-+-+-+-+-+...
                         <-      RFC 6282       ->
                             No RPL artifact

          Figure 16: Example ICMP Packet with RPI in Storing Mode

   The format in Figure 16 is logically equivalent to the non-compressed uncompressed
   format illustrated in Figure 17:

   +-+-+-+- ... -+-+-+-+ ... -+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+...
   |  IPv6 Header  | Hop-by-Hop |  RPI in       |  ICMP message ...
   |  NH = 58      | Header     |  RPL Option   |
   +-+-+-+- ... -+-+-+-+ ... -+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+...

               Figure 17: Uncompressed ICMP Packet with RPI

   For a UDP packet, the transport header can be compressed with 6LoWPAN
   HC [RFC6282] as illustrated in Figure 18:

   +-+ ... -+-+-...-+-+- ... -+-+-+-+ ... -+-+-+ ... -+-+-+-+-+...
   |11110001| RPI-  | NH=1        |11110CPP| Compressed | UDP
   |Page 1  | 6LoRH | LOWPAN_IPHC | UDP    | UDP header | Payload
   +-+ ... -+-+-...-+-+- ... -+-+-+-+ ... -+-+-+ ... -+-+-+-+-+...
                     <-         RFC 6282              ->
                                No RPL artifact

               Figure 18: Uncompressed ICMP Packet with RPI

   If the packet is received from the Internet in Storing mode, then the
   root is supposed to encapsulate the packet to insert the RPI.  The
   resulting format would be as represented in Figure 19:

 +-+ ... -+-+-...-+-+-- ... -+-+-+-+- ... -+-+ ... -+-+-+ ... -+-+-+...
 |11110001| RPI-  | IP-in-IP | NH=1        |11110CPP| Compressed | UDP
 |Page 1  | 6LoRH |  6LoRH   | LOWPAN_IPHC | UDP    | UDP header | Payld
 +-+ ... -+-+-...-+-+-- ... -+-+-+-+- ... -+-+ ... -+-+-+ ... -+-+-+...
                              <-         RFC 6282              ->
                                         No RPL artifact

            Figure 19: RPI Inserted by the Root in Storing Mode

A.2.  Example of a Downward Packet in Non-Storing Mode

   The example illustrated in Figure 20 is a classical packet in Non-
   Storing mode for a packet going down the DODAG following a source-
   routed path from the root.  Say that we have four forwarding hops to
   reach a destination.  In the non-compressed uncompressed form, when the root
   generates the packet, the last 3 hops are encoded in a Routing Header
   type
   Type 3 (SRH) and the first hop is the destination of the packet.  The
   intermediate hops perform a swap; the hop count indicates the current
   active hop as defined in RFC 2460 [RFC2460] and RFC 6554 [RFC6554].

   When compressed with this specification, the 4 hops are encoded in
   SRH-6LoRH when the root generates the packet, and the final
   destination is left in the LOWPAN_IPHC.  There is no swap; the
   forwarding node that corresponds to the first entry effectively
   consumes it when forwarding, which means that the size of the encoded
   packet decreases and that the hop information is lost.

   If the last hop in an SRH-6LoRH is not the final destination, then it
   removes the SRH-6LoRH before forwarding.

   In the particular example illustrated in Figure 20, all addresses in
   the DODAG are assigned from the same /112 prefix and the last 2
   octets encoding an identifier such as an IEEE 802.15.4 short address.
   In that case, all addresses can be compressed to 2 octets, using the
   root address as reference.  There will be one SRH_6LoRH header with,
   in this example, three compressed addresses:

 +-+ ... -+-+ ... +-+- ... -+-+- ... +-+-+-+ ... +-+-+ ... -+ ... +-...
 |11110001|SRH-6LoRH| RPI-  | IP-in-IP | NH=1      |11110CPP| UDP | UDP
 |Page 1  |Type1 S=2| 6LoRH |  6LoRH   |LOWPAN_IPHC| UDP    | hdr |Payld
 +-+ ... -+-+ ... +-+- ... -+-+-- ... -+-+-+ ... +-+-+ ... -+ ... +-...
            <-8bytes->                  <-        RFC 6282      ->
                                                No RPL artifact

               Figure 20: Example Compressed Packet with SRH

   One may note that the RPI is provided.  This is because the address
   of the root that is the source of the IP-in-IP header is elided and
   inferred from the RPLInstanceID in the RPI.  Once found from a local
   context, that address is used as a Compression Reference to expand
   addresses in the SRH-6LoRH.

   With the RPL specifications available at the time of writing, the
   root is the only node that may incorporate an SRH in an IP packet.
   When the root forwards a packet that it did not generate, it has to
   encapsulate the packet with IP-in-IP.

   But, if the root generates the packet towards a node in its DODAG,
   then it should avoid the extra IP-in-IP as illustrated in Figure 21:

   +- ...  -+-+-+ ... +-+-+-+ ... -+-+-+-+-+-+-+-++-+- ... -+-+-+-+-+...
   |11110001| SRH-6LoRH | NH=1       | 11110CPP  | Compressed | UDP
   |Page 1  | Type1 S=3 | LOWPAN_IPHC| LOWPAN-NHC| UDP header | Payload
   +- ...  -+-+-+ ... +-+-+-+ ... -+-+-+-+-+-+-+-++-+- ... -+-+-+-+-+...
                                          <-        RFC 6282        ->

        Figure 21: Compressed SRH 4*2bytes Entries Sourced by Root

   Note: The RPI is not represented, though RPL [RFC6550] generally
   expects it.  In this particular case, since the Compression Reference
   for the SRH-6LoRH is the source address in the LOWPAN_IPHC, and the
   routing is strict along the source route path, the RPI does not
   appear to be absolutely necessary.

   In Figure 21, all the nodes along the source route path share the
   same /112 prefix.  This is typical of IPv6 addresses derived from an
   IEEE802.15.4 short address, as long as all the nodes share the same
   PAN-ID.  In that case, a type Type 1 SRH-6LoRH header can be used for
   encoding.  The IPv6 address of the root is taken as reference, and
   only the last 2 octets of the address of the intermediate hops are
   encoded.  The Size of 3 indicates 4 hops, resulting in an SRH-6LoRH
   of 10 bytes.

A.3.  Example of SRH-6LoRH Life Cycle

   This section illustrates the operation specified in Section 5.6 of
   forwarding a packet with a compressed SRH along an A->B->C->D source
   route path.  The operation of popping addresses is exemplified at
   each hop.

   Packet as received by node A
   ----------------------------
     Type 3 SRH-6LoRH Size = 0   AAAA AAAA AAAA AAAA
     Type 1 SRH-6LoRH Size = 0                  BBBB
     Type 2 SRH-6LoRH Size = 1             CCCC CCCC
                                           DDDD DDDD

    Step 1: Popping BBBB, the first entry of the next SRH-6LoRH
    Step 2: If larger value (2 vs. 1), the SRH-6LoRH is removed

     Type 3 SRH-6LoRH Size = 0   AAAA AAAA AAAA AAAA
     Type 2 SRH-6LoRH Size = 1             CCCC CCCC
                                           DDDD DDDD

    Step 3: Recursion ended; coalescing BBBB with the first entry
     Type 3 SRH-6LoRH Size = 0   AAAA AAAA AAAA BBBB

    Step 4: Routing based on next segment endpoint to B

                      Figure 22: Processing at Node A
   Packet as received by node B
   ----------------------------
     Type 3 SRH-6LoRH Size = 0   AAAA AAAA AAAA BBBB
     Type 2 SRH-6LoRH Size = 1             CCCC CCCC
                                           DDDD DDDD

    Step 1: Popping CCCC CCCC, the first entry of the next SRH-6LoRH
    Step 2: Removing the first entry and decrementing the Size (by 1)

     Type 3 SRH-6LoRH Size = 0   AAAA AAAA AAAA BBBB
     Type 2 SRH-6LoRH Size = 0             DDDD DDDD

    Step 3: Recursion ended; coalescing CCCC CCCC with the first entry
     Type 3 SRH-6LoRH Size = 0   AAAA AAAA CCCC CCCC

    Step 4: Routing based on next segment endpoint to C

                      Figure 23: Processing at Node B

   Packet as received by node C
   ----------------------------

     Type 3 SRH-6LoRH Size = 0   AAAA AAAA CCCC CCCC
     Type 2 SRH-6LoRH Size = 0             DDDD DDDD

    Step 1: Popping DDDD DDDD, the first entry of the next SRH-6LoRH
    Step 2: The SRH-6LoRH is removed

     Type 3 SRH-6LoRH Size = 0   AAAA AAAA CCCC CCCC

    Step 3: Recursion ended; coalescing DDDD DDDDD with the first entry
     Type 3 SRH-6LoRH Size = 0   AAAA AAAA DDDD DDDD

    Step 4: Routing based on next segment endpoint to D

                      Figure 24: Processing at Node C
   Packet as received by node D
   ----------------------------
     Type 3 SRH-6LoRH Size = 0   AAAA AAAA DDDD DDDD

    Step 1: The SRH-6LoRH is removed
    Step 2: No more header; routing based on inner IP header

                      Figure 25: Processing at Node D

Acknowledgements

   The authors wish to thank Tom Phinney, Thomas Watteyne, Tengfei
   Chang, Martin Turon, James Woodyatt, Samita Chakrabarti, Jonathan
   Hui, Gabriel Montenegro, and Ralph Droms for constructive reviews to
   the design in the 6lo working group.  The overall discussion involved
   participants to the 6MAN, 6TiSCH, and ROLL WGs; thank you all.
   Special thanks to Michael Richardson and Ines Robles (the Chairs of
   the ROLL WG), Brian Haberman (the Internet Area AD), and Alvaro
   Retana and Adrian Farrel (Routing Area ADs) for driving this complex
   effort across working groups and areas.

Authors' Addresses

   Pascal Thubert (editor)
   Cisco Systems
   Building D - Regus
   45 Allee des Ormes
   BP1200
   MOUGINS - Sophia Antipolis  06254
   France

   Phone: +33 4 97 23 26 34
   Email: pthubert@cisco.com

   Carsten Bormann
   Universitaet Bremen TZI
   Postfach 330440
   Bremen  D-28359
   Germany

   Phone: +49-421-218-63921
   Email: cabo@tzi.org
   Laurent Toutain
   Institut MINES TELECOM; TELECOM Bretagne
   IMT Atlantique
   2 rue de la Chataigneraie
   CS 17607
   Cesson-Sevigne Cedex  35576
   France

   Email: Laurent.Toutain@telecom-bretagne.eu Laurent.Toutain@IMT-Atlantique.fr

   Robert Cragie
   ARM Ltd.
   110 Fulbourn Road
   Cambridge  CB1 9NJ
   United Kingdom

   Email: robert.cragie@gridmerge.com robert.cragie@arm.com