Difference between revisions of "RFC1122"

From RFC-Wiki
 
(One intermediate revision by the same user not shown)
Line 1: Line 1:
−
 
−
 
−
 
−
 
−
 
−
 
 
Network Working Group                    Internet Engineering Task Force
 
Network Working Group                    Internet Engineering Task Force
 
Request for Comments: 1122                            R. Braden, Editor
 
Request for Comments: 1122                            R. Braden, Editor
−
                                                            October 1989
+
                                                        October 1989
−
 
 
−
 
 
−
        Requirements for Internet Hosts -- Communication Layers
 
  
 +
    Requirements for Internet Hosts -- Communication Layers
  
−
Status of This Memo
+
'''Status of This Memo'''
  
−
  This RFC is an official specification for the Internet community.  It
+
This RFC is an official specification for the Internet community.  It
−
  incorporates by reference, amends, corrects, and supplements the
+
incorporates by reference, amends, corrects, and supplements the
−
  primary protocol standards documents relating to hosts.  Distribution
+
primary protocol standards documents relating to hosts.  Distribution
−
  of this document is unlimited.
+
of this document is unlimited.
  
 
Summary
 
Summary
  
−
  This is one RFC of a pair that defines and discusses the requirements
+
This is one RFC of a pair that defines and discusses the requirements
−
  for Internet host software.  This RFC covers the communications
+
for Internet host software.  This RFC covers the communications
−
  protocol layers: link layer, IP layer, and transport layer; its
+
protocol layers: link layer, IP layer, and transport layer; its
−
  companion RFC-1123 covers the application and support protocols.
+
companion RFC-1123 covers the application and support protocols.
  
 +
                        Table of Contents
  
 +
RFC1122                      INTRODUCTION                  October 1989
  
−
                          Table of Contents
+
RFC1122                      INTRODUCTION                  October 1989
−
 
 
  
 +
RFC1122                      INTRODUCTION                  October 1989
  
 +
RFC1122                      INTRODUCTION                  October 1989
  
−
  1.  INTRODUCTION ...............................................    5
+
== INTRODUCTION ==
−
      1.1  The Internet Architecture ..............................    6
 
−
        1.1.1  Internet Hosts ....................................    6
 
−
        1.1.2  Architectural Assumptions .........................    7
 
−
        1.1.3  Internet Protocol Suite ...........................    8
 
−
        1.1.4  Embedded Gateway Code .............................  10
 
−
      1.2  General Considerations .................................  12
 
−
        1.2.1  Continuing Internet Evolution .....................  12
 
−
        1.2.2  Robustness Principle ..............................  12
 
−
        1.2.3  Error Logging .....................................  13
 
−
        1.2.4  Configuration .....................................  14
 
−
      1.3  Reading this Document ..................................  15
 
−
        1.3.1  Organization ......................................  15
 
−
        1.3.2  Requirements ......................................  16
 
−
        1.3.3  Terminology .......................................  17
 
−
      1.4  Acknowledgments ........................................  20
 
  
−
  2. LINK LAYER ..................................................  21
+
This document is one of a pair that defines and discusses the
−
      2.1  INTRODUCTION ...........................................  21
+
requirements for host system implementations of the Internet protocol
 +
suite. This RFC covers the communication protocol layers:  link
 +
layer, IP layer, and transport layer. Its companion RFC,
 +
"Requirements for Internet Hosts -- Application and Support"
 +
[INTRO:1], covers the application layer protocols. This document
 +
should also be read in conjunction with "Requirements for Internet
 +
Gateways" [INTRO:2].
  
 +
These documents are intended to provide guidance for vendors,
 +
implementors, and users of Internet communication software.  They
 +
represent the consensus of a large body of technical experience and
 +
wisdom, contributed by the members of the Internet research and
 +
vendor communities.
  
 +
This RFC enumerates standard protocols that a host connected to the
 +
Internet must use, and it incorporates by reference the RFCs and
 +
other documents describing the current specifications for these
 +
protocols.  It corrects errors in the referenced documents and adds
 +
additional discussion and guidance for an implementor.
  
−
Internet Engineering Task Force                                [Page 1]
+
For each protocol, this document also contains an explicit set of
 +
requirements, recommendations, and options.  The reader must
 +
understand that the list of requirements in this document is
 +
incomplete by itself; the complete set of requirements for an
 +
Internet host is primarily defined in the standard protocol
 +
specification documents, with the corrections, amendments, and
 +
supplements contained in this RFC.
  
 +
A good-faith implementation of the protocols that was produced after
 +
careful reading of the RFC's and with some interaction with the
 +
Internet technical community, and that followed good communications
 +
software engineering practices, should differ from the requirements
 +
of this document in only minor ways.  Thus, in many cases, the
 +
"requirements" in this RFC are already stated or implied in the
 +
standard protocol documents, so that their inclusion here is, in a
 +
sense, redundant.  However, they were included because some past
 +
implementation has made the wrong choice, causing problems of
 +
interoperability, performance, and/or robustness.
  
 +
This document includes discussion and explanation of many of the
 +
requirements and recommendations.  A simple list of requirements
 +
would be dangerous, because:
  
 +
o    Some required features are more important than others, and some
 +
    features are optional.
  
 
RFC1122                      INTRODUCTION                  October 1989
 
RFC1122                      INTRODUCTION                  October 1989
  
 +
o    There may be valid reasons why particular vendor products that
 +
    are designed for restricted contexts might choose to use
 +
    different specifications.
  
−
      2.2  PROTOCOL WALK-THROUGH ..................................  21
+
However, the specifications of this document must be followed to meet
−
      2.3  SPECIFIC ISSUES ........................................  21
+
the general goal of arbitrary host interoperation across the
−
        2.3.1 Trailer Protocol Negotiation ......................  21
+
diversity and complexity of the Internet system.  Although most
−
        2.3.2  Address Resolution Protocol -- ARP ................  22
+
current implementations fail to meet these requirements in various
−
            2.3.2.1  ARP Cache Validation .........................  22
+
ways, some minor and some major, this specification is the ideal
−
            2.3.2.2  ARP Packet Queue .............................  24
+
towards which we need to move.
−
        2.3.3  Ethernet and IEEE 802 Encapsulation ...............  24
 
−
      2.4  LINK/INTERNET LAYER INTERFACE ..........................  25
 
−
      2.5  LINK LAYER REQUIREMENTS SUMMARY ........................   26
 
  
−
  3. INTERNET LAYER PROTOCOLS ....................................  27
+
These requirements are based on the current level of Internet
−
      3.1 INTRODUCTION ............................................  27
+
architecture.  This document will be updated as required to provide
−
      3.2  PROTOCOL WALK-THROUGH ..................................  29
+
additional clarifications or to include additional information in
−
        3.2.1 Internet Protocol -- IP ............................  29
+
those areas in which specifications are still evolving.
−
            3.2.1.1 Version Number ...............................  29
 
−
            3.2.1.2  Checksum .....................................  29
 
−
            3.2.1.3  Addressing ...................................  29
 
−
            3.2.1.4  Fragmentation and Reassembly .................  32
 
−
            3.2.1.5  Identification ...............................  32
 
−
            3.2.1.6  Type-of-Service ..............................  33
 
−
            3.2.1.7  Time-to-Live .................................  34
 
−
            3.2.1.8  Options ......................................  35
 
−
        3.2.2 Internet Control Message Protocol -- ICMP ..........  38
 
−
            3.2.2.1  Destination Unreachable ......................  39
 
−
            3.2.2.2  Redirect .....................................  40
 
−
            3.2.2.3  Source Quench ................................  41
 
−
            3.2.2.4  Time Exceeded ................................  41
 
−
            3.2.2.5  Parameter Problem ............................  42
 
−
            3.2.2.6  Echo Request/Reply ...........................  42
 
−
            3.2.2.7  Information Request/Reply ....................  43
 
−
            3.2.2.8  Timestamp and Timestamp Reply ................  43
 
−
            3.2.2.9  Address Mask Request/Reply ...................  45
 
−
        3.2.3  Internet Group Management Protocol IGMP ...........  47
 
−
      3.3  SPECIFIC ISSUES ........................................  47
 
−
        3.3.1  Routing Outbound Datagrams ........................  47
 
−
            3.3.1.1  Local/Remote Decision ........................  47
 
−
            3.3.1.2  Gateway Selection ............................  48
 
−
            3.3.1.3  Route Cache ..................................  49
 
−
            3.3.1.4  Dead Gateway Detection .......................  51
 
−
            3.3.1.5  New Gateway Selection ........................  55
 
−
            3.3.1.6  Initialization ...............................  56
 
−
        3.3.2  Reassembly ........................................  56
 
−
        3.3.3  Fragmentation .....................................  58
 
−
        3.3.4  Local Multihoming .................................  60
 
−
            3.3.4.1  Introduction .................................  60
 
−
            3.3.4.2  Multihoming Requirements .....................  61
 
−
            3.3.4.3  Choosing a Source Address ....................  64
 
−
        3.3.5  Source Route Forwarding ...........................   65
 
  
 +
This introductory section begins with a brief overview of the
 +
Internet architecture as it relates to hosts, and then gives some
 +
general advice to host software vendors.  Finally, there is some
 +
guidance on reading the rest of the document and some terminology.
  
 +
1.1  The Internet Architecture
  
−
Internet Engineering Task Force                                [Page 2]
+
  General background and discussion on the Internet architecture and
 +
  supporting protocol suite can be found in the DDN Protocol
 +
  Handbook [INTRO:3]; for background see for example [INTRO:9],
 +
  [INTRO:10], and [INTRO:11].  Reference [INTRO:5] describes the
 +
  procedure for obtaining Internet protocol documents, while
 +
  [INTRO:6] contains a list of the numbers assigned within Internet
 +
  protocols.
  
 +
  1.1.1  Internet Hosts
  
 +
      A host computer, or simply "host," is the ultimate consumer of
 +
      communication services.  A host generally executes application
 +
      programs on behalf of user(s), employing network and/or
 +
      Internet communication services in support of this function.
 +
      An Internet host corresponds to the concept of an "End-System"
 +
      used in the OSI protocol suite [INTRO:13].
  
 +
      An Internet communication system consists of interconnected
 +
      packet networks supporting communication among host computers
 +
      using the Internet protocols.  The networks are interconnected
 +
      using packet-switching computers called "gateways" or "IP
 +
      routers" by the Internet community, and "Intermediate Systems"
 +
      by the OSI world [INTRO:13].  The RFC "Requirements for
 +
      Internet Gateways" [INTRO:2] contains the official
 +
      specifications for Internet gateways.  That RFC together with
  
 
RFC1122                      INTRODUCTION                  October 1989
 
RFC1122                      INTRODUCTION                  October 1989
  
 +
      the present document and its companion [INTRO:1] define the
 +
      rules for the current realization of the Internet architecture.
  
−
        3.3.6 Broadcasts ........................................  66
+
      Internet hosts span a wide range of size, speed, and function.
−
        3.3.7  IP Multicasting ...................................  67
+
      They range in size from small microprocessors through
−
        3.3.8  Error Reporting ...................................  69
+
      workstations to mainframes and supercomputers.  In function,
−
       3.4  INTERNET/TRANSPORT LAYER INTERFACE .....................  69
+
      they range from single-purpose hosts (such as terminal servers)
−
       3.5  INTERNET LAYER REQUIREMENTS SUMMARY ....................   72
+
      to full-service hosts that support a variety of online network
 +
       services, typically including remote login, file transfer, and
 +
       electronic mail.
  
−
  4. TRANSPORT PROTOCOLS .........................................  77
+
       A host is generally said to be multihomed if it has more than
−
       4.1  USER DATAGRAM PROTOCOL -- UDP ..........................  77
+
       one interface to the same or to different networks.  See
−
        4.1.1  INTRODUCTION ......................................  77
+
      Section 1.1.3 on "Terminology".
−
        4.1.2  PROTOCOL WALK-THROUGH .............................  77
 
−
        4.1.3  SPECIFIC ISSUES ...................................  77
 
−
            4.1.3.1  Ports ........................................  77
 
−
            4.1.3.2  IP Options ...................................  77
 
−
            4.1.3.3  ICMP Messages ................................  78
 
−
            4.1.3.4  UDP Checksums ................................  78
 
−
            4.1.3.5  UDP Multihoming ..............................  79
 
−
            4.1.3.6  Invalid Addresses ............................  79
 
−
        4.1.4  UDP/APPLICATION LAYER INTERFACE ...................  79
 
−
        4.1.5  UDP REQUIREMENTS SUMMARY ..........................  80
 
−
       4.2  TRANSMISSION CONTROL PROTOCOL -- TCP ...................  82
 
−
        4.2.1  INTRODUCTION ......................................  82
 
−
        4.2.2  PROTOCOL WALK-THROUGH .............................  82
 
−
            4.2.2.1  Well-Known Ports .............................  82
 
−
            4.2.2.2  Use of Push ..................................  82
 
−
            4.2.2.3  Window Size ..................................  83
 
−
            4.2.2.4  Urgent Pointer ...............................  84
 
−
            4.2.2.5  TCP Options ..................................  85
 
−
            4.2.2.6  Maximum Segment Size Option ..................  85
 
−
            4.2.2.7  TCP Checksum .................................  86
 
−
            4.2.2.8  TCP Connection State Diagram .................  86
 
−
            4.2.2.9  Initial Sequence Number Selection ............  87
 
−
            4.2.2.10  Simultaneous Open Attempts ..................  87
 
−
            4.2.2.11  Recovery from Old Duplicate SYN .............  87
 
−
            4.2.2.12  RST Segment .................................  87
 
−
            4.2.2.13  Closing a Connection ........................  87
 
−
            4.2.2.14  Data Communication ..........................  89
 
−
            4.2.2.15  Retransmission Timeout ......................  90
 
−
            4.2.2.16  Managing the Window .........................  91
 
−
            4.2.2.17  Probing Zero Windows ........................  92
 
−
            4.2.2.18  Passive OPEN Calls ..........................  92
 
−
            4.2.2.19  Time to Live ................................  93
 
−
            4.2.2.20 Event Processing ............................  93
 
−
            4.2.2.21  Acknowledging Queued Segments ...............  94
 
−
        4.2.3  SPECIFIC ISSUES ...................................  95
 
−
            4.2.3.1 Retransmission Timeout Calculation ...........  95
 
−
            4.2.3.2  When to Send an ACK Segment ..................  96
 
−
            4.2.3.3  When to Send a Window Update .................  97
 
−
            4.2.3.4  When to Send Data ............................  98
 
  
 +
  1.1.2  Architectural Assumptions
  
 +
      The current Internet architecture is based on a set of
 +
      assumptions about the communication system.  The assumptions
 +
      most relevant to hosts are as follows:
  
−
Internet Engineering Task Force                                [Page 3]
+
      (a)  The Internet is a network of networks.
  
 +
          Each host is directly connected to some particular
 +
          network(s); its connection to the Internet is only
 +
          conceptual.  Two hosts on the same network communicate
 +
          with each other using the same set of protocols that they
 +
          would use to communicate with hosts on distant networks.
  
 +
      (b)  Gateways don't keep connection state information.
  
 +
          To improve robustness of the communication system,
 +
          gateways are designed to be stateless, forwarding each IP
 +
          datagram independently of other datagrams.  As a result,
 +
          redundant paths can be exploited to provide robust service
 +
          in spite of failures of intervening gateways and networks.
  
−
RFC1122                      INTRODUCTION                  October 1989
+
          All state information required for end-to-end flow control
 +
          and reliability is implemented in the hosts, in the
 +
          transport layer or in application programs.  All
 +
          connection control information is thus co-located with the
 +
          end points of the communication, so it will be lost only
 +
          if an end point fails.
  
 +
      (c)  Routing complexity should be in the gateways.
  
−
            4.2.3.5  TCP Connection Failures ......................  100
+
          Routing is a complex and difficult problem, and ought to
−
            4.2.3.6  TCP Keep-Alives ..............................  101
+
          be performed by the gateways, not the hosts.  An important
−
            4.2.3.7  TCP Multihoming ..............................  103
 
−
            4.2.3.8  IP Options ...................................  103
 
−
            4.2.3.9  ICMP Messages ................................  103
 
−
            4.2.3.10  Remote Address Validation ...................  104
 
−
            4.2.3.11  TCP Traffic Patterns ........................  104
 
−
            4.2.3.12  Efficiency ..................................  105
 
−
        4.2.4  TCP/APPLICATION LAYER INTERFACE ...................  106
 
−
            4.2.4.1  Asynchronous Reports .........................  106
 
−
            4.2.4.2  Type-of-Service ..............................  107
 
−
            4.2.4.3  Flush Call ...................................  107
 
−
            4.2.4.4  Multihoming ..................................  108
 
−
        4.2.5  TCP REQUIREMENT SUMMARY ...........................  108
 
  
−
  5.  REFERENCES .................................................  112
+
RFC1122                      INTRODUCTION                  October 1989
  
 +
          objective is to insulate host software from changes caused
 +
          by the inevitable evolution of the Internet routing
 +
          architecture.
  
 +
      (d)  The System must tolerate wide network variation.
  
 +
          A basic objective of the Internet design is to tolerate a
 +
          wide range of network characteristics -- e.g., bandwidth,
 +
          delay, packet loss, packet reordering, and maximum packet
 +
          size.  Another objective is robustness against failure of
 +
          individual networks, gateways, and hosts, using whatever
 +
          bandwidth is still available.  Finally, the goal is full
 +
          "open system interconnection": an Internet host must be
 +
          able to interoperate robustly and effectively with any
 +
          other Internet host, across diverse Internet paths.
  
 +
          Sometimes host implementors have designed for less
 +
          ambitious goals.  For example, the LAN environment is
 +
          typically much more benign than the Internet as a whole;
 +
          LANs have low packet loss and delay and do not reorder
 +
          packets.  Some vendors have fielded host implementations
 +
          that are adequate for a simple LAN environment, but work
 +
          badly for general interoperation.  The vendor justifies
 +
          such a product as being economical within the restricted
 +
          LAN market.  However, isolated LANs seldom stay isolated
 +
          for long; they are soon gatewayed to each other, to
 +
          organization-wide internets, and eventually to the global
 +
          Internet system.  In the end, neither the customer nor the
 +
          vendor is served by incomplete or substandard Internet
 +
          host software.
  
 +
          The requirements spelled out in this document are designed
 +
          for a full-function Internet host, capable of full
 +
          interoperation over an arbitrary Internet path.
  
 +
  1.1.3  Internet Protocol Suite
  
 +
      To communicate using the Internet system, a host must implement
 +
      the layered set of protocols comprising the Internet protocol
 +
      suite.  A host typically must implement at least one protocol
 +
      from each layer.
  
 +
      The protocol layers used in the Internet architecture are as
 +
      follows [INTRO:4]:
  
 +
      o  Application Layer
  
 +
RFC1122                      INTRODUCTION                  October 1989
  
 +
          The application layer is the top layer of the Internet
 +
          protocol suite.  The Internet suite does not further
 +
          subdivide the application layer, although some of the
 +
          Internet application layer protocols do contain some
 +
          internal sub-layering.  The application layer of the
 +
          Internet suite essentially combines the functions of the
 +
          top two layers -- Presentation and Application -- of the
 +
          OSI reference model.
  
 +
          We distinguish two categories of application layer
 +
          protocols:  user protocols that provide service directly
 +
          to users, and support protocols that provide common system
 +
          functions.  Requirements for user and support protocols
 +
          will be found in the companion RFC [INTRO:1].
  
 +
          The most common Internet user protocols are:
  
 +
            o  Telnet (remote login)
 +
            o  FTP    (file transfer)
 +
            o  SMTP  (electronic mail delivery)
  
 +
          There are a number of other standardized user protocols
 +
          [INTRO:4] and many private user protocols.
  
 +
          Support protocols, used for host name mapping, booting,
 +
          and management, include SNMP, BOOTP, RARP, and the Domain
 +
          Name System (DNS) protocols.
  
 +
      o  Transport Layer
  
 +
          The transport layer provides end-to-end communication
 +
          services for applications.  There are two primary
 +
          transport layer protocols at present:
  
 +
            o Transmission Control Protocol (TCP)
 +
            o User Datagram Protocol (UDP)
  
 +
          TCP is a reliable connection-oriented transport service
 +
          that provides end-to-end reliability, resequencing, and
 +
          flow control.  UDP is a connectionless ("datagram")
 +
          transport service.
  
 +
          Other transport protocols have been developed by the
 +
          research community, and the set of official Internet
 +
          transport protocols may be expanded in the future.
  
 +
          Transport layer protocols are discussed in Chapter 4.
  
 +
RFC1122                      INTRODUCTION                  October 1989
  
 +
      o  Internet Layer
  
 +
          All Internet transport protocols use the Internet Protocol
 +
          (IP) to carry data from source host to destination host.
 +
          IP is a connectionless or datagram internetwork service,
 +
          providing no end-to-end delivery guarantees. Thus, IP
 +
          datagrams may arrive at the destination host damaged,
 +
          duplicated, out of order, or not at all.  The layers above
 +
          IP are responsible for reliable delivery service when it
 +
          is required.  The IP protocol includes provision for
 +
          addressing, type-of-service specification, fragmentation
 +
          and reassembly, and security information.
  
 +
          The datagram or connectionless nature of the IP protocol
 +
          is a fundamental and characteristic feature of the
 +
          Internet architecture.  Internet IP was the model for the
 +
          OSI Connectionless Network Protocol [INTRO:12].
  
 +
          ICMP is a control protocol that is considered to be an
 +
          integral part of IP, although it is architecturally
 +
          layered upon IP, i.e., it uses IP to carry its data end-
 +
          to-end just as a transport protocol like TCP or UDP does.
 +
          ICMP provides error reporting, congestion reporting, and
 +
          first-hop gateway redirection.
  
 +
          IGMP is an Internet layer protocol used for establishing
 +
          dynamic host groups for IP multicasting.
  
 +
          The Internet layer protocols IP, ICMP, and IGMP are
 +
          discussed in Chapter 3.
  
 +
      o  Link Layer
  
 +
          To communicate on its directly-connected network, a host
 +
          must implement the communication protocol used to
 +
          interface to that network.  We call this a link layer or
 +
          media-access layer protocol.
  
 +
          There is a wide variety of link layer protocols,
 +
          corresponding to the many different types of networks.
 +
          See Chapter 2.
  
 +
  1.1.4  Embedded Gateway Code
  
 +
      Some Internet host software includes embedded gateway
 +
      functionality, so that these hosts can forward packets as a
  
−
Internet Engineering Task Force                                [Page 4]
+
RFC1122                      INTRODUCTION                  October 1989
  
 +
      gateway would, while still performing the application layer
 +
      functions of a host.
  
 +
      Such dual-purpose systems must follow the Gateway Requirements
 +
      RFC [INTRO:2]  with respect to their gateway functions, and
 +
      must follow the present document with respect to their host
 +
      functions.  In all overlapping cases, the two specifications
 +
      should be in agreement.
  
 +
      There are varying opinions in the Internet community about
 +
      embedded gateway functionality.  The main arguments are as
 +
      follows:
  
−
RFC1122                      INTRODUCTION                  October 1989
+
      o    Pro: in a local network environment where networking is
 +
          informal, or in isolated internets, it may be convenient
 +
          and economical to use existing host systems as gateways.
  
 +
          There is also an architectural argument for embedded
 +
          gateway functionality: multihoming is much more common
 +
          than originally foreseen, and multihoming forces a host to
 +
          make routing decisions as if it were a gateway.  If the
 +
          multihomed  host contains an embedded gateway, it will
 +
          have full routing knowledge and as a result will be able
 +
          to make more optimal routing decisions.
  
−
1.  INTRODUCTION
+
      o    Con: Gateway algorithms and protocols are still changing,
 +
          and they will continue to change as the Internet system
 +
          grows larger.  Attempting to include a general gateway
 +
          function within the host IP layer will force host system
 +
          maintainers to track these (more frequent) changes.  Also,
 +
          a larger pool of gateway implementations will make
 +
          coordinating the changes more difficult.  Finally, the
 +
          complexity of a gateway IP layer is somewhat greater than
 +
          that of a host, making the implementation and operation
 +
          tasks more complex.
  
−
  This document is one of a pair that defines and discusses the
+
          In addition, the style of operation of some hosts is not
−
  requirements for host system implementations of the Internet protocol
+
          appropriate for providing stable and robust gateway
−
  suite.  This RFC covers the communication protocol layers:  link
+
          service.
−
  layer, IP layer, and transport layer.  Its companion RFC,
 
−
  "Requirements for Internet Hosts -- Application and Support"
 
−
  [INTRO:1], covers the application layer protocols.  This document
 
−
  should also be read in conjunction with "Requirements for Internet
 
−
  Gateways" [INTRO:2].
 
  
−
  These documents are intended to provide guidance for vendors,
+
      There is considerable merit in both of these viewpoints.  One
−
  implementors, and users of Internet communication software.  They
+
      conclusion can be drawn: an host administrator must have
−
  represent the consensus of a large body of technical experience and
+
      conscious control over whether or not a given host acts as a
−
  wisdom, contributed by the members of the Internet research and
+
      gateway.  See Section 3.1 for the detailed requirements.
−
  vendor communities.
 
  
−
  This RFC enumerates standard protocols that a host connected to the
+
RFC1122                      INTRODUCTION                  October 1989
−
  Internet must use, and it incorporates by reference the RFCs and
 
−
  other documents describing the current specifications for these
 
−
  protocols.  It corrects errors in the referenced documents and adds
 
−
  additional discussion and guidance for an implementor.
 
−
 
 
−
  For each protocol, this document also contains an explicit set of
 
−
  requirements, recommendations, and options.  The reader must
 
−
  understand that the list of requirements in this document is
 
−
  incomplete by itself; the complete set of requirements for an
 
−
  Internet host is primarily defined in the standard protocol
 
−
  specification documents, with the corrections, amendments, and
 
−
  supplements contained in this RFC.
 
  
−
  A good-faith implementation of the protocols that was produced after
+
1.2 General Considerations
−
  careful reading of the RFC's and with some interaction with the
 
−
  Internet technical community, and that followed good communications
 
−
  software engineering practices, should differ from the requirements
 
−
  of this document in only minor ways.  Thus, in many cases, the
 
−
  "requirements" in this RFC are already stated or implied in the
 
−
  standard protocol documents, so that their inclusion here is, in a
 
−
  sense, redundant.  However, they were included because some past
 
−
  implementation has made the wrong choice, causing problems of
 
−
  interoperability, performance, and/or robustness.
 
  
−
   This document includes discussion and explanation of many of the
+
   There are two important lessons that vendors of Internet host
−
   requirements and recommendations.  A simple list of requirements
+
   software have learned and which a new vendor should consider
−
   would be dangerous, because:
+
   seriously.
  
−
   o    Some required features are more important than others, and some
+
   1.2.1  Continuing Internet Evolution
−
        features are optional.
 
  
 +
      The enormous growth of the Internet has revealed problems of
 +
      management and scaling in a large datagram-based packet
 +
      communication system.  These problems are being addressed, and
 +
      as a result there will be continuing evolution of the
 +
      specifications described in this document.  These changes will
 +
      be carefully planned and controlled, since there is extensive
 +
      participation in this planning by the vendors and by the
 +
      organizations responsible for operations of the networks.
  
 +
      Development, evolution, and revision are characteristic of
 +
      computer network protocols today, and this situation will
 +
      persist for some years.  A vendor who develops computer
 +
      communication software for the Internet protocol suite (or any
 +
      other protocol suite!) and then fails to maintain and update
 +
      that software for changing specifications is going to leave a
 +
      trail of unhappy customers.  The Internet is a large
 +
      communication network, and the users are in constant contact
 +
      through it.  Experience has shown that knowledge of
 +
      deficiencies in vendor software propagates quickly through the
 +
      Internet technical community.
  
−
Internet Engineering Task Force                                [Page 5]
+
  1.2.2  Robustness Principle
  
 +
      At every layer of the protocols, there is a general rule whose
 +
      application can lead to enormous benefits in robustness and
 +
      interoperability [IP:1]:
  
 +
            "Be liberal in what you accept, and
 +
              conservative in what you send"
  
 +
      Software should be written to deal with every conceivable
 +
      error, no matter how unlikely; sooner or later a packet will
 +
      come in with that particular combination of errors and
 +
      attributes, and unless the software is prepared, chaos can
 +
      ensue.  In general, it is best to assume that the network is
 +
      filled with malevolent entities that will send in packets
 +
      designed to have the worst possible effect.  This assumption
 +
      will lead to suitable protective design, although the most
 +
      serious problems in the Internet have been caused by
 +
      unenvisaged mechanisms triggered by low-probability events;
  
 
RFC1122                      INTRODUCTION                  October 1989
 
RFC1122                      INTRODUCTION                  October 1989
  
 +
      mere human malice would never have taken so devious a course!
  
−
  o    There may be valid reasons why particular vendor products that
+
      Adaptability to change must be designed into all levels of
−
        are designed for restricted contexts might choose to use
+
      Internet host software.  As a simple example, consider a
−
        different specifications.
+
      protocol specification that contains an enumeration of values
 +
      for a particular header field -- e.g., a type field, a port
 +
      number, or an error code; this enumeration must be assumed to
 +
      be incomplete.  Thus, if a protocol specification defines four
 +
      possible error codes, the software must not break when a fifth
 +
      code shows up.  An undefined code might be logged (see below),
 +
      but it must not cause a failure.
  
−
  However, the specifications of this document must be followed to meet
+
      The second part of the principle is almost as important:
−
  the general goal of arbitrary host interoperation across the
+
      software on other hosts may contain deficiencies that make it
−
  diversity and complexity of the Internet system.  Although most
+
      unwise to exploit legal but obscure protocol features.  It is
−
  current implementations fail to meet these requirements in various
+
      unwise to stray far from the obvious and simple, lest untoward
−
  ways, some minor and some major, this specification is the ideal
+
      effects result elsewhere.  A corollary of this is "watch out
−
  towards which we need to move.
+
      for misbehaving hosts"; host software should be prepared, not
 +
      just to survive other misbehaving hosts, but also to cooperate
 +
      to limit the amount of disruption such hosts can cause to the
 +
      shared communication facility.
  
−
   These requirements are based on the current level of Internet
+
   1.2.3 Error Logging
−
  architecture.  This document will be updated as required to provide
 
−
  additional clarifications or to include additional information in
 
−
  those areas in which specifications are still evolving.
 
  
−
  This introductory section begins with a brief overview of the
+
      The Internet includes a great variety of host and gateway
−
  Internet architecture as it relates to hosts, and then gives some
+
      systems, each implementing many protocols and protocol layers,
−
  general advice to host software vendors.  Finally, there is some
+
      and some of these contain bugs and mis-features in their
−
  guidance on reading the rest of the document and some terminology.
+
      Internet protocol software.  As a result of complexity,
 +
      diversity, and distribution of function, the diagnosis of
 +
      Internet problems is often very difficult.
  
−
  1.1 The Internet Architecture
+
      Problem diagnosis will be aided if host implementations include
 +
      a carefully designed facility for logging erroneous or
 +
      "strange" protocol events.  It is important to include as much
 +
      diagnostic information as possible when an error is logged.  In
 +
      particular, it is often useful to record the header(s) of a
 +
      packet that caused an error.  However, care must be taken to
 +
      ensure that error logging does not consume prohibitive amounts
 +
      of resources or otherwise interfere with the operation of the
 +
      host.
  
−
       General background and discussion on the Internet architecture and
+
       There is a tendency for abnormal but harmless protocol events
−
       supporting protocol suite can be found in the DDN Protocol
+
       to overflow error logging files; this can be avoided by using a
−
       Handbook [INTRO:3]; for background see for example [INTRO:9],
+
       "circular" log, or by enabling logging only while diagnosing a
−
       [INTRO:10], and [INTRO:11].  Reference [INTRO:5] describes the
+
       known failure.  It may be useful to filter and count duplicate
−
       procedure for obtaining Internet protocol documents, while
+
      successive messages.  One strategy that seems to work well is:
−
       [INTRO:6] contains a list of the numbers assigned within Internet
+
       (1) always count abnormalities and make such counts accessible
−
      protocols.
+
       through the management protocol (see [INTRO:1]); and (2) allow
  
−
      1.1.1  Internet Hosts
+
RFC1122                      INTRODUCTION                  October 1989
  
−
        A host computer, or simply "host," is the ultimate consumer of
+
      the logging of a great variety of events to be selectively
−
        communication services.  A host generally executes application
+
      enabled.  For example, it might useful to be able to "log
−
        programs on behalf of user(s), employing network and/or
+
      everything" or to "log everything for host X".
−
        Internet communication services in support of this function.
 
−
        An Internet host corresponds to the concept of an "End-System"
 
−
        used in the OSI protocol suite [INTRO:13].
 
  
−
        An Internet communication system consists of interconnected
+
      Note that different managements may have differing policies
−
        packet networks supporting communication among host computers
+
      about the amount of error logging that they want normally
−
        using the Internet protocols.  The networks are interconnected
+
      enabled in a host.  Some will say, "if it doesn't hurt me, I
−
        using packet-switching computers called "gateways" or "IP
+
      don't want to know about it", while others will want to take a
−
        routers" by the Internet community, and "Intermediate Systems"
+
      more watchful and aggressive attitude about detecting and
−
        by the OSI world [INTRO:13]. The RFC "Requirements for
+
      removing protocol abnormalities.
−
        Internet Gateways" [INTRO:2] contains the official
 
−
        specifications for Internet gateways.  That RFC together with
 
  
 +
  1.2.4  Configuration
  
 +
      It would be ideal if a host implementation of the Internet
 +
      protocol suite could be entirely self-configuring.  This would
 +
      allow the whole suite to be implemented in ROM or cast into
 +
      silicon, it would simplify diskless workstations, and it would
 +
      be an immense boon to harried LAN administrators as well as
 +
      system vendors.  We have not reached this ideal; in fact, we
 +
      are not even close.
  
−
Internet Engineering Task Force                                [Page 6]
+
      At many points in this document, you will find a requirement
 +
      that a parameter be a configurable option.  There are several
 +
      different reasons behind such requirements.  In a few cases,
 +
      there is current uncertainty or disagreement about the best
 +
      value, and it may be necessary to update the recommended value
 +
      in the future.  In other cases, the value really depends on
 +
      external factors -- e.g., the size of the host and the
 +
      distribution of its communication load, or the speeds and
 +
      topology of nearby networks -- and self-tuning algorithms are
 +
      unavailable and may be insufficient.  In some cases,
 +
      configurability is needed because of administrative
 +
      requirements.
  
 +
      Finally, some configuration options are required to communicate
 +
      with obsolete or incorrect implementations of the protocols,
 +
      distributed without sources, that unfortunately persist in many
 +
      parts of the Internet.  To make correct systems coexist with
 +
      these faulty systems, administrators often have to "mis-
 +
      configure" the correct systems.  This problem will correct
 +
      itself gradually as the faulty systems are retired, but it
 +
      cannot be ignored by vendors.
  
 +
      When we say that a parameter must be configurable, we do not
 +
      intend to require that its value be explicitly read from a
 +
      configuration file at every boot time.  We recommend that
 +
      implementors set up a default for each parameter, so a
 +
      configuration file is only necessary to override those defaults
  
 +
RFC1122                      INTRODUCTION                  October 1989
  
−
RFC1122                      INTRODUCTION                  October 1989
+
      that are inappropriate in a particular installation.  Thus, the
 +
      configurability requirement is an assurance that it will be
 +
      POSSIBLE to override the default when necessary, even in a
 +
      binary-only or ROM-based product.
  
 +
      This document requires a particular value for such defaults in
 +
      some cases.  The choice of default is a sensitive issue when
 +
      the configuration item controls the accommodation to existing
 +
      faulty systems.  If the Internet is to converge successfully to
 +
      complete interoperability, the default values built into
 +
      implementations must implement the official protocol, not
 +
      "mis-configurations" to accommodate faulty implementations.
 +
      Although marketing considerations have led some vendors to
 +
      choose mis-configuration defaults, we urge vendors to choose
 +
      defaults that will conform to the standard.
  
−
        the present document and its companion [INTRO:1] define the
+
      Finally, we note that a vendor needs to provide adequate
−
        rules for the current realization of the Internet architecture.
+
      documentation on all configuration parameters, their limits and
 +
      effects.
  
−
        Internet hosts span a wide range of size, speed, and function.
+
1.3 Reading this Document
−
        They range in size from small microprocessors through
 
−
        workstations to mainframes and supercomputers.  In function,
 
−
        they range from single-purpose hosts (such as terminal servers)
 
−
        to full-service hosts that support a variety of online network
 
−
        services, typically including remote login, file transfer, and
 
−
        electronic mail.
 
  
−
        A host is generally said to be multihomed if it has more than
+
  1.3.1  Organization
−
        one interface to the same or to different networks.  See
 
−
        Section 1.1.3 on "Terminology".
 
  
−
       1.1.2 Architectural Assumptions
+
       Protocol layering, which is generally used as an organizing
 +
      principle in implementing network software, has also been used
 +
      to organize this document.  In describing the rules, we assume
 +
      that an implementation does strictly mirror the layering of the
 +
      protocols.  Thus, the following three major sections specify
 +
      the requirements for the link layer, the internet layer, and
 +
      the transport layer, respectively. A companion RFC [INTRO:1]
 +
      covers application level software.  This layerist organization
 +
      was chosen for simplicity and clarity.
  
−
        The current Internet architecture is based on a set of
+
      However, strict layering is an imperfect model, both for the
−
        assumptions about the communication system.  The assumptions
+
      protocol suite and for recommended implementation approaches.
−
        most relevant to hosts are as follows:
+
      Protocols in different layers interact in complex and sometimes
 +
      subtle ways, and particular functions often involve multiple
 +
      layers.  There are many design choices in an implementation,
 +
      many of which involve creative "breaking" of strict layering.
 +
      Every implementor is urged to read references [INTRO:7] and
 +
      [INTRO:8].
  
−
        (a)  The Internet is a network of networks.
+
      This document describes the conceptual service interface
 +
      between layers using a functional ("procedure call") notation,
 +
      like that used in the TCP specification [TCP:1]. A host
 +
      implementation must support the logical information flow
  
−
              Each host is directly connected to some particular
+
RFC1122                      INTRODUCTION                  October 1989
−
              network(s); its connection to the Internet is only
 
−
              conceptual.  Two hosts on the same network communicate
 
−
              with each other using the same set of protocols that they
 
−
              would use to communicate with hosts on distant networks.
 
  
−
        (b) Gateways don't keep connection state information.
+
      implied by these calls, but need not literally implement the
 +
      calls themselves. For example, many implementations reflect
 +
      the coupling between the transport layer and the IP layer by
 +
      giving them shared access to common data structures.  These
 +
      data structures, rather than explicit procedure calls, are then
 +
      the agency for passing much of the information that is
 +
      required.
  
−
              To improve robustness of the communication system,
+
      In general, each major section of this document is organized
−
              gateways are designed to be stateless, forwarding each IP
+
      into the following subsections:
−
              datagram independently of other datagrams.  As a result,
 
−
              redundant paths can be exploited to provide robust service
 
−
              in spite of failures of intervening gateways and networks.
 
  
−
              All state information required for end-to-end flow control
+
      (1) Introduction
−
              and reliability is implemented in the hosts, in the
 
−
              transport layer or in application programs. All
 
−
              connection control information is thus co-located with the
 
−
              end points of the communication, so it will be lost only
 
−
              if an end point fails.
 
  
−
        (c)  Routing complexity should be in the gateways.
+
      (2)  Protocol Walk-Through -- considers the protocol
 +
          specification documents section-by-section, correcting
 +
          errors, stating requirements that may be ambiguous or
 +
          ill-defined, and providing further clarification or
 +
          explanation.
  
−
              Routing is a complex and difficult problem, and ought to
+
      (3)  Specific Issues -- discusses protocol design and
−
              be performed by the gateways, not the hosts. An important
+
          implementation issues that were not included in the walk-
 +
          through.
  
 +
      (4)  Interfaces -- discusses the service interface to the next
 +
          higher layer.
  
 +
      (5)  Summary -- contains a summary of the requirements of the
 +
          section.
  
−
Internet Engineering Task Force                                [Page 7]
+
      Under many of the individual topics in this document, there is
 +
      parenthetical material labeled "DISCUSSION" or
 +
      "IMPLEMENTATION". This material is intended to give
 +
      clarification and explanation of the preceding requirements
 +
      text.  It also includes some suggestions on possible future
 +
      directions or developments.  The implementation material
 +
      contains suggested approaches that an implementor may want to
 +
      consider.
  
 +
      The summary sections are intended to be guides and indexes to
 +
      the text, but are necessarily cryptic and incomplete.  The
 +
      summaries should never be used or referenced separately from
 +
      the complete RFC.
  
 +
  1.3.2  Requirements
  
 +
      In this document, the words that are used to define the
 +
      significance of each particular requirement are capitalized.
  
 
RFC1122                      INTRODUCTION                  October 1989
 
RFC1122                      INTRODUCTION                  October 1989
  
 +
      These words are:
  
−
              objective is to insulate host software from changes caused
+
      *    "MUST"
−
              by the inevitable evolution of the Internet routing
 
−
              architecture.
 
  
−
        (d)  The System must tolerate wide network variation.
+
          This word or the adjective "REQUIRED" means that the item
 +
          is an absolute requirement of the specification.
  
−
              A basic objective of the Internet design is to tolerate a
+
      *    "SHOULD"
−
              wide range of network characteristics -- e.g., bandwidth,
 
−
              delay, packet loss, packet reordering, and maximum packet
 
−
              size.  Another objective is robustness against failure of
 
−
              individual networks, gateways, and hosts, using whatever
 
−
              bandwidth is still available.  Finally, the goal is full
 
−
              "open system interconnection": an Internet host must be
 
−
              able to interoperate robustly and effectively with any
 
−
              other Internet host, across diverse Internet paths.
 
  
−
              Sometimes host implementors have designed for less
+
          This word or the adjective "RECOMMENDED" means that there
−
              ambitious goals.  For example, the LAN environment is
+
          may exist valid reasons in particular circumstances to
−
              typically much more benign than the Internet as a whole;
+
          ignore this item, but the full implications should be
−
              LANs have low packet loss and delay and do not reorder
+
          understood and the case carefully weighed before choosing
−
              packets.  Some vendors have fielded host implementations
+
          a different course.
−
              that are adequate for a simple LAN environment, but work
 
−
              badly for general interoperation.  The vendor justifies
 
−
              such a product as being economical within the restricted
 
−
              LAN market.  However, isolated LANs seldom stay isolated
 
−
              for long; they are soon gatewayed to each other, to
 
−
              organization-wide internets, and eventually to the global
 
−
              Internet system.  In the end, neither the customer nor the
 
−
              vendor is served by incomplete or substandard Internet
 
−
              host software.
 
  
−
              The requirements spelled out in this document are designed
+
      *    "MAY"
−
              for a full-function Internet host, capable of full
 
−
              interoperation over an arbitrary Internet path.
 
  
 +
          This word or the adjective "OPTIONAL" means that this item
 +
          is truly optional.  One vendor may choose to include the
 +
          item because a particular marketplace requires it or
 +
          because it enhances the product, for example; another
 +
          vendor may omit the same item.
  
−
       1.1.3  Internet Protocol Suite
+
       An implementation is not compliant if it fails to satisfy one
 +
      or more of the MUST requirements for the protocols it
 +
      implements. An implementation that satisfies all the MUST and
 +
      all the SHOULD requirements for its protocols is said to be
 +
      "unconditionally compliant"; one that satisfies all the MUST
 +
      requirements but not all the SHOULD requirements for its
 +
      protocols is said to be "conditionally compliant".
  
−
        To communicate using the Internet system, a host must implement
+
  1.3.3 Terminology
−
        the layered set of protocols comprising the Internet protocol
 
−
        suite.  A host typically must implement at least one protocol
 
−
        from each layer.
 
  
−
        The protocol layers used in the Internet architecture are as
+
      This document uses the following technical terms:
−
        follows [INTRO:4]:
 
  
 +
      Segment
 +
          A segment is the unit of end-to-end transmission in the
 +
          TCP protocol.  A segment consists of a TCP header followed
 +
          by application data.  A segment is transmitted by
 +
          encapsulation inside an IP datagram.
  
−
        o Application Layer
+
      Message
 +
          In this description of the lower-layer protocols, a
 +
          message is the unit of transmission in a transport layer
 +
          protocol. In particular, a TCP segment is a message.  A
 +
          message consists of a transport protocol header followed
 +
          by application protocol data.  To be transmitted end-to-
  
 +
RFC1122                      INTRODUCTION                  October 1989
  
 +
          end through the Internet, a message must be encapsulated
 +
          inside a datagram.
  
−
Internet Engineering Task Force                                [Page 8]
+
      IP Datagram
 +
          An IP datagram is the unit of end-to-end transmission in
 +
          the IP protocol.  An IP datagram consists of an IP header
 +
          followed by transport layer data, i.e., of an IP header
 +
          followed by a message.
  
 +
          In the description of the internet layer (Section 3), the
 +
          unqualified term "datagram" should be understood to refer
 +
          to an IP datagram.
  
 +
      Packet
 +
          A packet is the unit of data passed across the interface
 +
          between the internet layer and the link layer.  It
 +
          includes an IP header and data.  A packet may be a
 +
          complete IP datagram or a fragment of an IP datagram.
  
 +
      Frame
 +
          A frame is the unit of transmission in a link layer
 +
          protocol, and consists of a link-layer header followed by
 +
          a packet.
  
−
RFC1122                      INTRODUCTION                  October 1989
+
      Connected Network
 +
          A network to which a host is interfaced is often known as
 +
          the "local network" or the "subnetwork" relative to that
 +
          host.  However, these terms can cause confusion, and
 +
          therefore we use the term "connected network" in this
 +
          document.
  
 +
      Multihomed
 +
          A host is said to be multihomed if it has multiple IP
 +
          addresses.  For a discussion of multihoming, see Section
 +
          3.3.4 below.
  
−
              The application layer is the top layer of the Internet
+
      Physical network interface
−
              protocol suite.  The Internet suite does not further
+
          This is a physical interface to a connected network and
−
              subdivide the application layer, although some of the
+
          has a (possibly unique) link-layer address.  Multiple
−
              Internet application layer protocols do contain some
+
          physical network interfaces on a single host may share the
−
              internal sub-layering.  The application layer of the
+
          same link-layer address, but the address must be unique
−
              Internet suite essentially combines the functions of the
+
          for different hosts on the same physical network.
−
              top two layers -- Presentation and Application -- of the
 
−
              OSI reference model.
 
  
−
              We distinguish two categories of application layer
+
      Logical [network] interface
−
              protocols:  user protocols that provide service directly
+
          We define a logical [network] interface to be a logical
−
              to users, and support protocols that provide common system
+
          path, distinguished by a unique IP address, to a connected
−
              functions.  Requirements for user and support protocols
+
          network.  See Section 3.3.4.
−
              will be found in the companion RFC [INTRO:1].
 
  
−
              The most common Internet user protocols are:
+
RFC1122                      INTRODUCTION                  October 1989
  
−
                o  Telnet (remote login)
+
      Specific-destination address
−
                o  FTP    (file transfer)
+
          This is the effective destination address of a datagram,
−
                o  SMTP  (electronic mail delivery)
+
          even if it is broadcast or multicast; see Section 3.2.1.3.
  
−
              There are a number of other standardized user protocols
+
      Path
−
              [INTRO:4] and many private user protocols.
+
          At a given moment, all the IP datagrams from a particular
 +
          source host to a particular destination host will
 +
          typically traverse the same sequence of gateways.  We use
 +
          the term "path" for this sequence.  Note that a path is
 +
          uni-directional; it is not unusual to have different paths
 +
          in the two directions between a given host pair.
  
−
              Support protocols, used for host name mapping, booting,
+
      MTU
−
              and management, include SNMP, BOOTP, RARP, and the Domain
+
          The maximum transmission unit, i.e., the size of the
−
              Name System (DNS) protocols.
+
          largest packet that can be transmitted.
  
 +
      The terms frame, packet, datagram, message, and segment are
 +
      illustrated by the following schematic diagrams:
  
−
        o  Transport Layer
+
      A. Transmission on connected network:
 +
        _______________________________________________
 +
      | LL hdr | IP hdr |        (data)              |
 +
      |________|________|_____________________________|
  
−
              The transport layer provides end-to-end communication
+
        <---------- Frame ----------------------------->
−
              services for applications.  There are two primary
+
                <----------Packet -------------------->
−
              transport layer protocols at present:
 
  
−
                 o Transmission Control Protocol (TCP)
+
      B. Before IP fragmentation or after IP reassembly:
−
                 o User Datagram Protocol (UDP)
+
                ______________________________________
 +
                 | IP hdr | transport| Application Data |
 +
                 |________|____hdr___|__________________|
  
−
              TCP is a reliable connection-oriented transport service
+
                <--------  Datagram ------------------>
−
              that provides end-to-end reliability, resequencing, and
+
                          <-------- Message ----------->
−
              flow control. UDP is a connectionless ("datagram")
+
        or, for TCP:
−
              transport service.
+
                ______________________________________
 +
                | IP hdr | TCP hdr | Application Data |
 +
                |________|__________|__________________|
  
−
              Other transport protocols have been developed by the
+
                <--------  Datagram ------------------>
−
              research community, and the set of official Internet
+
                          <-------- Segment ----------->
−
              transport protocols may be expanded in the future.
 
  
−
              Transport layer protocols are discussed in Chapter 4.
+
RFC1122                      INTRODUCTION                  October 1989
  
 +
1.4  Acknowledgments
  
 +
  This document incorporates contributions and comments from a large
 +
  group of Internet protocol experts, including representatives of
 +
  university and research labs, vendors, and government agencies.
 +
  It was assembled primarily by the Host Requirements Working Group
 +
  of the Internet Engineering Task Force (IETF).
  
−
Internet Engineering Task Force                                [Page 9]
+
  The Editor would especially like to acknowledge the tireless
 +
  dedication of the following people, who attended many long
 +
  meetings and generated 3 million bytes of electronic mail over the
 +
  past 18 months in pursuit of this document: Philip Almquist, Dave
 +
  Borman (Cray Research), Noel Chiappa, Dave Crocker (DEC), Steve
 +
  Deering (Stanford), Mike Karels (Berkeley), Phil Karn (Bellcore),
 +
  John Lekashman (NASA), Charles Lynn (BBN), Keith McCloghrie (TWG),
 +
  Paul Mockapetris (ISI), Thomas Narten (Purdue), Craig Partridge
 +
  (BBN), Drew Perkins (CMU), and James Van Bokkelen (FTP Software).
  
 +
  In addition, the following people made major contributions to the
 +
  effort: Bill Barns (Mitre), Steve Bellovin (AT&T), Mike Brescia
 +
  (BBN), Ed Cain (DCA), Annette DeSchon (ISI), Martin Gross (DCA),
 +
  Phill Gross (NRI), Charles Hedrick (Rutgers), Van Jacobson (LBL),
 +
  John Klensin (MIT), Mark Lottor (SRI), Milo Medin (NASA), Bill
 +
  Melohn (Sun Microsystems), Greg Minshall (Kinetics), Jeff Mogul
 +
  (DEC), John Mullen (CMC), Jon Postel (ISI), John Romkey (Epilogue
 +
  Technology), and Mike StJohns (DCA).  The following also made
 +
  significant contributions to particular areas: Eric Allman
 +
  (Berkeley), Rob Austein (MIT), Art Berggreen (ACC), Keith Bostic
 +
  (Berkeley), Vint Cerf (NRI), Wayne Hathaway (NASA), Matt Korn
 +
  (IBM), Erik Naggum (Naggum Software, Norway), Robert Ullmann
 +
  (Prime Computer), David Waitzman (BBN), Frank Wancho (USA), Arun
 +
  Welch (Ohio State), Bill Westfield (Cisco), and Rayan Zachariassen
 +
  (Toronto).
  
 +
  We are grateful to all, including any contributors who may have
 +
  been inadvertently omitted from this list.
  
 +
RFC1122                        LINK LAYER                  October 1989
  
−
RFC1122                      INTRODUCTION                  October 1989
+
== LINK LAYER ==
  
 +
2.1  INTRODUCTION
  
−
        o Internet Layer
+
  All Internet systems, both hosts and gateways, have the same
 +
  requirements for link layer protocols. These requirements are
 +
  given in Chapter 3 of "Requirements for Internet Gateways"
 +
  [INTRO:2], augmented with the material in this section.
  
−
              All Internet transport protocols use the Internet Protocol
+
2.2 PROTOCOL WALK-THROUGH
−
              (IP) to carry data from source host to destination host.
 
−
              IP is a connectionless or datagram internetwork service,
 
−
              providing no end-to-end delivery guarantees. Thus, IP
 
−
              datagrams may arrive at the destination host damaged,
 
−
              duplicated, out of order, or not at all.  The layers above
 
−
              IP are responsible for reliable delivery service when it
 
−
              is required.  The IP protocol includes provision for
 
−
              addressing, type-of-service specification, fragmentation
 
−
              and reassembly, and security information.
 
  
−
              The datagram or connectionless nature of the IP protocol
+
  None.
−
              is a fundamental and characteristic feature of the
 
−
              Internet architecture.  Internet IP was the model for the
 
−
              OSI Connectionless Network Protocol [INTRO:12].
 
  
−
              ICMP is a control protocol that is considered to be an
+
2.3  SPECIFIC ISSUES
−
              integral part of IP, although it is architecturally
 
−
              layered upon IP, i.e., it uses IP to carry its data end-
 
−
              to-end just as a transport protocol like TCP or UDP does.
 
−
              ICMP provides error reporting, congestion reporting, and
 
−
              first-hop gateway redirection.
 
  
−
              IGMP is an Internet layer protocol used for establishing
+
  2.3.1  Trailer Protocol Negotiation
−
              dynamic host groups for IP multicasting.
 
  
−
              The Internet layer protocols IP, ICMP, and IGMP are
+
      The trailer protocol [LINK:1] for link-layer encapsulation MAY
−
              discussed in Chapter 3.
+
      be used, but only when it has been verified that both systems
 +
      (host or gateway) involved in the link-layer communication
 +
      implement trailers.  If the system does not dynamically
 +
      negotiate use of the trailer protocol on a per-destination
 +
      basis, the default configuration MUST disable the protocol.
  
 +
      DISCUSSION:
 +
          The trailer protocol is a link-layer encapsulation
 +
          technique that rearranges the data contents of packets
 +
          sent on the physical network.  In some cases, trailers
 +
          improve the throughput of higher layer protocols by
 +
          reducing the amount of data copying within the operating
 +
          system.  Higher layer protocols are unaware of trailer
 +
          use, but both the sending and receiving host MUST
 +
          understand the protocol if it is used.
  
−
        o Link Layer
+
          Improper use of trailers can result in very confusing
 +
          symptoms. Only packets with specific size attributes are
 +
          encapsulated using trailers, and typically only a small
 +
          fraction of the packets being exchanged have these
 +
          attributes.  Thus, if a system using trailers exchanges
 +
          packets with a system that does not, some packets
 +
          disappear into a black hole while others are delivered
 +
          successfully.
  
−
              To communicate on its directly-connected network, a host
+
      IMPLEMENTATION:
−
              must implement the communication protocol used to
+
          On an Ethernet, packets encapsulated with trailers use a
−
              interface to that network.  We call this a link layer or
+
          distinct Ethernet type [LINK:1], and trailer negotiation
−
              media-access layer protocol.
+
          is performed at the time that ARP is used to discover the
 +
          link-layer address of a destination system.
  
−
              There is a wide variety of link layer protocols,
+
RFC1122                        LINK LAYER                  October 1989
−
              corresponding to the many different types of networks.
 
−
              See Chapter 2.
 
  
 +
          Specifically, the ARP exchange is completed in the usual
 +
          manner using the normal IP protocol type, but a host that
 +
          wants to speak trailers will send an additional "trailer
 +
          ARP reply" packet, i.e., an ARP reply that specifies the
 +
          trailer encapsulation protocol type but otherwise has the
 +
          format of a normal ARP reply.  If a host configured to use
 +
          trailers receives a trailer ARP reply message from a
 +
          remote machine, it can add that machine to the list of
 +
          machines that understand trailers, e.g., by marking the
 +
          corresponding entry in the ARP cache.
  
−
      1.1.4  Embedded Gateway Code
+
          Hosts wishing to receive trailer encapsulations send
 +
          trailer ARP replies whenever they complete exchanges of
 +
          normal ARP messages for IP. Thus, a host that received an
 +
          ARP request for its IP protocol address would send a
 +
          trailer ARP reply in addition to the normal IP ARP reply;
 +
          a host that sent the IP ARP request would send a trailer
 +
          ARP reply when it received the corresponding IP ARP reply.
 +
          In this way, either the requesting or responding host in
 +
          an IP ARP exchange may request that it receive trailer
 +
          encapsulations.
  
−
        Some Internet host software includes embedded gateway
+
          This scheme, using extra trailer ARP reply packets rather
−
        functionality, so that these hosts can forward packets as a
+
          than sending an ARP request for the trailer protocol type,
 +
          was designed to avoid a continuous exchange of ARP packets
 +
          with a misbehaving host that, contrary to any
 +
          specification or common sense, responded to an ARP reply
 +
          for trailers with another ARP reply for IP.  This problem
 +
          is avoided by sending a trailer ARP reply in response to
 +
          an IP ARP reply only when the IP ARP reply answers an
 +
          outstanding request; this is true when the hardware
 +
          address for the host is still unknown when the IP ARP
 +
          reply is received.  A trailer ARP reply may always be sent
 +
          along with an IP ARP reply responding to an IP ARP
 +
          request.
  
 +
  2.3.2  Address Resolution Protocol -- ARP
  
 +
      2.3.2.1  ARP Cache Validation
  
−
Internet Engineering Task Force                                [Page 10]
+
        An implementation of the Address Resolution Protocol (ARP)
 +
        [LINK:2] MUST provide a mechanism to flush out-of-date cache
 +
        entries.  If this mechanism involves a timeout, it SHOULD be
 +
        possible to configure the timeout value.
  
 +
        A mechanism to prevent ARP flooding (repeatedly sending an
 +
        ARP Request for the same IP address, at a high rate) MUST be
 +
        included.  The recommended maximum rate is 1 per second per
  
 +
RFC1122                        LINK LAYER                  October 1989
  
 +
        destination.
  
−
RFC1122                      INTRODUCTION                  October 1989
+
        DISCUSSION:
 +
              The ARP specification [LINK:2] suggests but does not
 +
              require a timeout mechanism to invalidate cache entries
 +
              when hosts change their Ethernet addresses.  The
 +
              prevalence of proxy ARP (see Section 2.4 of [INTRO:2])
 +
              has significantly increased the likelihood that cache
 +
              entries in hosts will become invalid, and therefore
 +
              some ARP-cache invalidation mechanism is now required
 +
              for hosts.  Even in the absence of proxy ARP, a long-
 +
              period cache timeout is useful in order to
 +
              automatically correct any bad ARP data that might have
 +
              been cached.
  
 +
        IMPLEMENTATION:
 +
              Four mechanisms have been used, sometimes in
 +
              combination, to flush out-of-date cache entries.
  
−
        gateway would, while still performing the application layer
+
              (1)  Timeout -- Periodically time out cache entries,
−
        functions of a host.
+
                  even if they are in use.  Note that this timeout
 +
                  should be restarted when the cache entry is
 +
                  "refreshed" (by observing the source fields,
 +
                  regardless of target address, of an ARP broadcast
 +
                  from the system in question).  For proxy ARP
 +
                  situations, the timeout needs to be on the order
 +
                  of a minute.
  
−
        Such dual-purpose systems must follow the Gateway Requirements
+
              (2)  Unicast Poll -- Actively poll the remote host by
−
        RFC [INTRO:2]  with respect to their gateway functions, and
+
                  periodically sending a point-to-point ARP Request
−
        must follow the present document with respect to their host
+
                  to it, and delete the entry if no ARP Reply is
−
        functions.  In all overlapping cases, the two specifications
+
                  received from N successive polls.  Again, the
−
        should be in agreement.
+
                  timeout should be on the order of a minute, and
 +
                  typically N is 2.
  
−
        There are varying opinions in the Internet community about
+
              (3)  Link-Layer Advice -- If the link-layer driver
−
        embedded gateway functionality. The main arguments are as
+
                  detects a delivery problem, flush the
−
        follows:
+
                  corresponding ARP cache entry.
  
−
        o    Pro: in a local network environment where networking is
+
              (4)  Higher-layer Advice -- Provide a call from the
−
              informal, or in isolated internets, it may be convenient
+
                  Internet layer to the link layer to indicate a
−
              and economical to use existing host systems as gateways.
+
                  delivery problem.  The effect of this call would
 +
                  be to invalidate the corresponding cache entry.
 +
                  This call would be analogous to the
 +
                  "ADVISE_DELIVPROB()" call from the transport layer
 +
                  to the Internet layer (see Section 3.4), and in
 +
                  fact the ADVISE_DELIVPROB routine might in turn
 +
                  call the link-layer advice routine to invalidate
  
−
              There is also an architectural argument for embedded
+
RFC1122                        LINK LAYER                  October 1989
−
              gateway functionality: multihoming is much more common
 
−
              than originally foreseen, and multihoming forces a host to
 
−
              make routing decisions as if it were a gateway.  If the
 
−
              multihomed  host contains an embedded gateway, it will
 
−
              have full routing knowledge and as a result will be able
 
−
              to make more optimal routing decisions.
 
  
−
        o    Con: Gateway algorithms and protocols are still changing,
+
                  the ARP cache entry.
−
              and they will continue to change as the Internet system
 
−
              grows larger.  Attempting to include a general gateway
 
−
              function within the host IP layer will force host system
 
−
              maintainers to track these (more frequent) changes.  Also,
 
−
              a larger pool of gateway implementations will make
 
−
              coordinating the changes more difficult.  Finally, the
 
−
              complexity of a gateway IP layer is somewhat greater than
 
−
              that of a host, making the implementation and operation
 
−
              tasks more complex.
 
  
−
               In addition, the style of operation of some hosts is not
+
               Approaches (1) and (2) involve ARP cache timeouts on
−
               appropriate for providing stable and robust gateway
+
              the order of a minute or less.  In the absence of proxy
−
               service.
+
              ARP, a timeout this short could create noticeable
 +
              overhead traffic on a very large Ethernet.  Therefore,
 +
               it may be necessary to configure a host to lengthen the
 +
               ARP cache timeout.
  
−
        There is considerable merit in both of these viewpoints. One
+
      2.3.2.2  ARP Packet Queue
−
        conclusion can be drawn: an host administrator must have
 
−
        conscious control over whether or not a given host acts as a
 
−
        gateway.  See Section 3.1 for the detailed requirements.
 
  
 +
        The link layer SHOULD save (rather than discard) at least
 +
        one (the latest) packet of each set of packets destined to
 +
        the same unresolved IP address, and transmit the saved
 +
        packet when the address has been resolved.
  
 +
        DISCUSSION:
 +
              Failure to follow this recommendation causes the first
 +
              packet of every exchange to be lost.  Although higher-
 +
              layer protocols can generally cope with packet loss by
 +
              retransmission, packet loss does impact performance.
 +
              For example, loss of a TCP open request causes the
 +
              initial round-trip time estimate to be inflated.  UDP-
 +
              based applications such as the Domain Name System are
 +
              more seriously affected.
  
 +
  2.3.3  Ethernet and IEEE 802 Encapsulation
  
 +
      The IP encapsulation for Ethernets is described in RFC-894
 +
      [LINK:3], while RFC-1042 [LINK:4] describes the IP
 +
      encapsulation for IEEE 802 networks.  RFC-1042 elaborates and
 +
      replaces the discussion in Section 3.4 of [INTRO:2].
  
 +
      Every Internet host connected to a 10Mbps Ethernet cable:
  
 +
      o    MUST be able to send and receive packets using RFC-894
 +
          encapsulation;
  
−
Internet Engineering Task Force                                [Page 11]
+
      o    SHOULD be able to receive RFC-1042 packets, intermixed
 +
          with RFC-894 packets; and
  
 +
      o    MAY be able to send packets using RFC-1042 encapsulation.
  
 +
      An Internet host that implements sending both the RFC-894 and
 +
      the RFC-1042 encapsulations MUST provide a configuration switch
 +
      to select which is sent, and this switch MUST default to RFC-
 +
      894.
  
 +
RFC1122                        LINK LAYER                  October 1989
  
−
RFC1122                      INTRODUCTION                  October 1989
+
      Note that the standard IP encapsulation in RFC-1042 does not
 +
      use the protocol id value (K1=6) that IEEE reserved for IP;
 +
      instead, it uses a value (K1=170) that implies an extension
 +
      (the "SNAP") which can be used to hold the Ether-Type field.
 +
      An Internet system MUST NOT send 802 packets using K1=6.
  
 +
      Address translation from Internet addresses to link-layer
 +
      addresses on Ethernet and IEEE 802 networks MUST be managed by
 +
      the Address Resolution Protocol (ARP).
  
−
  1.2  General Considerations
+
      The MTU for an Ethernet is 1500 and for 802.3 is 1492.
  
−
       There are two important lessons that vendors of Internet host
+
       DISCUSSION:
−
      software have learned and which a new vendor should consider
+
          The IEEE 802.3 specification provides for operation over a
−
      seriously.
+
          10Mbps Ethernet cable, in which case Ethernet and IEEE
 +
          802.3 frames can be physically intermixed.  A receiver can
 +
          distinguish Ethernet and 802.3 frames by the value of the
 +
          802.3 Length field; this two-octet field coincides in the
 +
          header with the Ether-Type field of an Ethernet frame.  In
 +
          particular, the 802.3 Length field must be less than or
 +
          equal to 1500, while all valid Ether-Type values are
 +
          greater than 1500.
  
−
      1.2.1  Continuing Internet Evolution
+
          Another compatibility problem arises with link-layer
 +
          broadcasts. A broadcast sent with one framing will not be
 +
          seen by hosts that can receive only the other framing.
  
−
        The enormous growth of the Internet has revealed problems of
+
          The provisions of this section were designed to provide
−
        management and scaling in a large datagram-based packet
+
          direct interoperation between 894-capable and 1042-capable
−
        communication system. These problems are being addressed, and
+
          systems on the same cable, to the maximum extent possible.
−
        as a result there will be continuing evolution of the
+
          It is intended to support the present situation where
−
        specifications described in this document.  These changes will
+
          894-only systems predominate, while providing an easy
−
        be carefully planned and controlled, since there is extensive
+
          transition to a possible future in which 1042-capable
−
        participation in this planning by the vendors and by the
+
          systems become common.
−
        organizations responsible for operations of the networks.
 
  
−
        Development, evolution, and revision are characteristic of
+
          Note that 894-only systems cannot interoperate directly
−
        computer network protocols today, and this situation will
+
          with 1042-only systems.  If the two system types are set
−
        persist for some years. A vendor who develops computer
+
          up as two different logical networks on the same cable,
−
        communication software for the Internet protocol suite (or any
+
          they can communicate only through an IP gateway.
−
        other protocol suite!) and then fails to maintain and update
+
          Furthermore, it is not useful or even possible for a
−
        that software for changing specifications is going to leave a
+
          dual-format host to discover automatically which format to
−
        trail of unhappy customers.  The Internet is a large
+
          send, because of the problem of link-layer broadcasts.
−
        communication network, and the users are in constant contact
 
−
        through it.  Experience has shown that knowledge of
 
−
        deficiencies in vendor software propagates quickly through the
 
−
        Internet technical community.
 
  
−
      1.2.2 Robustness Principle
+
2.4 LINK/INTERNET LAYER INTERFACE
  
−
        At every layer of the protocols, there is a general rule whose
+
  The packet receive interface between the IP layer and the link
−
        application can lead to enormous benefits in robustness and
+
  layer MUST include a flag to indicate whether the incoming packet
−
        interoperability [IP:1]:
+
  was addressed to a link-layer broadcast address.
  
−
                "Be liberal in what you accept, and
+
RFC1122                        LINK LAYER                  October 1989
−
                conservative in what you send"
 
  
−
        Software should be written to deal with every conceivable
+
  DISCUSSION
−
        error, no matter how unlikely; sooner or later a packet will
+
        Although the IP layer does not generally know link layer
−
        come in with that particular combination of errors and
+
        addresses (since every different network medium typically has
−
        attributes, and unless the software is prepared, chaos can
+
        a different address format), the broadcast address on a
−
        ensue.  In general, it is best to assume that the network is
+
        broadcast-capable medium is an important special case.  See
−
        filled with malevolent entities that will send in packets
+
        Section 3.2.2, especially the DISCUSSION concerning broadcast
−
        designed to have the worst possible effect. This assumption
+
        storms.
−
        will lead to suitable protective design, although the most
 
−
        serious problems in the Internet have been caused by
 
−
        unenvisaged mechanisms triggered by low-probability events;
 
  
 +
  The packet send interface between the IP and link layers MUST
 +
  include the 5-bit TOS field (see Section 3.2.1.6).
  
 +
  The link layer MUST NOT report a Destination Unreachable error to
 +
  IP solely because there is no ARP cache entry for a destination.
  
−
Internet Engineering Task Force                                [Page 12]
+
2.5  LINK LAYER REQUIREMENTS SUMMARY
  
 +
                                              |      | | | |S| |
 +
                                              |      | | | |H| |F
 +
                                              |      | | | |O|M|o
 +
                                              |      | |S| |U|U|o
 +
                                              |      | |H| |L|S|t
 +
                                              |      |M|O| |D|T|n
 +
                                              |      |U|U|M| | |o
 +
                                              |      |S|L|A|N|N|t
 +
                                              |      |T|D|Y|O|O|t
 +
FEATURE                                          |SECTION| | | |T|T|e
 +
--------------------------------------------------|-------|-|-|-|-|-|--
 +
                                              |      | | | | | |
 +
Trailer encapsulation                            |2.3.1  | | |x| | |
 +
Send Trailers by default without negotiation      |2.3.1  | | | | |x|
 +
ARP                                              |2.3.2  | | | | | |
 +
  Flush out-of-date ARP cache entries            |2.3.2.1|x| | | | |
 +
  Prevent ARP floods                              |2.3.2.1|x| | | | |
 +
  Cache timeout configurable                      |2.3.2.1| |x| | | |
 +
  Save at least one (latest) unresolved pkt      |2.3.2.2| |x| | | |
 +
Ethernet and IEEE 802 Encapsulation              |2.3.3  | | | | | |
 +
  Host able to:                                  |2.3.3  | | | | | |
 +
Send & receive RFC-894 encapsulation          |2.3.3  |x| | | | |
 +
Receive RFC-1042 encapsulation                |2.3.3  | |x| | | |
 +
Send RFC-1042 encapsulation                  |2.3.3  | | |x| | |
 +
  Then config. sw. to select, RFC-894 dflt    |2.3.3  |x| | | | |
 +
  Send K1=6 encapsulation                        |2.3.3  | | | | |x|
 +
  Use ARP on Ethernet and IEEE 802 nets          |2.3.3  |x| | | | |
 +
Link layer report b'casts to IP layer            |2.4    |x| | | | |
 +
IP layer pass TOS to link layer                  |2.4    |x| | | | |
 +
No ARP cache entry treated as Dest. Unreach.      |2.4    | | | | |x|
  
 +
RFC1122                      INTERNET LAYER                October 1989
  
 +
== INTERNET LAYER PROTOCOLS ==
  
−
RFC1122                      INTRODUCTION                 October 1989
+
3.1 INTRODUCTION
  
 +
  The Robustness Principle: "Be liberal in what you accept, and
 +
  conservative in what you send" is particularly important in the
 +
  Internet layer, where one misbehaving host can deny Internet
 +
  service to many other hosts.
  
−
        mere human malice would never have taken so devious a course!
+
  The protocol standards used in the Internet layer are:
  
−
        Adaptability to change must be designed into all levels of
+
  o    RFC-791 [IP:1] defines the IP protocol and gives an
−
        Internet host software.  As a simple example, consider a
+
        introduction to the architecture of the Internet.
−
        protocol specification that contains an enumeration of values
 
−
        for a particular header field -- e.g., a type field, a port
 
−
        number, or an error code; this enumeration must be assumed to
 
−
        be incomplete.  Thus, if a protocol specification defines four
 
−
        possible error codes, the software must not break when a fifth
 
−
        code shows up.  An undefined code might be logged (see below),
 
−
        but it must not cause a failure.
 
  
−
        The second part of the principle is almost as important:
+
  o    RFC-792 [IP:2] defines ICMP, which provides routing,
−
        software on other hosts may contain deficiencies that make it
+
        diagnostic and error functionality for IP.  Although ICMP
−
        unwise to exploit legal but obscure protocol features.  It is
+
        messages are encapsulated within IP datagrams, ICMP
−
        unwise to stray far from the obvious and simple, lest untoward
+
        processing is considered to be (and is typically implemented
−
        effects result elsewhere.  A corollary of this is "watch out
+
        as) part of the IP layer.  See Section 3.2.2.
−
        for misbehaving hosts"; host software should be prepared, not
 
−
        just to survive other misbehaving hosts, but also to cooperate
 
−
        to limit the amount of disruption such hosts can cause to the
 
−
        shared communication facility.
 
  
−
      1.2.3  Error Logging
+
  o    RFC-950 [IP:3] defines the mandatory subnet extension to the
 +
        addressing architecture.
  
−
        The Internet includes a great variety of host and gateway
+
  o    RFC-1112 [IP:4] defines the Internet Group Management
−
        systems, each implementing many protocols and protocol layers,
+
        Protocol IGMP, as part of a recommended extension to hosts
−
        and some of these contain bugs and mis-features in their
+
        and to the host-gateway interface to support Internet-wide
−
        Internet protocol software.  As a result of complexity,
+
        multicasting at the IP level.  See Section 3.2.3.
−
        diversity, and distribution of function, the diagnosis of
 
−
        Internet problems is often very difficult.
 
  
−
        Problem diagnosis will be aided if host implementations include
+
        The target of an IP multicast may be an arbitrary group of
−
        a carefully designed facility for logging erroneous or
+
        Internet hosts.  IP multicasting is designed as a natural
−
        "strange" protocol events.  It is important to include as much
+
        extension of the link-layer multicasting facilities of some
−
        diagnostic information as possible when an error is logged.  In
+
        networks, and it provides a standard means for local access
−
        particular, it is often useful to record the header(s) of a
+
        to such link-layer multicasting facilities.
−
        packet that caused an error.  However, care must be taken to
 
−
        ensure that error logging does not consume prohibitive amounts
 
−
        of resources or otherwise interfere with the operation of the
 
−
        host.
 
  
−
        There is a tendency for abnormal but harmless protocol events
+
  Other important references are listed in Section 5 of this
−
        to overflow error logging files; this can be avoided by using a
+
  document.
−
        "circular" log, or by enabling logging only while diagnosing a
 
−
        known failure.  It may be useful to filter and count duplicate
 
−
        successive messages. One strategy that seems to work well is:
 
−
        (1) always count abnormalities and make such counts accessible
 
−
        through the management protocol (see [INTRO:1]); and (2) allow
 
  
 +
  The Internet layer of host software MUST implement both IP and
 +
  ICMP.  See Section 3.3.7 for the requirements on support of IGMP.
  
 +
  The host IP layer has two basic functions:  (1) choose the "next
 +
  hop" gateway or host for outgoing IP datagrams and (2) reassemble
 +
  incoming IP datagrams.  The IP layer may also (3) implement
 +
  intentional fragmentation of outgoing datagrams.  Finally, the IP
 +
  layer must (4) provide diagnostic and error functionality.  We
 +
  expect that IP layer functions may increase somewhat in the
 +
  future, as further Internet control and management facilities are
 +
  developed.
  
−
Internet Engineering Task Force                                [Page 13]
+
RFC1122                      INTERNET LAYER                October 1989
  
 +
  For normal datagrams, the processing is straightforward.  For
 +
  incoming datagrams, the IP layer:
  
 +
  (1)  verifies that the datagram is correctly formatted;
  
 +
  (2)  verifies that it is destined to the local host;
  
−
RFC1122                      INTRODUCTION                  October 1989
+
  (3)  processes options;
  
 +
  (4)  reassembles the datagram if necessary; and
  
−
        the logging of a great variety of events to be selectively
+
  (5)  passes the encapsulated message to the appropriate
−
        enabled.  For example, it might useful to be able to "log
+
        transport-layer protocol module.
−
        everything" or to "log everything for host X".
 
  
−
        Note that different managements may have differing policies
+
  For outgoing datagrams, the IP layer:
−
        about the amount of error logging that they want normally
 
−
        enabled in a host.  Some will say, "if it doesn't hurt me, I
 
−
        don't want to know about it", while others will want to take a
 
−
        more watchful and aggressive attitude about detecting and
 
−
        removing protocol abnormalities.
 
  
−
      1.2.4 Configuration
+
  (1) sets any fields not set by the transport layer;
  
−
        It would be ideal if a host implementation of the Internet
+
  (2)  selects the correct first hop on the connected network (a
−
        protocol suite could be entirely self-configuring.  This would
+
        process called "routing");
−
        allow the whole suite to be implemented in ROM or cast into
 
−
        silicon, it would simplify diskless workstations, and it would
 
−
        be an immense boon to harried LAN administrators as well as
 
−
        system vendors.  We have not reached this ideal; in fact, we
 
−
        are not even close.
 
  
−
        At many points in this document, you will find a requirement
+
  (3) fragments the datagram if necessary and if intentional
−
        that a parameter be a configurable option. There are several
+
        fragmentation is implemented (see Section 3.3.3); and
−
        different reasons behind such requirements.  In a few cases,
 
−
        there is current uncertainty or disagreement about the best
 
−
        value, and it may be necessary to update the recommended value
 
−
        in the future.  In other cases, the value really depends on
 
−
        external factors -- e.g., the size of the host and the
 
−
        distribution of its communication load, or the speeds and
 
−
        topology of nearby networks -- and self-tuning algorithms are
 
−
        unavailable and may be insufficient.  In some cases,
 
−
        configurability is needed because of administrative
 
−
        requirements.
 
  
−
        Finally, some configuration options are required to communicate
+
  (4)  passes the packet(s) to the appropriate link-layer driver.
−
        with obsolete or incorrect implementations of the protocols,
 
−
        distributed without sources, that unfortunately persist in many
 
−
        parts of the Internet.  To make correct systems coexist with
 
−
        these faulty systems, administrators often have to "mis-
 
−
        configure" the correct systems.  This problem will correct
 
−
        itself gradually as the faulty systems are retired, but it
 
−
        cannot be ignored by vendors.
 
  
−
        When we say that a parameter must be configurable, we do not
+
  A host is said to be multihomed if it has multiple IP addresses.
−
        intend to require that its value be explicitly read from a
+
  Multihoming introduces considerable confusion and complexity into
−
        configuration file at every boot time.  We recommend that
+
  the protocol suite, and it is an area in which the Internet
−
        implementors set up a default for each parameter, so a
+
  architecture falls seriously short of solving all problems.  There
−
        configuration file is only necessary to override those defaults
+
  are two distinct problem areas in multihoming:
  
 +
  (1)  Local multihoming --  the host itself is multihomed; or
  
 +
  (2)  Remote multihoming -- the local host needs to communicate
 +
        with a remote multihomed host.
  
−
Internet Engineering Task Force                                [Page 14]
+
  At present, remote multihoming MUST be handled at the application
 +
  layer, as discussed in the companion RFC [INTRO:1].  A host MAY
 +
  support local multihoming, which is discussed in this document,
 +
  and in particular in Section 3.3.4.
  
 +
  Any host that forwards datagrams generated by another host is
 +
  acting as a gateway and MUST also meet the specifications laid out
 +
  in the gateway requirements RFC [INTRO:2].  An Internet host that
 +
  includes embedded gateway code MUST have a configuration switch to
 +
  disable the gateway function, and this switch MUST default to the
  
 +
RFC1122                      INTERNET LAYER                October 1989
  
 +
  non-gateway mode.  In this mode, a datagram arriving through one
 +
  interface will not be forwarded to another host or gateway (unless
 +
  it is source-routed), regardless of whether the host is single-
 +
  homed or multihomed.  The host software MUST NOT automatically
 +
  move into gateway mode if the host has more than one interface, as
 +
  the operator of the machine may neither want to provide that
 +
  service nor be competent to do so.
  
−
RFC1122                      INTRODUCTION                  October 1989
+
  In the following, the action specified in certain cases is to
 +
  "silently discard" a received datagram.  This means that the
 +
  datagram will be discarded without further processing and that the
 +
  host will not send any ICMP error message (see Section 3.2.2) as a
 +
  result.  However, for diagnosis of problems a host SHOULD provide
 +
  the capability of logging the error (see Section 1.2.3), including
 +
  the contents of the silently-discarded datagram, and SHOULD record
 +
  the event in a statistics counter.
  
 +
  DISCUSSION:
 +
        Silent discard of erroneous datagrams is generally intended
 +
        to prevent "broadcast storms".
  
−
        that are inappropriate in a particular installation.  Thus, the
+
3.2 PROTOCOL WALK-THROUGH
−
        configurability requirement is an assurance that it will be
 
−
        POSSIBLE to override the default when necessary, even in a
 
−
        binary-only or ROM-based product.
 
  
−
        This document requires a particular value for such defaults in
+
  3.2.1 Internet Protocol -- IP
−
        some cases. The choice of default is a sensitive issue when
 
−
        the configuration item controls the accommodation to existing
 
−
        faulty systems. If the Internet is to converge successfully to
 
−
        complete interoperability, the default values built into
 
−
        implementations must implement the official protocol, not
 
−
        "mis-configurations" to accommodate faulty implementations.
 
−
        Although marketing considerations have led some vendors to
 
−
        choose mis-configuration defaults, we urge vendors to choose
 
−
        defaults that will conform to the standard.
 
  
−
        Finally, we note that a vendor needs to provide adequate
+
      3.2.1.1  Version Number: RFC-791 Section 3.1
−
        documentation on all configuration parameters, their limits and
 
−
        effects.
 
  
 +
        A datagram whose version number is not 4 MUST be silently
 +
        discarded.
  
−
  1.3 Reading this Document
+
      3.2.1.2  Checksum: RFC-791 Section 3.1
  
−
      1.3.1  Organization
+
        A host MUST verify the IP header checksum on every received
 +
        datagram and silently discard every datagram that has a bad
 +
        checksum.
  
−
        Protocol layering, which is generally used as an organizing
+
      3.2.1.3 Addressing: RFC-791 Section 3.2
−
        principle in implementing network software, has also been used
 
−
        to organize this document. In describing the rules, we assume
 
−
        that an implementation does strictly mirror the layering of the
 
−
        protocols. Thus, the following three major sections specify
 
−
        the requirements for the link layer, the internet layer, and
 
−
        the transport layer, respectively.  A companion RFC [INTRO:1]
 
−
        covers application level software.  This layerist organization
 
−
        was chosen for simplicity and clarity.
 
  
−
         However, strict layering is an imperfect model, both for the
+
         There are now five classes of IP addresses: Class A through
−
        protocol suite and for recommended implementation approaches.
+
         Class E.  Class D addresses are used for IP multicasting
−
        Protocols in different layers interact in complex and sometimes
+
         [IP:4], while Class E addresses are reserved for
−
        subtle ways, and particular functions often involve multiple
+
         experimental use.
−
         layers.  There are many design choices in an implementation,
 
−
         many of which involve creative "breaking" of strict layering.
 
−
        Every implementor is urged to read references [INTRO:7] and
 
−
         [INTRO:8].
 
  
−
         This document describes the conceptual service interface
+
         A multicast (Class D) address is a 28-bit logical address
−
         between layers using a functional ("procedure call") notation,
+
         that stands for a group of hosts, and may be either
−
         like that used in the TCP specification [TCP:1].  A host
+
         permanent or transient.  Permanent multicast addresses are
−
         implementation must support the logical information flow
+
         allocated by the Internet Assigned Number Authority
 +
        [INTRO:6], while transient addresses may be allocated
  
 +
RFC1122                      INTERNET LAYER                October 1989
  
 +
        dynamically to transient groups.  Group membership is
 +
        determined dynamically using IGMP [IP:4].
  
−
Internet Engineering Task Force                                [Page 15]
+
        We now summarize the important special cases for Class A, B,
 +
        and C IP addresses, using the following notation for an IP
 +
        address:
  
 +
            { <Network-number>, <Host-number> }
  
 +
        or
 +
            { <Network-number>, <Subnet-number>, <Host-number> }
  
 +
        and the notation "-1" for a field that contains all 1 bits.
 +
        This notation is not intended to imply that the 1-bits in an
 +
        address mask need be contiguous.
  
−
RFC1122                      INTRODUCTION                  October 1989
+
        (a)  { 0, 0 }
  
 +
              This host on this network.  MUST NOT be sent, except as
 +
              a source address as part of an initialization procedure
 +
              by which the host learns its own IP address.
  
−
        implied by these calls, but need not literally implement the
+
              See also Section 3.3.6 for a non-standard use of {0,0}.
−
        calls themselves. For example, many implementations reflect
 
−
        the coupling between the transport layer and the IP layer by
 
−
        giving them shared access to common data structures. These
 
−
        data structures, rather than explicit procedure calls, are then
 
−
        the agency for passing much of the information that is
 
−
        required.
 
  
−
         In general, each major section of this document is organized
+
         (b)  { 0, <Host-number> }
−
        into the following subsections:
 
  
−
        (1) Introduction
+
              Specified host on this network. It MUST NOT be sent,
 +
              except as a source address as part of an initialization
 +
              procedure by which the host learns its full IP address.
  
−
         (2)  Protocol Walk-Through -- considers the protocol
+
         (c)  { -1, -1 }
−
              specification documents section-by-section, correcting
 
−
              errors, stating requirements that may be ambiguous or
 
−
              ill-defined, and providing further clarification or
 
−
              explanation.
 
  
−
        (3) Specific Issues -- discusses protocol design and
+
              Limited broadcast. It MUST NOT be used as a source
−
              implementation issues that were not included in the walk-
+
               address.
−
               through.
 
  
−
        (4)  Interfaces -- discusses the service interface to the next
+
              A datagram with this destination address will be
−
               higher layer.
+
              received by every host on the connected physical
 +
               network but will not be forwarded outside that network.
  
−
         (5)  Summary -- contains a summary of the requirements of the
+
         (d)  { <Network-number>, -1 }
−
              section.
 
  
 +
              Directed broadcast to the specified network.  It MUST
 +
              NOT be used as a source address.
  
−
         Under many of the individual topics in this document, there is
+
         (e)  { <Network-number>, <Subnet-number>, -1 }
−
        parenthetical material labeled "DISCUSSION" or
 
−
        "IMPLEMENTATION". This material is intended to give
 
−
        clarification and explanation of the preceding requirements
 
−
        text.  It also includes some suggestions on possible future
 
−
        directions or developments.  The implementation material
 
−
        contains suggested approaches that an implementor may want to
 
−
        consider.
 
  
−
        The summary sections are intended to be guides and indexes to
+
              Directed broadcast to the specified subnet.  It MUST
−
        the text, but are necessarily cryptic and incomplete.  The
+
              NOT be used as a source address.
−
        summaries should never be used or referenced separately from
 
−
        the complete RFC.
 
  
−
      1.3.2  Requirements
+
RFC1122                      INTERNET LAYER                October 1989
  
−
         In this document, the words that are used to define the
+
         (f)  { <Network-number>, -1, -1 }
−
        significance of each particular requirement are capitalized.
 
  
 +
              Directed broadcast to all subnets of the specified
 +
              subnetted network.  It MUST NOT be used as a source
 +
              address.
  
 +
        (g)  { 127, <any> }
  
−
Internet Engineering Task Force                                [Page 16]
+
              Internal host loopback address.  Addresses of this form
 +
              MUST NOT appear outside a host.
  
 +
        The <Network-number> is administratively assigned so that
 +
        its value will be unique in the entire world.
  
 +
        IP addresses are not permitted to have the value 0 or -1 for
 +
        any of the <Host-number>, <Network-number>, or <Subnet-
 +
        number> fields (except in the special cases listed above).
 +
        This implies that each of these fields will be at least two
 +
        bits long.
  
 +
        For further discussion of broadcast addresses, see Section
 +
        3.3.6.
  
−
RFC1122                      INTRODUCTION                  October 1989
+
        A host MUST support the subnet extensions to IP [IP:3].  As
 +
        a result, there will be an address mask of the form:
 +
        {-1, -1, 0} associated with each of the host's local IP
 +
        addresses; see Sections 3.2.2.9 and 3.3.1.1.
  
 +
        When a host sends any datagram, the IP source address MUST
 +
        be one of its own IP addresses (but not a broadcast or
 +
        multicast address).
  
−
         These words are:
+
         A host MUST silently discard an incoming datagram that is
 +
        not destined for the host.  An incoming datagram is destined
 +
        for the host if the datagram's destination address field is:
  
−
         *    "MUST"
+
         (1)  (one of) the host's IP address(es); or
  
−
              This word or the adjective "REQUIRED" means that the item
+
        (2)  an IP broadcast address valid for the connected
−
               is an absolute requirement of the specification.
+
               network; or
  
−
         *    "SHOULD"
+
         (3)  the address for a multicast group of which the host is
 +
              a member on the incoming physical interface.
  
−
              This word or the adjective "RECOMMENDED" means that there
+
        For most purposes, a datagram addressed to a broadcast or
−
              may exist valid reasons in particular circumstances to
+
        multicast destination is processed as if it had been
−
              ignore this item, but the full implications should be
+
        addressed to one of the host's IP addresses; we use the term
−
              understood and the case carefully weighed before choosing
+
        "specific-destination address" for the equivalent local IP
−
              a different course.
 
  
−
        *    "MAY"
+
RFC1122                      INTERNET LAYER                October 1989
  
−
              This word or the adjective "OPTIONAL" means that this item
+
        address of the host.  The specific-destination address is
−
              is truly optional.  One vendor may choose to include the
+
        defined to be the destination address in the IP header
−
              item because a particular marketplace requires it or
+
        unless the header contains a broadcast or multicast address,
−
              because it enhances the product, for example; another
+
        in which case the specific-destination is an IP address
−
              vendor may omit the same item.
+
        assigned to the physical interface on which the datagram
 +
        arrived.
  
 +
        A host MUST silently discard an incoming datagram containing
 +
        an IP source address that is invalid by the rules of this
 +
        section.  This validation could be done in either the IP
 +
        layer or by each protocol in the transport layer.
  
−
         An implementation is not compliant if it fails to satisfy one
+
         DISCUSSION:
−
        or more of the MUST requirements for the protocols it
+
              A mis-addressed datagram might be caused by a link-
−
        implements.  An implementation that satisfies all the MUST and
+
              layer broadcast of a unicast datagram or by a gateway
−
        all the SHOULD requirements for its protocols is said to be
+
              or host that is confused or mis-configured.
−
        "unconditionally compliant"; one that satisfies all the MUST
 
−
        requirements but not all the SHOULD requirements for its
 
−
        protocols is said to be "conditionally compliant".
 
  
−
      1.3.3 Terminology
+
              An architectural goal for Internet hosts was to allow
 +
              IP addresses to be featureless 32-bit numbers, avoiding
 +
              algorithms that required a knowledge of the IP address
 +
              format. Otherwise, any future change in the format or
 +
              interpretation of IP addresses will require host
 +
              software changes.  However, validation of broadcast and
 +
              multicast addresses violates this goal; a few other
 +
              violations are described elsewhere in this document.
  
−
        This document uses the following technical terms:
+
              Implementers should be aware that applications
 +
              depending upon the all-subnets directed broadcast
 +
              address (f) may be unusable on some networks.  All-
 +
              subnets broadcast is not widely implemented in vendor
 +
              gateways at present, and even when it is implemented, a
 +
              particular network administration may disable it in the
 +
              gateway configuration.
  
−
        Segment
+
      3.2.1.4 Fragmentation and Reassembly: RFC-791 Section 3.2
−
              A segment is the unit of end-to-end transmission in the
 
−
              TCP protocol. A segment consists of a TCP header followed
 
−
              by application data.  A segment is transmitted by
 
−
              encapsulation inside an IP datagram.
 
  
−
         Message
+
         The Internet model requires that every host support
−
              In this description of the lower-layer protocols, a
+
        reassembly.  See Sections 3.3.2 and 3.3.3 for the
−
              message is the unit of transmission in a transport layer
+
        requirements on fragmentation and reassembly.
−
              protocol.  In particular, a TCP segment is a message. A
 
−
              message consists of a transport protocol header followed
 
−
              by application protocol data. To be transmitted end-to-
 
  
 +
      3.2.1.5  Identification: RFC-791 Section 3.2
  
 +
        When sending an identical copy of an earlier datagram, a
 +
        host MAY optionally retain the same Identification field in
 +
        the copy.
  
−
Internet Engineering Task Force                                [Page 17]
+
RFC1122                      INTERNET LAYER                October 1989
  
 +
        DISCUSSION:
 +
              Some Internet protocol experts have maintained that
 +
              when a host sends an identical copy of an earlier
 +
              datagram, the new copy should contain the same
 +
              Identification value as the original.  There are two
 +
              suggested advantages:  (1) if the datagrams are
 +
              fragmented and some of the fragments are lost, the
 +
              receiver may be able to reconstruct a complete datagram
 +
              from fragments of the original and the copies; (2) a
 +
              congested gateway might use the IP Identification field
 +
              (and Fragment Offset) to discard duplicate datagrams
 +
              from the queue.
  
 +
              However, the observed patterns of datagram loss in the
 +
              Internet do not favor the probability of retransmitted
 +
              fragments filling reassembly gaps, while other
 +
              mechanisms (e.g., TCP repacketizing upon
 +
              retransmission) tend to prevent retransmission of an
 +
              identical datagram [IP:9].  Therefore, we believe that
 +
              retransmitting the same Identification field is not
 +
              useful.  Also, a connectionless transport protocol like
 +
              UDP would require the cooperation of the application
 +
              programs to retain the same Identification value in
 +
              identical datagrams.
  
 +
      3.2.1.6  Type-of-Service: RFC-791 Section 3.2
  
−
RFC1122                      INTRODUCTION                  October 1989
+
        The "Type-of-Service" byte in the IP header is divided into
 +
        two sections:  the Precedence field (high-order 3 bits), and
 +
        a field that is customarily called "Type-of-Service" or
 +
        "TOS" (low-order 5 bits).  In this document, all references
 +
        to "TOS" or the "TOS field" refer to the low-order 5 bits
 +
        only.
  
 +
        The Precedence field is intended for Department of Defense
 +
        applications of the Internet protocols.  The use of non-zero
 +
        values in this field is outside the scope of this document
 +
        and the IP standard specification.  Vendors should consult
 +
        the Defense Communication Agency (DCA) for guidance on the
 +
        IP Precedence field and its implications for other protocol
 +
        layers.  However, vendors should note that the use of
 +
        precedence will most likely require that its value be passed
 +
        between protocol layers in just the same way as the TOS
 +
        field is passed.
  
−
              end through the Internet, a message must be encapsulated
+
        The IP layer MUST provide a means for the transport layer to
−
              inside a datagram.
+
        set the TOS field of every datagram that is sent; the
 +
        default is all zero bits. The IP layer SHOULD pass received
  
−
        IP Datagram
+
RFC1122                      INTERNET LAYER                October 1989
−
              An IP datagram is the unit of end-to-end transmission in
 
−
              the IP protocol.  An IP datagram consists of an IP header
 
−
              followed by transport layer data, i.e., of an IP header
 
−
              followed by a message.
 
  
−
              In the description of the internet layer (Section 3), the
+
        TOS values up to the transport layer.
−
              unqualified term "datagram" should be understood to refer
 
−
              to an IP datagram.
 
  
−
         Packet
+
         The particular link-layer mappings of TOS contained in RFC-
−
              A packet is the unit of data passed across the interface
+
        795 SHOULD NOT be implemented.
−
              between the internet layer and the link layer.  It
 
−
              includes an IP header and data.  A packet may be a
 
−
              complete IP datagram or a fragment of an IP datagram.
 
  
−
         Frame
+
         DISCUSSION:
−
               A frame is the unit of transmission in a link layer
+
               While the TOS field has been little used in the past,
−
               protocol, and consists of a link-layer header followed by
+
              it is expected to play an increasing role in the near
−
               a packet.
+
              future.  The TOS field is expected to be used to
 +
              control two aspects of gateway operations: routing and
 +
               queueing algorithms.  See Section 2 of [INTRO:1] for
 +
              the requirements on application programs to specify TOS
 +
               values.
  
−
        Connected Network
+
              The TOS field may also be mapped into link-layer
−
               A network to which a host is interfaced is often known as
+
               service selectors.  This has been applied to provide
−
               the "local network" or the "subnetwork" relative to that
+
               effective sharing of serial lines by different classes
−
               host.  However, these terms can cause confusion, and
+
               of TCP traffic, for example.  However, the mappings
−
               therefore we use the term "connected network" in this
+
               suggested in RFC-795 for networks that were included in
−
               document.
+
               the Internet as of 1981 are now obsolete.
  
−
        Multihomed
+
      3.2.1.7 Time-to-Live: RFC-791 Section 3.2
−
              A host is said to be multihomed if it has multiple IP
 
−
              addresses.  For a discussion of multihoming, see Section
 
−
              3.3.4 below.
 
  
−
         Physical network interface
+
         A host MUST NOT send a datagram with a Time-to-Live (TTL)
−
              This is a physical interface to a connected network and
+
        value of zero.
−
              has a (possibly unique) link-layer address.  Multiple
 
−
              physical network interfaces on a single host may share the
 
−
              same link-layer address, but the address must be unique
 
−
              for different hosts on the same physical network.
 
  
−
         Logical [network] interface
+
         A host MUST NOT discard a datagram just because it was
−
              We define a logical [network] interface to be a logical
+
        received with TTL less than 2.
−
              path, distinguished by a unique IP address, to a connected
 
−
              network.  See Section 3.3.4.
 
  
 +
        The IP layer MUST provide a means for the transport layer to
 +
        set the TTL field of every datagram that is sent.  When a
 +
        fixed TTL value is used, it MUST be configurable.  The
 +
        current suggested value will be published in the "Assigned
 +
        Numbers" RFC.
  
 +
        DISCUSSION:
 +
              The TTL field has two functions: limit the lifetime of
 +
              TCP segments (see RFC-793 [TCP:1], p. 28), and
 +
              terminate Internet routing loops.  Although TTL is a
 +
              time in seconds, it also has some attributes of a hop-
 +
              count, since each gateway is required to reduce the TTL
 +
              field by at least one.
  
 +
              The intent is that TTL expiration will cause a datagram
 +
              to be discarded by a gateway but not by the destination
 +
              host; however, hosts that act as gateways by forwarding
 +
              datagrams must follow the gateway rules for TTL.
  
−
Internet Engineering Task Force                                [Page 18]
+
RFC1122                      INTERNET LAYER                October 1989
  
 +
              A higher-layer protocol may want to set the TTL in
 +
              order to implement an "expanding scope" search for some
 +
              Internet resource.  This is used by some diagnostic
 +
              tools, and is expected to be useful for locating the
 +
              "nearest" server of a given class using IP
 +
              multicasting, for example.  A particular transport
 +
              protocol may also want to specify its own TTL bound on
 +
              maximum datagram lifetime.
  
 +
              A fixed value must be at least big enough for the
 +
              Internet "diameter," i.e., the longest possible path.
 +
              A reasonable value is about twice the diameter, to
 +
              allow for continued Internet growth.
  
 +
      3.2.1.8  Options: RFC-791 Section 3.2
  
−
RFC1122                      INTRODUCTION                  October 1989
+
        There MUST be a means for the transport layer to specify IP
 +
        options to be included in transmitted IP datagrams (see
 +
        Section 3.4).
  
 +
        All IP options (except NOP or END-OF-LIST) received in
 +
        datagrams MUST be passed to the transport layer (or to ICMP
 +
        processing when the datagram is an ICMP message).  The IP
 +
        and transport layer MUST each interpret those IP options
 +
        that they understand and silently ignore the others.
  
−
         Specific-destination address
+
         Later sections of this document discuss specific IP option
−
              This is the effective destination address of a datagram,
+
        support required by each of ICMP, TCP, and UDP.
−
              even if it is broadcast or multicast; see Section 3.2.1.3.
 
  
−
         Path
+
         DISCUSSION:
−
               At a given moment, all the IP datagrams from a particular
+
               Passing all received IP options to the transport layer
−
               source host to a particular destination host will
+
              is a deliberate "violation of strict layering" that is
−
               typically traverse the same sequence of gateways.  We use
+
               designed to ease the introduction of new transport-
−
               the term "path" for this sequence.  Note that a path is
+
               relevant IP options in the future.  Each layer must
−
               uni-directional; it is not unusual to have different paths
+
               pick out any options that are relevant to its own
−
               in the two directions between a given host pair.
+
              processing and ignore the rest.  For this purpose,
 +
               every IP option except NOP and END-OF-LIST will include
 +
               a specification of its own length.
  
−
        MTU
+
              This document does not define the order in which a
−
               The maximum transmission unit, i.e., the size of the
+
              receiver must process multiple options in the same IP
−
               largest packet that can be transmitted.
+
               header. Hosts sending multiple options must be aware
 +
              that this introduces an ambiguity in the meaning of
 +
               certain options when combined with a source-route
 +
              option.
  
 +
        IMPLEMENTATION:
 +
              The IP layer must not crash as the result of an option
  
−
        The terms frame, packet, datagram, message, and segment are
+
RFC1122                      INTERNET LAYER                October 1989
−
        illustrated by the following schematic diagrams:
 
  
−
        A. Transmission on connected network:
+
              length that is outside the possible range. For
−
          _______________________________________________
+
              example, erroneous option lengths have been observed to
−
          | LL hdr | IP hdr |        (data)              |
+
              put some IP implementations into infinite loops.
−
          |________|________|_____________________________|
 
  
−
          <---------- Frame ----------------------------->
+
        Here are the requirements for specific IP options:
−
                    <----------Packet -------------------->
 
  
 +
        (a)  Security Option
  
−
        B. Before IP fragmentation or after IP reassembly:
+
              Some environments require the Security option in every
−
                    ______________________________________
+
              datagram; such a requirement is outside the scope of
−
                  | IP hdr | transport| Application Data |
+
              this document and the IP standard specification. Note,
−
                  |________|____hdr___|__________________|
+
              however, that the security options described in RFC-791
 +
              and RFC-1038 are obsolete.  For DoD applications,
 +
              vendors should consult [IP:8] for guidance.
  
−
                    <-------- Datagram ------------------>
+
        (b) Stream Identifier Option
−
                            <-------- Message ----------->
 
−
          or, for TCP:
 
−
                    ______________________________________
 
−
                  | IP hdr |  TCP hdr | Application Data |
 
−
                  |________|__________|__________________|
 
  
−
                    <--------  Datagram ------------------>
+
              This option is obsolete; it SHOULD NOT be sent, and it
−
                            <-------- Segment ----------->
+
              MUST be silently ignored if received.
  
 +
        (c)  Source Route Options
  
 +
              A host MUST support originating a source route and MUST
 +
              be able to act as the final destination of a source
 +
              route.
  
 +
              If host receives a datagram containing a completed
 +
              source route (i.e., the pointer points beyond the last
 +
              field), the datagram has reached its final destination;
 +
              the option as received (the recorded route) MUST be
 +
              passed up to the transport layer (or to ICMP message
 +
              processing).  This recorded route will be reversed and
 +
              used to form a return source route for reply datagrams
 +
              (see discussion of IP Options in Section 4).  When a
 +
              return source route is built, it MUST be correctly
 +
              formed even if the recorded route included the source
 +
              host (see case (B) in the discussion below).
  
 +
              An IP header containing more than one Source Route
 +
              option MUST NOT be sent; the effect on routing of
 +
              multiple Source Route options is implementation-
 +
              specific.
  
 +
              Section 3.3.5 presents the rules for a host acting as
 +
              an intermediate hop in a source route, i.e., forwarding
  
 +
RFC1122                      INTERNET LAYER                October 1989
  
 +
              a source-routed datagram.
  
−
Internet Engineering Task Force                                [Page 19]
+
              DISCUSSION:
 +
                  If a source-routed datagram is fragmented, each
 +
                  fragment will contain a copy of the source route.
 +
                  Since the processing of IP options (including a
 +
                  source route) must precede reassembly, the
 +
                  original datagram will not be reassembled until
 +
                  the final destination is reached.
  
 +
                  Suppose a source routed datagram is to be routed
 +
                  from host S to host D via gateways G1, G2, ... Gn.
 +
                  There was an ambiguity in the specification over
 +
                  whether the source route option in a datagram sent
 +
                  out by S should be (A) or (B):
  
 +
                      (A):  {>>G2, G3, ... Gn, D}    <--- CORRECT
  
 +
                      (B):  {S, >>G2, G3, ... Gn, D}  <---- WRONG
  
−
RFC1122                      INTRODUCTION                  October 1989
+
                  (where >> represents the pointer).  If (A) is
 +
                  sent, the datagram received at D will contain the
 +
                  option: {G1, G2, ... Gn >>}, with S and D as the
 +
                  IP source and destination addresses.  If (B) were
 +
                  sent, the datagram received at D would again
 +
                  contain S and D as the same IP source and
 +
                  destination addresses, but the option would be:
 +
                  {S, G1, ...Gn >>}; i.e., the originating host
 +
                  would be the first hop in the route.
  
 +
        (d)  Record Route Option
  
−
  1.4  Acknowledgments
+
              Implementation of originating and processing the Record
 +
              Route option is OPTIONAL.
  
−
      This document incorporates contributions and comments from a large
+
        (e) Timestamp Option
−
      group of Internet protocol experts, including representatives of
 
−
      university and research labs, vendors, and government agencies.
 
−
      It was assembled primarily by the Host Requirements Working Group
 
−
      of the Internet Engineering Task Force (IETF).
 
  
−
      The Editor would especially like to acknowledge the tireless
+
              Implementation of originating and processing the
−
      dedication of the following people, who attended many long
+
              Timestamp option is OPTIONAL.  If it is implemented,
−
      meetings and generated 3 million bytes of electronic mail over the
+
              the following rules apply:
−
      past 18 months in pursuit of this document: Philip Almquist, Dave
 
−
      Borman (Cray Research), Noel Chiappa, Dave Crocker (DEC), Steve
 
−
      Deering (Stanford), Mike Karels (Berkeley), Phil Karn (Bellcore),
 
−
      John Lekashman (NASA), Charles Lynn (BBN), Keith McCloghrie (TWG),
 
−
      Paul Mockapetris (ISI), Thomas Narten (Purdue), Craig Partridge
 
−
      (BBN), Drew Perkins (CMU), and James Van Bokkelen (FTP Software).
 
  
−
      In addition, the following people made major contributions to the
+
              o    The originating host MUST record a timestamp in a
−
      effort: Bill Barns (Mitre), Steve Bellovin (AT&T), Mike Brescia
+
                  Timestamp option whose Internet address fields are
−
      (BBN), Ed Cain (DCA), Annette DeSchon (ISI), Martin Gross (DCA),
+
                  not pre-specified or whose first pre-specified
−
      Phill Gross (NRI), Charles Hedrick (Rutgers), Van Jacobson (LBL),
+
                  address is the host's interface address.
−
      John Klensin (MIT), Mark Lottor (SRI), Milo Medin (NASA), Bill
 
−
      Melohn (Sun Microsystems), Greg Minshall (Kinetics), Jeff Mogul
 
−
      (DEC), John Mullen (CMC), Jon Postel (ISI), John Romkey (Epilogue
 
−
      Technology), and Mike StJohns (DCA).  The following also made
 
−
      significant contributions to particular areas: Eric Allman
 
−
      (Berkeley), Rob Austein (MIT), Art Berggreen (ACC), Keith Bostic
 
−
      (Berkeley), Vint Cerf (NRI), Wayne Hathaway (NASA), Matt Korn
 
−
      (IBM), Erik Naggum (Naggum Software, Norway), Robert Ullmann
 
−
      (Prime Computer), David Waitzman (BBN), Frank Wancho (USA), Arun
 
−
      Welch (Ohio State), Bill Westfield (Cisco), and Rayan Zachariassen
 
−
      (Toronto).
 
  
−
      We are grateful to all, including any contributors who may have
+
RFC1122                      INTERNET LAYER                October 1989
−
      been inadvertently omitted from this list.
 
  
 +
              o    The destination host MUST (if possible) add the
 +
                  current timestamp to a Timestamp option before
 +
                  passing the option to the transport layer or to
 +
                  ICMP for processing.
  
 +
              o    A timestamp value MUST follow the rules given in
 +
                  Section 3.2.2.8 for the ICMP Timestamp message.
  
 +
  3.2.2 Internet Control Message Protocol -- ICMP
  
 +
      ICMP messages are grouped into two classes.
  
 +
      *
 +
          ICMP error messages:
  
 +
            Destination Unreachable  (see Section 3.2.2.1)
 +
            Redirect                  (see Section 3.2.2.2)
 +
            Source Quench            (see Section 3.2.2.3)
 +
            Time Exceeded            (see Section 3.2.2.4)
 +
            Parameter Problem        (see Section 3.2.2.5)
  
 +
      *
 +
          ICMP query messages:
  
 +
            Echo                    (see Section 3.2.2.6)
 +
            Information              (see Section 3.2.2.7)
 +
            Timestamp                (see Section 3.2.2.8)
 +
            Address Mask            (see Section 3.2.2.9)
  
 +
      If an ICMP message of unknown type is received, it MUST be
 +
      silently discarded.
  
 +
      Every ICMP error message includes the Internet header and at
 +
      least the first 8 data octets of the datagram that triggered
 +
      the error; more than 8 octets MAY be sent; this header and data
 +
      MUST be unchanged from the received datagram.
  
 +
      In those cases where the Internet layer is required to pass an
 +
      ICMP error message to the transport layer, the IP protocol
 +
      number MUST be extracted from the original header and used to
 +
      select the appropriate transport protocol entity to handle the
 +
      error.
  
 +
      An ICMP error message SHOULD be sent with normal (i.e., zero)
 +
      TOS bits.
  
 +
RFC1122                      INTERNET LAYER                October 1989
  
 +
      An ICMP error message MUST NOT be sent as the result of
 +
      receiving:
  
−
Internet Engineering Task Force                                [Page 20]
+
      *    an ICMP error message, or
  
 +
      *    a datagram destined to an IP broadcast or IP multicast
 +
          address, or
  
 +
      *    a datagram sent as a link-layer broadcast, or
  
 +
      *    a non-initial fragment, or
  
−
RFC1122                        LINK LAYER                  October 1989
+
      *    a datagram whose source address does not define a single
 +
          host -- e.g., a zero address, a loopback address, a
 +
          broadcast address, a multicast address, or a Class E
 +
          address.
  
 +
      NOTE: THESE RESTRICTIONS TAKE PRECEDENCE OVER ANY REQUIREMENT
 +
      ELSEWHERE IN THIS DOCUMENT FOR SENDING ICMP ERROR MESSAGES.
  
−
2. LINK LAYER
+
      DISCUSSION:
 +
          These rules will prevent the "broadcast storms" that have
 +
          resulted from hosts returning ICMP error messages in
 +
          response to broadcast datagrams.  For example, a broadcast
 +
          UDP segment to a non-existent port could trigger a flood
 +
          of ICMP Destination Unreachable datagrams from all
 +
          machines that do not have a client for that destination
 +
          port.  On a large Ethernet, the resulting collisions can
 +
          render the network useless for a second or more.
  
−
  2.1 INTRODUCTION
+
          Every datagram that is broadcast on the connected network
 +
          should have a valid IP broadcast address as its IP
 +
          destination (see Section 3.3.6).  However, some hosts
 +
          violate this rule.  To be certain to detect broadcast
 +
          datagrams, therefore, hosts are required to check for a
 +
          link-layer broadcast as well as an IP-layer broadcast
 +
          address.
  
−
       All Internet systems, both hosts and gateways, have the same
+
       IMPLEMENTATION:
−
      requirements for link layer protocols.  These requirements are
+
          This requires that the link layer inform the IP layer when
−
      given in Chapter 3 of "Requirements for Internet Gateways"
+
          a link-layer broadcast datagram has been received; see
−
      [INTRO:2], augmented with the material in this section.
+
          Section 2.4.
  
−
  2.2  PROTOCOL WALK-THROUGH
+
      3.2.2.1 Destination Unreachable: RFC-792
  
−
      None.
+
        The following additional codes are hereby defined:
  
−
  2.3  SPECIFIC ISSUES
+
                6 = destination network unknown
  
−
      2.3.1  Trailer Protocol Negotiation
+
RFC1122                      INTERNET LAYER                October 1989
  
−
        The trailer protocol [LINK:1] for link-layer encapsulation MAY
+
                7 = destination host unknown
−
        be used, but only when it has been verified that both systems
 
−
        (host or gateway) involved in the link-layer communication
 
−
        implement trailers.  If the system does not dynamically
 
−
        negotiate use of the trailer protocol on a per-destination
 
−
        basis, the default configuration MUST disable the protocol.
 
  
−
        DISCUSSION:
+
                8 = source host isolated
−
              The trailer protocol is a link-layer encapsulation
 
−
              technique that rearranges the data contents of packets
 
−
              sent on the physical network.  In some cases, trailers
 
−
              improve the throughput of higher layer protocols by
 
−
              reducing the amount of data copying within the operating
 
−
              system.  Higher layer protocols are unaware of trailer
 
−
              use, but both the sending and receiving host MUST
 
−
              understand the protocol if it is used.
 
  
−
              Improper use of trailers can result in very confusing
+
                9 = communication with destination network
−
              symptoms.  Only packets with specific size attributes are
+
                        administratively prohibited
−
              encapsulated using trailers, and typically only a small
 
−
              fraction of the packets being exchanged have these
 
−
              attributes.  Thus, if a system using trailers exchanges
 
−
              packets with a system that does not, some packets
 
−
              disappear into a black hole while others are delivered
 
−
              successfully.
 
  
−
        IMPLEMENTATION:
+
                10 = communication with destination host
−
              On an Ethernet, packets encapsulated with trailers use a
+
                        administratively prohibited
−
              distinct Ethernet type [LINK:1], and trailer negotiation
 
−
              is performed at the time that ARP is used to discover the
 
−
              link-layer address of a destination system.
 
  
 +
                11 = network unreachable for type of service
  
 +
                12 = host unreachable for type of service
  
−
Internet Engineering Task Force                                [Page 21]
+
        A host SHOULD generate Destination Unreachable messages with
 +
        code:
  
 +
        2    (Protocol Unreachable), when the designated transport
 +
              protocol is not supported; or
  
 +
        3    (Port Unreachable), when the designated transport
 +
              protocol (e.g., UDP) is unable to demultiplex the
 +
              datagram but has no protocol mechanism to inform the
 +
              sender.
  
 +
        A Destination Unreachable message that is received MUST be
 +
        reported to the transport layer.  The transport layer SHOULD
 +
        use the information appropriately; for example, see Sections
 +
        4.1.3.3, 4.2.3.9, and 4.2.4 below.  A transport protocol
 +
        that has its own mechanism for notifying the sender that a
 +
        port is unreachable (e.g., TCP, which sends RST segments)
 +
        MUST nevertheless accept an ICMP Port Unreachable for the
 +
        same purpose.
  
−
RFC1122                        LINK LAYER                  October 1989
+
        A Destination Unreachable message that is received with code
 +
        0 (Net), 1 (Host), or 5 (Bad Source Route) may result from a
 +
        routing transient and MUST therefore be interpreted as only
 +
        a hint, not proof, that the specified destination is
 +
        unreachable [IP:11].  For example, it MUST NOT be used as
 +
        proof of a dead gateway (see Section 3.3.1).
  
 +
      3.2.2.2  Redirect: RFC-792
  
−
              Specifically, the ARP exchange is completed in the usual
+
        A host SHOULD NOT send an ICMP Redirect message; Redirects
−
              manner using the normal IP protocol type, but a host that
+
        are to be sent only by gateways.
−
              wants to speak trailers will send an additional "trailer
 
−
              ARP reply" packet, i.e., an ARP reply that specifies the
 
−
              trailer encapsulation protocol type but otherwise has the
 
−
              format of a normal ARP reply.  If a host configured to use
 
−
              trailers receives a trailer ARP reply message from a
 
−
              remote machine, it can add that machine to the list of
 
−
              machines that understand trailers, e.g., by marking the
 
−
              corresponding entry in the ARP cache.
 
  
−
              Hosts wishing to receive trailer encapsulations send
+
        A host receiving a Redirect message MUST update its routing
−
              trailer ARP replies whenever they complete exchanges of
+
        information accordingly.  Every host MUST be prepared to
−
              normal ARP messages for IP.  Thus, a host that received an
 
−
              ARP request for its IP protocol address would send a
 
−
              trailer ARP reply in addition to the normal IP ARP reply;
 
−
              a host that sent the IP ARP request would send a trailer
 
−
              ARP reply when it received the corresponding IP ARP reply.
 
−
              In this way, either the requesting or responding host in
 
−
              an IP ARP exchange may request that it receive trailer
 
−
              encapsulations.
 
  
−
              This scheme, using extra trailer ARP reply packets rather
+
RFC1122                      INTERNET LAYER                October 1989
−
              than sending an ARP request for the trailer protocol type,
+
 
−
              was designed to avoid a continuous exchange of ARP packets
+
        accept both Host and Network Redirects and to process them
−
              with a misbehaving host that, contrary to any
+
        as described in Section 3.3.1.2 below.
−
              specification or common sense, responded to an ARP reply
+
 
−
              for trailers with another ARP reply for IP. This problem
+
        A Redirect message SHOULD be silently discarded if the new
−
               is avoided by sending a trailer ARP reply in response to
+
        gateway address it specifies is not on the same connected
−
               an IP ARP reply only when the IP ARP reply answers an
+
        (sub-) net through which the Redirect arrived [INTRO:2,
−
               outstanding request; this is true when the hardware
+
        Appendix A], or if the source of the Redirect is not the
−
               address for the host is still unknown when the IP ARP
+
        current first-hop gateway for the specified destination (see
−
               reply is received.  A trailer ARP reply may always be sent
+
        Section 3.3.1).
−
               along with an IP ARP reply responding to an IP ARP
+
 
−
               request.
+
      3.2.2.3  Source Quench: RFC-792
 +
 
 +
        A host MAY send a Source Quench message if it is
 +
        approaching, or has reached, the point at which it is forced
 +
        to discard incoming datagrams due to a shortage of
 +
        reassembly buffers or other resources.  See Section 2.2.3 of
 +
        [INTRO:2] for suggestions on when to send Source Quench.
 +
 
 +
        If a Source Quench message is received, the IP layer MUST
 +
        report it to the transport layer (or ICMP processing). In
 +
        general, the transport or application layer SHOULD implement
 +
        a mechanism to respond to Source Quench for any protocol
 +
        that can send a sequence of datagrams to the same
 +
        destination and which can reasonably be expected to maintain
 +
        enough state information to make this feasible.  See Section
 +
        4 for the handling of Source Quench by TCP and UDP.
 +
 
 +
        DISCUSSION:
 +
               A Source Quench may be generated by the target host or
 +
               by some gateway in the path of a datagram.  The host
 +
               receiving a Source Quench should throttle itself back
 +
               for a period of time, then gradually increase the
 +
               transmission rate again.  The mechanism to respond to
 +
              Source Quench may be in the transport layer (for
 +
               connection-oriented protocols like TCP) or in the
 +
              application layer (for protocols that are built on top
 +
               of UDP).
  
−
      2.3.2  Address Resolution Protocol -- ARP
+
              A mechanism has been proposed [IP:14] to make the IP
 +
              layer respond directly to Source Quench by controlling
 +
              the rate at which datagrams are sent, however, this
 +
              proposal is currently experimental and not currently
 +
              recommended.
  
−
        2.3.2.1 ARP Cache Validation
+
      3.2.2.4 Time Exceeded: RFC-792
  
−
            An implementation of the Address Resolution Protocol (ARP)
+
        An incoming Time Exceeded message MUST be passed to the
−
            [LINK:2] MUST provide a mechanism to flush out-of-date cache
+
        transport layer.
−
            entries.  If this mechanism involves a timeout, it SHOULD be
 
−
            possible to configure the timeout value.
 
  
−
            A mechanism to prevent ARP flooding (repeatedly sending an
+
RFC1122                      INTERNET LAYER                October 1989
−
            ARP Request for the same IP address, at a high rate) MUST be
 
−
            included.  The recommended maximum rate is 1 per second per
 
  
 +
        DISCUSSION:
 +
              A gateway will send a Time Exceeded Code 0 (In Transit)
 +
              message when it discards a datagram due to an expired
 +
              TTL field.  This indicates either a gateway routing
 +
              loop or too small an initial TTL value.
  
 +
              A host may receive a Time Exceeded Code 1 (Reassembly
 +
              Timeout) message from a destination host that has timed
 +
              out and discarded an incomplete datagram; see Section
 +
              3.3.2 below.  In the future, receipt of this message
 +
              might be part of some "MTU discovery" procedure, to
 +
              discover the maximum datagram size that can be sent on
 +
              the path without fragmentation.
  
−
Internet Engineering Task Force                                [Page 22]
+
      3.2.2.5  Parameter Problem: RFC-792
  
 +
        A host SHOULD generate Parameter Problem messages.  An
 +
        incoming Parameter Problem message MUST be passed to the
 +
        transport layer, and it MAY be reported to the user.
  
 +
        DISCUSSION:
 +
              The ICMP Parameter Problem message is sent to the
 +
              source host for any problem not specifically covered by
 +
              another ICMP message.  Receipt of a Parameter Problem
 +
              message generally indicates some local or remote
 +
              implementation error.
  
 +
        A new variant on the Parameter Problem message is hereby
 +
        defined:
 +
          Code 1 = required option is missing.
  
−
RFC1122                        LINK LAYER                  October 1989
+
        DISCUSSION:
 +
              This variant is currently in use in the military
 +
              community for a missing security option.
  
 +
      3.2.2.6  Echo Request/Reply: RFC-792
  
−
            destination.
+
        Every host MUST implement an ICMP Echo server function that
 +
        receives Echo Requests and sends corresponding Echo Replies.
 +
        A host SHOULD also implement an application-layer interface
 +
        for sending an Echo Request and receiving an Echo Reply, for
 +
        diagnostic purposes.
  
−
            DISCUSSION:
+
        An ICMP Echo Request destined to an IP broadcast or IP
−
                The ARP specification [LINK:2] suggests but does not
+
        multicast address MAY be silently discarded.
−
                require a timeout mechanism to invalidate cache entries
 
−
                when hosts change their Ethernet addresses.  The
 
−
                prevalence of proxy ARP (see Section 2.4 of [INTRO:2])
 
−
                has significantly increased the likelihood that cache
 
−
                entries in hosts will become invalid, and therefore
 
−
                some ARP-cache invalidation mechanism is now required
 
−
                for hosts.  Even in the absence of proxy ARP, a long-
 
−
                period cache timeout is useful in order to
 
−
                automatically correct any bad ARP data that might have
 
−
                been cached.
 
  
−
            IMPLEMENTATION:
+
RFC1122                      INTERNET LAYER                October 1989
−
                Four mechanisms have been used, sometimes in
 
−
                combination, to flush out-of-date cache entries.
 
  
−
                (1)  Timeout -- Periodically time out cache entries,
+
        DISCUSSION:
−
                      even if they are in use.  Note that this timeout
+
              This neutral provision results from a passionate debate
−
                      should be restarted when the cache entry is
+
              between those who feel that ICMP Echo to a broadcast
−
                      "refreshed" (by observing the source fields,
+
              address provides a valuable diagnostic capability and
−
                      regardless of target address, of an ARP broadcast
+
              those who feel that misuse of this feature can too
−
                      from the system in question).  For proxy ARP
+
              easily create packet storms.
−
                      situations, the timeout needs to be on the order
 
−
                      of a minute.
 
  
−
                (2)  Unicast Poll -- Actively poll the remote host by
+
        The IP source address in an ICMP Echo Reply MUST be the same
−
                      periodically sending a point-to-point ARP Request
+
        as the specific-destination address (defined in Section
−
                      to it, and delete the entry if no ARP Reply is
+
        3.2.1.3) of the corresponding ICMP Echo Request message.
−
                      received from N successive polls. Again, the
 
−
                      timeout should be on the order of a minute, and
 
−
                      typically N is 2.
 
  
−
                (3) Link-Layer Advice -- If the link-layer driver
+
        Data received in an ICMP Echo Request MUST be entirely
−
                      detects a delivery problem, flush the
+
        included in the resulting Echo Reply. However, if sending
−
                      corresponding ARP cache entry.
+
        the Echo Reply requires intentional fragmentation that is
 +
        not implemented, the datagram MUST be truncated to maximum
 +
        transmission size (see Section 3.3.3) and sent.
  
−
                (4)  Higher-layer Advice -- Provide a call from the
+
        Echo Reply messages MUST be passed to the ICMP user
−
                      Internet layer to the link layer to indicate a
+
        interface, unless the corresponding Echo Request originated
−
                      delivery problem.  The effect of this call would
+
        in the IP layer.
−
                      be to invalidate the corresponding cache entry.
 
−
                      This call would be analogous to the
 
−
                      "ADVISE_DELIVPROB()" call from the transport layer
 
−
                      to the Internet layer (see Section 3.4), and in
 
−
                      fact the ADVISE_DELIVPROB routine might in turn
 
−
                      call the link-layer advice routine to invalidate
 
  
 +
        If a Record Route and/or Time Stamp option is received in an
 +
        ICMP Echo Request, this option (these options) SHOULD be
 +
        updated to include the current host and included in the IP
 +
        header of the Echo Reply message, without "truncation".
 +
        Thus, the recorded route will be for the entire round trip.
  
 +
        If a Source Route option is received in an ICMP Echo
 +
        Request, the return route MUST be reversed and used as a
 +
        Source Route option for the Echo Reply message.
  
−
Internet Engineering Task Force                                [Page 23]
+
      3.2.2.7  Information Request/Reply: RFC-792
  
 +
        A host SHOULD NOT implement these messages.
  
 +
        DISCUSSION:
 +
              The Information Request/Reply pair was intended to
 +
              support self-configuring systems such as diskless
 +
              workstations, to allow them to discover their IP
 +
              network numbers at boot time.  However, the RARP and
 +
              BOOTP protocols provide better mechanisms for a host to
 +
              discover its own IP address.
  
 +
      3.2.2.8  Timestamp and Timestamp Reply: RFC-792
  
−
RFC1122                        LINK LAYER                  October 1989
+
        A host MAY implement Timestamp and Timestamp Reply.  If they
 +
        are implemented, the following rules MUST be followed.
  
 +
RFC1122                      INTERNET LAYER                October 1989
  
−
                      the ARP cache entry.
+
        o    The ICMP Timestamp server function returns a Timestamp
 +
              Reply to every Timestamp message that is received.  If
 +
              this function is implemented, it SHOULD be designed for
 +
              minimum variability in delay (e.g., implemented in the
 +
              kernel to avoid delay in scheduling a user process).
  
−
                Approaches (1) and (2) involve ARP cache timeouts on
+
        The following cases for Timestamp are to be handled
−
                the order of a minute or less.  In the absence of proxy
+
        according to the corresponding rules for ICMP Echo:
−
                ARP, a timeout this short could create noticeable
 
−
                overhead traffic on a very large Ethernet.  Therefore,
 
−
                it may be necessary to configure a host to lengthen the
 
−
                ARP cache timeout.
 
  
−
         2.3.2.2  ARP Packet Queue
+
         o    An ICMP Timestamp Request message to an IP broadcast or
 +
              IP multicast address MAY be silently discarded.
  
−
            The link layer SHOULD save (rather than discard) at least
+
        o    The IP source address in an ICMP Timestamp Reply MUST
−
            one (the latest) packet of each set of packets destined to
+
              be the same as the specific-destination address of the
−
            the same unresolved IP address, and transmit the saved
+
              corresponding Timestamp Request message.
−
            packet when the address has been resolved.
 
  
−
            DISCUSSION:
+
        o    If a Source-route option is received in an ICMP Echo
−
                Failure to follow this recommendation causes the first
+
              Request, the return route MUST be reversed and used as
−
                packet of every exchange to be lost.  Although higher-
+
              a Source Route option for the Timestamp Reply message.
−
                layer protocols can generally cope with packet loss by
 
−
                retransmission, packet loss does impact performance.
 
−
                For example, loss of a TCP open request causes the
 
−
                initial round-trip time estimate to be inflated.  UDP-
 
−
                based applications such as the Domain Name System are
 
−
                more seriously affected.
 
  
−
      2.3.3  Ethernet and IEEE 802 Encapsulation
+
        o    If a Record Route and/or Timestamp option is received
 +
              in a Timestamp Request, this (these) option(s) SHOULD
 +
              be updated to include the current host and included in
 +
              the IP header of the Timestamp Reply message.
  
−
         The IP encapsulation for Ethernets is described in RFC-894
+
         o    Incoming Timestamp Reply messages MUST be passed up to
−
        [LINK:3], while RFC-1042 [LINK:4] describes the IP
+
              the ICMP user interface.
−
        encapsulation for IEEE 802 networks.  RFC-1042 elaborates and
 
−
        replaces the discussion in Section 3.4 of [INTRO:2].
 
  
−
         Every Internet host connected to a 10Mbps Ethernet cable:
+
         The preferred form for a timestamp value (the "standard
 +
        value") is in units of milliseconds since midnight Universal
 +
        Time.  However, it may be difficult to provide this value
 +
        with millisecond resolution.  For example, many systems use
 +
        clocks that update only at line frequency, 50 or 60 times
 +
        per second.  Therefore, some latitude is allowed in a
 +
        "standard value":
  
−
         o    MUST be able to send and receive packets using RFC-894
+
         (a)  A "standard value" MUST be updated at least 15 times
−
               encapsulation;
+
              per second (i.e., at most the six low-order bits of the
 +
               value may be undefined).
  
−
         o    SHOULD be able to receive RFC-1042 packets, intermixed
+
         (b)  The accuracy of a "standard value" MUST approximate
−
               with RFC-894 packets; and
+
              that of operator-set CPU clocks, i.e., correct within a
 +
               few minutes.
  
−
        o    MAY be able to send packets using RFC-1042 encapsulation.
+
RFC1122                      INTERNET LAYER                October 1989
  
 +
      3.2.2.9  Address Mask Request/Reply: RFC-950
  
−
         An Internet host that implements sending both the RFC-894 and
+
         A host MUST support the first, and MAY implement all three,
−
         the RFC-1042 encapsulations MUST provide a configuration switch
+
         of the following methods for determining the address mask(s)
−
         to select which is sent, and this switch MUST default to RFC-
+
         corresponding to its IP address(es):
−
        894.
 
  
 +
        (1)  static configuration information;
  
 +
        (2)  obtaining the address mask(s) dynamically as a side-
 +
              effect of the system initialization process (see
 +
              [INTRO:1]); and
  
−
Internet Engineering Task Force                                [Page 24]
+
        (3)  sending ICMP Address Mask Request(s) and receiving ICMP
 +
              Address Mask Reply(s).
  
 +
        The choice of method to be used in a particular host MUST be
 +
        configurable.
  
 +
        When method (3), the use of Address Mask messages, is
 +
        enabled, then:
  
 +
        (a)  When it initializes, the host MUST broadcast an Address
 +
              Mask Request message on the connected network
 +
              corresponding to the IP address.  It MUST retransmit
 +
              this message a small number of times if it does not
 +
              receive an immediate Address Mask Reply.
  
−
RFC1122                        LINK LAYER                  October 1989
+
        (b)  Until it has received an Address Mask Reply, the host
 +
              SHOULD assume a mask appropriate for the address class
 +
              of the IP address, i.e., assume that the connected
 +
              network is not subnetted.
  
 +
        (c)  The first Address Mask Reply message received MUST be
 +
              used to set the address mask corresponding to the
 +
              particular local IP address.  This is true even if the
 +
              first Address Mask Reply message is "unsolicited", in
 +
              which case it will have been broadcast and may arrive
 +
              after the host has ceased to retransmit Address Mask
 +
              Requests.  Once the mask has been set by an Address
 +
              Mask Reply, later Address Mask Reply messages MUST be
 +
              (silently) ignored.
  
−
         Note that the standard IP encapsulation in RFC-1042 does not
+
         Conversely, if Address Mask messages are disabled, then no
−
         use the protocol id value (K1=6) that IEEE reserved for IP;
+
         ICMP Address Mask Requests will be sent, and any ICMP
−
         instead, it uses a value (K1=170) that implies an extension
+
         Address Mask Replies received for that local IP address MUST
−
         (the "SNAP") which can be used to hold the Ether-Type field.
+
         be (silently) ignored.
−
        An Internet system MUST NOT send 802 packets using K1=6.
 
  
−
         Address translation from Internet addresses to link-layer
+
         A host SHOULD make some reasonableness check on any address
−
        addresses on Ethernet and IEEE 802 networks MUST be managed by
 
−
        the Address Resolution Protocol (ARP).
 
  
−
        The MTU for an Ethernet is 1500 and for 802.3 is 1492.
+
RFC1122                      INTERNET LAYER                October 1989
  
−
         DISCUSSION:
+
         mask it installs; see IMPLEMENTATION section below.
−
              The IEEE 802.3 specification provides for operation over a
 
−
              10Mbps Ethernet cable, in which case Ethernet and IEEE
 
−
              802.3 frames can be physically intermixed.  A receiver can
 
−
              distinguish Ethernet and 802.3 frames by the value of the
 
−
              802.3 Length field; this two-octet field coincides in the
 
−
              header with the Ether-Type field of an Ethernet frame.  In
 
−
              particular, the 802.3 Length field must be less than or
 
−
              equal to 1500, while all valid Ether-Type values are
 
−
              greater than 1500.
 
  
−
              Another compatibility problem arises with link-layer
+
        A system MUST NOT send an Address Mask Reply unless it is an
−
              broadcasts.  A broadcast sent with one framing will not be
+
        authoritative agent for address masks.  An authoritative
−
              seen by hosts that can receive only the other framing.
+
        agent may be a host or a gateway, but it MUST be explicitly
 +
        configured as a address mask agent.  Receiving an address
 +
        mask via an Address Mask Reply does not give the receiver
 +
        authority and MUST NOT be used as the basis for issuing
 +
        Address Mask Replies.
  
−
              The provisions of this section were designed to provide
+
        With a statically configured address mask, there SHOULD be
−
              direct interoperation between 894-capable and 1042-capable
+
        an additional configuration flag that determines whether the
−
              systems on the same cable, to the maximum extent possible.
+
        host is to act as an authoritative agent for this mask,
−
              It is intended to support the present situation where
+
        i.e., whether it will answer Address Mask Request messages
−
              894-only systems predominate, while providing an easy
+
        using this mask.
−
              transition to a possible future in which 1042-capable
 
−
              systems become common.
 
  
−
              Note that 894-only systems cannot interoperate directly
+
        If it is configured as an agent, the host MUST broadcast an
−
              with 1042-only systems.  If the two system types are set
+
        Address Mask Reply for the mask on the appropriate interface
−
              up as two different logical networks on the same cable,
+
        when it initializes.
−
              they can communicate only through an IP gateway.
 
−
              Furthermore, it is not useful or even possible for a
 
−
              dual-format host to discover automatically which format to
 
−
              send, because of the problem of link-layer broadcasts.
 
  
−
  2.4  LINK/INTERNET LAYER INTERFACE
+
        See "System Initialization" in [INTRO:1] for more
 +
        information about the use of Address Mask Request/Reply
 +
        messages.
  
−
      The packet receive interface between the IP layer and the link
+
        DISCUSSION
−
      layer MUST include a flag to indicate whether the incoming packet
+
              Hosts that casually send Address Mask Replies with
−
      was addressed to a link-layer broadcast address.
+
              invalid address masks have often been a serious
 +
              nuisance.  To prevent this, Address Mask Replies ought
 +
              to be sent only by authoritative agents that have been
 +
              selected by explicit administrative action.
  
 +
              When an authoritative agent receives an Address Mask
 +
              Request message, it will send a unicast Address Mask
 +
              Reply to the source IP address.  If the network part of
 +
              this address is zero (see (a) and (b) in 3.2.1.3), the
 +
              Reply will be broadcast.
  
 +
              Getting no reply to its Address Mask Request messages,
 +
              a host will assume there is no agent and use an
 +
              unsubnetted mask, but the agent may be only temporarily
 +
              unreachable.  An agent will broadcast an unsolicited
 +
              Address Mask Reply whenever it initializes, in order to
 +
              update the masks of all hosts that have initialized in
 +
              the meantime.
  
−
Internet Engineering Task Force                                [Page 25]
+
        IMPLEMENTATION:
 +
              The following reasonableness check on an address mask
 +
              is suggested: the mask is not all 1 bits, and it is
  
 +
RFC1122                      INTERNET LAYER                October 1989
  
 +
              either zero or else the 8 highest-order bits are on.
  
 +
  3.2.3  Internet Group Management Protocol IGMP
  
−
RFC1122                        LINK LAYER                  October 1989
+
      IGMP [IP:4] is a protocol used between hosts and gateways on a
 +
      single network to establish hosts' membership in particular
 +
      multicast groups.  The gateways use this information, in
 +
      conjunction with a multicast routing protocol, to support IP
 +
      multicasting across the Internet.
  
 +
      At this time, implementation of IGMP is OPTIONAL; see Section
 +
      3.3.7 for more information.  Without IGMP, a host can still
 +
      participate in multicasting local to its connected networks.
  
−
      DISCUSSION
+
3.3 SPECIFIC ISSUES
−
          Although the IP layer does not generally know link layer
 
−
          addresses (since every different network medium typically has
 
−
          a different address format), the broadcast address on a
 
−
          broadcast-capable medium is an important special case.  See
 
−
          Section 3.2.2, especially the DISCUSSION concerning broadcast
 
−
          storms.
 
  
−
      The packet send interface between the IP and link layers MUST
+
  3.3.1 Routing Outbound Datagrams
−
      include the 5-bit TOS field (see Section 3.2.1.6).
 
  
−
       The link layer MUST NOT report a Destination Unreachable error to
+
       The IP layer chooses the correct next hop for each datagram it
−
       IP solely because there is no ARP cache entry for a destination.
+
      sends.  If the destination is on a connected network, the
 +
      datagram is sent directly to the destination host; otherwise,
 +
       it has to be routed to a gateway on a connected network.
  
−
  2.5 LINK LAYER REQUIREMENTS SUMMARY
+
      3.3.1.1 Local/Remote Decision
  
−
                                                  |      | | | |S| |
+
        To decide if the destination is on a connected network, the
−
                                                  |      | | | |H| |F
+
        following algorithm MUST be used [see IP:3]:
−
                                                  |      | | | |O|M|o
+
 
−
                                                  |      | |S| |U|U|o
+
        (a)  The address mask (particular to a local IP address for
−
                                                  |      | |H| |L|S|t
+
              a multihomed host) is a 32-bit mask that selects the
−
                                                  |      |M|O| |D|T|n
+
              network number and subnet number fields of the
−
                                                  |      |U|U|M| | |o
+
              corresponding IP address.
−
                                                  |      |S|L|A|N|N|t
 
−
                                                  |      |T|D|Y|O|O|t
 
−
FEATURE                                          |SECTION| | | |T|T|e
 
−
--------------------------------------------------|-------|-|-|-|-|-|--
 
−
                                                  |      | | | | | |
 
−
Trailer encapsulation                            |2.3.1  | | |x| | |
 
−
Send Trailers by default without negotiation      |2.3.1  | | | | |x|
 
−
ARP                                              |2.3.2  | | | | | |
 
−
  Flush out-of-date ARP cache entries            |2.3.2.1|x| | | | |
 
−
  Prevent ARP floods                              |2.3.2.1|x| | | | |
 
−
  Cache timeout configurable                      |2.3.2.1| |x| | | |
 
−
  Save at least one (latest) unresolved pkt      |2.3.2.2| |x| | | |
 
−
Ethernet and IEEE 802 Encapsulation              |2.3.3 | | | | | |
 
−
  Host able to:                                  |2.3.3  | | | | | |
 
−
    Send & receive RFC-894 encapsulation          |2.3.3  |x| | | | |
 
−
    Receive RFC-1042 encapsulation                |2.3.3  | |x| | | |
 
−
    Send RFC-1042 encapsulation                  |2.3.3  | | |x| | |
 
−
      Then config. sw. to select, RFC-894 dflt    |2.3.3  |x| | | | |
 
−
  Send K1=6 encapsulation                        |2.3.3  | | | | |x|
 
−
  Use ARP on Ethernet and IEEE 802 nets          |2.3.3  |x| | | | |
 
−
Link layer report b'casts to IP layer            |2.4    |x| | | | |
 
−
IP layer pass TOS to link layer                  |2.4    |x| | | | |
 
−
No ARP cache entry treated as Dest. Unreach.      |2.4    | | | | |x|
 
  
 +
        (b)  If the IP destination address bits extracted by the
 +
              address mask match the IP source address bits extracted
 +
              by the same mask, then the destination is on the
 +
              corresponding connected network, and the datagram is to
 +
              be transmitted directly to the destination host.
  
 +
        (c)  If not, then the destination is accessible only through
 +
              a gateway.  Selection of a gateway is described below
 +
              (3.3.1.2).
  
 +
        A special-case destination address is handled as follows:
  
 +
        *    For a limited broadcast or a multicast address, simply
 +
              pass the datagram to the link layer for the appropriate
 +
              interface.
  
−
Internet Engineering Task Force                                [Page 26]
+
RFC1122                      INTERNET LAYER                October 1989
  
 +
        *    For a (network or subnet) directed broadcast, the
 +
              datagram can use the standard routing algorithms.
  
 +
        The host IP layer MUST operate correctly in a minimal
 +
        network environment, and in particular, when there are no
 +
        gateways.  For example, if the IP layer of a host insists on
 +
        finding at least one gateway to initialize, the host will be
 +
        unable to operate on a single isolated broadcast net.
  
 +
      3.3.1.2  Gateway Selection
  
−
RFC1122                      INTERNET LAYER                October 1989
+
        To efficiently route a series of datagrams to the same
 +
        destination, the source host MUST keep a "route cache" of
 +
        mappings to next-hop gateways.  A host uses the following
 +
        basic algorithm on this cache to route a datagram; this
 +
        algorithm is designed to put the primary routing burden on
 +
        the gateways [IP:11].
  
 +
        (a)  If the route cache contains no information for a
 +
              particular destination, the host chooses a "default"
 +
              gateway and sends the datagram to it.  It also builds a
 +
              corresponding Route Cache entry.
  
−
3. INTERNET LAYER PROTOCOLS
+
        (b)  If that gateway is not the best next hop to the
 +
              destination, the gateway will forward the datagram to
 +
              the best next-hop gateway and return an ICMP Redirect
 +
              message to the source host.
  
−
  3.1 INTRODUCTION
+
        (c)  When it receives a Redirect, the host updates the
 +
              next-hop gateway in the appropriate route cache entry,
 +
              so later datagrams to the same destination will go
 +
              directly to the best gateway.
  
−
      The Robustness Principle: "Be liberal in what you accept, and
+
        Since the subnet mask appropriate to the destination address
−
      conservative in what you send" is particularly important in the
+
        is generally not known, a Network Redirect message SHOULD be
−
      Internet layer, where one misbehaving host can deny Internet
+
        treated identically to a Host Redirect message; i.e., the
−
      service to many other hosts.
+
        cache entry for the destination host (only) would be updated
 +
        (or created, if an entry for that host did not exist) for
 +
        the new gateway.
  
−
      The protocol standards used in the Internet layer are:
+
        DISCUSSION:
 +
              This recommendation is to protect against gateways that
 +
              erroneously send Network Redirects for a subnetted
 +
              network, in violation of the gateway requirements
 +
              [INTRO:2].
  
−
      o    RFC-791 [IP:1] defines the IP protocol and gives an
+
        When there is no route cache entry for the destination host
−
          introduction to the architecture of the Internet.
+
        address (and the destination is not on the connected
  
−
      o    RFC-792 [IP:2] defines ICMP, which provides routing,
+
RFC1122                      INTERNET LAYER                October 1989
−
          diagnostic and error functionality for IP.  Although ICMP
 
−
          messages are encapsulated within IP datagrams, ICMP
 
−
          processing is considered to be (and is typically implemented
 
−
          as) part of the IP layer.  See Section 3.2.2.
 
  
−
      o    RFC-950 [IP:3] defines the mandatory subnet extension to the
+
        network), the IP layer MUST pick a gateway from its list of
−
          addressing architecture.
+
        "default" gateways.  The IP layer MUST support multiple
 +
        default gateways.
  
−
      o    RFC-1112 [IP:4] defines the Internet Group Management
+
        As an extra feature, a host IP layer MAY implement a table
−
          Protocol IGMP, as part of a recommended extension to hosts
+
        of "static routes".  Each such static route MAY include a
−
          and to the host-gateway interface to support Internet-wide
+
        flag specifying whether it may be overridden by ICMP
−
          multicasting at the IP level.  See Section 3.2.3.
+
        Redirects.
  
−
          The target of an IP multicast may be an arbitrary group of
+
        DISCUSSION:
−
          Internet hosts.  IP multicasting is designed as a natural
+
              A host generally needs to know at least one default
−
          extension of the link-layer multicasting facilities of some
+
              gateway to get started.  This information can be
−
          networks, and it provides a standard means for local access
+
              obtained from a configuration file or else from the
−
          to such link-layer multicasting facilities.
+
              host startup sequence, e.g., the BOOTP protocol (see
 +
              [INTRO:1]).
  
−
      Other important references are listed in Section 5 of this
+
              It has been suggested that a host can augment its list
−
      document.
+
              of default gateways by recording any new gateways it
 +
              learns about.  For example, it can record every gateway
 +
              to which it is ever redirected.  Such a feature, while
 +
              possibly useful in some circumstances, may cause
 +
              problems in other cases (e.g., gateways are not all
 +
              equal), and it is not recommended.
  
−
      The Internet layer of host software MUST implement both IP and
+
              A static route is typically a particular preset mapping
−
      ICMP.  See Section 3.3.7 for the requirements on support of IGMP.
+
              from destination host or network into a particular
 +
              next-hop gateway; it might also depend on the Type-of-
 +
              Service (see next section).  Static routes would be set
 +
              up by system administrators to override the normal
 +
              automatic routing mechanism, to handle exceptional
 +
              situations. However, any static routing information is
 +
              a potential source of failure as configurations change
 +
              or equipment fails.
  
−
       The host IP layer has two basic functions:  (1) choose the "next
+
       3.3.1.3 Route Cache
−
      hop" gateway or host for outgoing IP datagrams and (2) reassemble
 
−
      incoming IP datagrams. The IP layer may also (3) implement
 
−
      intentional fragmentation of outgoing datagrams. Finally, the IP
 
−
      layer must (4) provide diagnostic and error functionality.  We
 
−
      expect that IP layer functions may increase somewhat in the
 
−
      future, as further Internet control and management facilities are
 
−
      developed.
 
  
 +
        Each route cache entry needs to include the following
 +
        fields:
  
 +
        (1)  Local IP address (for a multihomed host)
  
−
Internet Engineering Task Force                                [Page 27]
+
        (2)  Destination IP address
  
 +
        (3)  Type(s)-of-Service
  
 +
        (4)  Next-hop gateway IP address
  
 +
        Field (2) MAY be the full IP address of the destination
  
 
RFC1122                      INTERNET LAYER                October 1989
 
RFC1122                      INTERNET LAYER                October 1989
  
 +
        host, or only the destination network number.  Field (3),
 +
        the TOS, SHOULD be included.
  
−
      For normal datagrams, the processing is straightforward.  For
+
        See Section 3.3.4.2 for a discussion of the implications of
−
      incoming datagrams, the IP layer:
+
        multihoming for the lookup procedure in this cache.
  
−
      (1)  verifies that the datagram is correctly formatted;
+
        DISCUSSION:
 +
              Including the Type-of-Service field in the route cache
 +
              and considering it in the host route algorithm will
 +
              provide the necessary mechanism for the future when
 +
              Type-of-Service routing is commonly used in the
 +
              Internet.  See Section 3.2.1.6.
  
−
      (2) verifies that it is destined to the local host;
+
              Each route cache entry defines the endpoints of an
 +
              Internet path. Although the connecting path may change
 +
              dynamically in an arbitrary way, the transmission
 +
              characteristics of the path tend to remain
 +
              approximately constant over a time period longer than a
 +
              single typical host-host transport connection.
 +
              Therefore, a route cache entry is a natural place to
 +
              cache data on the properties of the path.  Examples of
 +
              such properties might be the maximum unfragmented
 +
              datagram size (see Section 3.3.3), or the average
 +
              round-trip delay measured by a transport protocol.
 +
              This data will generally be both gathered and used by a
 +
              higher layer protocol, e.g., by TCP, or by an
 +
              application using UDP.  Experiments are currently in
 +
              progress on caching path properties in this manner.
  
−
      (3) processes options;
+
              There is no consensus on whether the route cache should
 +
              be keyed on destination host addresses alone, or allow
 +
              both host and network addresses. Those who favor the
 +
              use of only host addresses argue that:
  
−
      (4)  reassembles the datagram if necessary; and
+
              (1)  As required in Section 3.3.1.2, Redirect messages
 +
                  will generally result in entries keyed on
 +
                  destination host addresses; the simplest and most
 +
                  general scheme would be to use host addresses
 +
                  always.
  
−
      (5)  passes the encapsulated message to the appropriate
+
              (2)  The IP layer may not always know the address mask
−
          transport-layer protocol module.
+
                  for a network address in a complex subnetted
 +
                  environment.
  
−
      For outgoing datagrams, the IP layer:
+
              (3)  The use of only host addresses allows the
 +
                  destination address to be used as a pure 32-bit
 +
                  number, which may allow the Internet architecture
 +
                  to be more easily extended in the future without
  
−
      (1)  sets any fields not set by the transport layer;
+
RFC1122                      INTERNET LAYER                October 1989
  
−
      (2)  selects the correct first hop on the connected network (a
+
                  any change to the hosts.
−
          process called "routing");
 
  
−
      (3)  fragments the datagram if necessary and if intentional
+
              The opposing view is that allowing a mixture of
−
          fragmentation is implemented (see Section 3.3.3); and
+
              destination hosts and networks in the route cache:
  
−
      (4)  passes the packet(s) to the appropriate link-layer driver.
+
              (1)  Saves memory space.
  
 +
              (2)  Leads to a simpler data structure, easily
 +
                  combining the cache with the tables of default and
 +
                  static routes (see below).
  
−
      A host is said to be multihomed if it has multiple IP addresses.
+
              (3)  Provides a more useful place to cache path
−
      Multihoming introduces considerable confusion and complexity into
+
                  properties, as discussed earlier.
−
      the protocol suite, and it is an area in which the Internet
 
−
      architecture falls seriously short of solving all problems. There
 
−
      are two distinct problem areas in multihoming:
 
  
−
      (1)  Local multihoming --  the host itself is multihomed; or
+
        IMPLEMENTATION:
−
 
+
              The cache needs to be large enough to include entries
−
      (2)  Remote multihoming -- the local host needs to communicate
+
              for the maximum number of destination hosts that may be
−
          with a remote multihomed host.
+
              in use at one time.
−
 
 
−
      At present, remote multihoming MUST be handled at the application
 
−
      layer, as discussed in the companion RFC [INTRO:1].  A host MAY
 
−
      support local multihoming, which is discussed in this document,
 
−
      and in particular in Section 3.3.4.
 
−
 
 
−
      Any host that forwards datagrams generated by another host is
 
−
      acting as a gateway and MUST also meet the specifications laid out
 
−
      in the gateway requirements RFC [INTRO:2]. An Internet host that
 
−
      includes embedded gateway code MUST have a configuration switch to
 
−
      disable the gateway function, and this switch MUST default to the
 
  
 +
              A route cache entry may also include control
 +
              information used to choose an entry for replacement.
 +
              This might take the form of a "recently used" bit, a
 +
              use count, or a last-used timestamp, for example.  It
 +
              is recommended that it include the time of last
 +
              modification of the entry, for diagnostic purposes.
  
 +
              An implementation may wish to reduce the overhead of
 +
              scanning the route cache for every datagram to be
 +
              transmitted.  This may be accomplished with a hash
 +
              table to speed the lookup, or by giving a connection-
 +
              oriented transport protocol a "hint" or temporary
 +
              handle on the appropriate cache entry, to be passed to
 +
              the IP layer with each subsequent datagram.
  
−
Internet Engineering Task Force                                [Page 28]
+
              Although we have described the route cache, the lists
 +
              of default gateways, and a table of static routes as
 +
              conceptually distinct, in practice they may be combined
 +
              into a single "routing table" data structure.
  
 +
      3.3.1.4  Dead Gateway Detection
  
 +
        The IP layer MUST be able to detect the failure of a "next-
 +
        hop" gateway that is listed in its route cache and to choose
 +
        an alternate gateway (see Section 3.3.1.5).
  
 +
        Dead gateway detection is covered in some detail in RFC-816
 +
        [IP:11]. Experience to date has not produced a complete
  
 
RFC1122                      INTERNET LAYER                October 1989
 
RFC1122                      INTERNET LAYER                October 1989
  
 +
        algorithm which is totally satisfactory, though it has
 +
        identified several forbidden paths and promising techniques.
  
−
      non-gateway mode.  In this mode, a datagram arriving through one
+
        *    A particular gateway SHOULD NOT be used indefinitely in
−
      interface will not be forwarded to another host or gateway (unless
+
              the absence of positive indications that it is
−
      it is source-routed), regardless of whether the host is single-
+
              functioning.
−
      homed or multihomed.  The host software MUST NOT automatically
 
−
      move into gateway mode if the host has more than one interface, as
 
−
      the operator of the machine may neither want to provide that
 
−
      service nor be competent to do so.
 
  
−
      In the following, the action specified in certain cases is to
+
        *    Active probes such as "pinging" (i.e., using an ICMP
−
      "silently discard" a received datagram.  This means that the
+
              Echo Request/Reply exchange) are expensive and scale
−
      datagram will be discarded without further processing and that the
+
              poorly.  In particular, hosts MUST NOT actively check
−
      host will not send any ICMP error message (see Section 3.2.2) as a
+
              the status of a first-hop gateway by simply pinging the
−
      result.  However, for diagnosis of problems a host SHOULD provide
+
              gateway continuously.
−
      the capability of logging the error (see Section 1.2.3), including
 
−
      the contents of the silently-discarded datagram, and SHOULD record
 
−
      the event in a statistics counter.
 
  
−
      DISCUSSION:
+
        *    Even when it is the only effective way to verify a
−
          Silent discard of erroneous datagrams is generally intended
+
              gateway's status, pinging MUST be used only when
−
          to prevent "broadcast storms".
+
              traffic is being sent to the gateway and when there is
 +
              no other positive indication to suggest that the
 +
              gateway is functioning.
  
−
   3.2  PROTOCOL WALK-THROUGH
+
        *   To avoid pinging, the layers above and/or below the
 +
              Internet layer SHOULD be able to give "advice" on the
 +
              status of route cache entries when either positive
 +
              (gateway OK) or negative (gateway dead) information is
 +
              available.
  
−
      3.2.1 Internet Protocol -- IP
+
        DISCUSSION:
 +
              If an implementation does not include an adequate
 +
              mechanism for detecting a dead gateway and re-routing,
 +
              a gateway failure may cause datagrams to apparently
 +
              vanish into a "black hole". This failure can be
 +
              extremely confusing for users and difficult for network
 +
              personnel to debug.
  
−
        3.2.1.1 Version Number: RFC-791 Section 3.1
+
              The dead-gateway detection mechanism must not cause
 +
              unacceptable load on the host, on connected networks,
 +
              or on first-hop gateway(s).  The exact constraints on
 +
              the timeliness of dead gateway detection and on
 +
              acceptable load may vary somewhat depending on the
 +
              nature of the host's mission, but a host generally
 +
              needs to detect a failed first-hop gateway quickly
 +
              enough that transport-layer connections will not break
 +
              before an alternate gateway can be selected.
  
−
            A datagram whose version number is not 4 MUST be silently
+
              Passing advice from other layers of the protocol stack
−
            discarded.
+
              complicates the interfaces between the layers, but it
 +
              is the preferred approach to dead gateway detection.
 +
              Advice can come from almost any part of the IP/TCP
  
−
        3.2.1.2  Checksum: RFC-791 Section 3.1
+
RFC1122                      INTERNET LAYER                October 1989
  
−
            A host MUST verify the IP header checksum on every received
+
              architecture, but it is expected to come primarily from
−
            datagram and silently discard every datagram that has a bad
+
              the transport and link layers.  Here are some possible
−
            checksum.
+
              sources for gateway advice:
  
−
        3.2.1.3  Addressing: RFC-791 Section 3.2
+
              o    TCP or any connection-oriented transport protocol
 +
                  should be able to give negative advice, e.g.,
 +
                  triggered by excessive retransmissions.
  
−
            There are now five classes of IP addresses: Class A through
+
              o    TCP may give positive advice when (new) data is
−
            Class E.  Class D addresses are used for IP multicasting
+
                  acknowledged.  Even though the route may be
−
            [IP:4], while Class E addresses are reserved for
+
                  asymmetric, an ACK for new data proves that the
−
            experimental use.
+
                  acknowleged data must have been transmitted
 +
                  successfully.
  
−
            A multicast (Class D) address is a 28-bit logical address
+
              o    An ICMP Redirect message from a particular gateway
−
            that stands for a group of hosts, and may be either
+
                  should be used as positive advice about that
−
            permanent or transient. Permanent multicast addresses are
+
                  gateway.
−
            allocated by the Internet Assigned Number Authority
 
−
            [INTRO:6], while transient addresses may be allocated
 
  
 +
              o    Link-layer information that reliably detects and
 +
                  reports host failures (e.g., ARPANET Destination
 +
                  Dead messages) should be used as negative advice.
  
 +
              o    Failure to ARP or to re-validate ARP mappings may
 +
                  be used as negative advice for the corresponding
 +
                  IP address.
  
−
Internet Engineering Task Force                                [Page 29]
+
              o    Packets arriving from a particular link-layer
 +
                  address are evidence that the system at this
 +
                  address is alive.  However, turning this
 +
                  information into advice about gateways requires
 +
                  mapping the link-layer address into an IP address,
 +
                  and then checking that IP address against the
 +
                  gateways pointed to by the route cache.  This is
 +
                  probably prohibitively inefficient.
  
 +
              Note that positive advice that is given for every
 +
              datagram received may cause unacceptable overhead in
 +
              the implementation.
  
 +
              While advice might be passed using required arguments
 +
              in all interfaces to the IP layer, some transport and
 +
              application layer protocols cannot deduce the correct
 +
              advice.  These interfaces must therefore allow a
 +
              neutral value for advice, since either always-positive
 +
              or always-negative advice leads to incorrect behavior.
  
 +
              There is another technique for dead gateway detection
 +
              that has been commonly used but is not recommended.
  
 
RFC1122                      INTERNET LAYER                October 1989
 
RFC1122                      INTERNET LAYER                October 1989
  
 +
              This technique depends upon the host passively
 +
              receiving ("wiretapping") the Interior Gateway Protocol
 +
              (IGP) datagrams that the gateways are broadcasting to
 +
              each other.  This approach has the drawback that a host
 +
              needs to recognize all the interior gateway protocols
 +
              that gateways may use (see [INTRO:2]).  In addition, it
 +
              only works on a broadcast network.
  
−
            dynamically to transient groups.  Group membership is
+
              At present, pinging (i.e., using ICMP Echo messages) is
−
            determined dynamically using IGMP [IP:4].
+
              the mechanism for gateway probing when absolutely
 +
              required.  A successful ping guarantees that the
 +
              addressed interface and its associated machine are up,
 +
              but it does not guarantee that the machine is a gateway
 +
              as opposed to a host.  The normal inference is that if
 +
              a Redirect or other evidence indicates that a machine
 +
              was a gateway, successful pings will indicate that the
 +
              machine is still up and hence still a gateway.
 +
              However, since a host silently discards packets that a
 +
              gateway would forward or redirect, this assumption
 +
              could sometimes fail. To avoid this problem, a new
 +
              ICMP message under development will ask "are you a
 +
              gateway?"
  
−
            We now summarize the important special cases for Class A, B,
+
        IMPLEMENTATION:
−
            and C IP addresses, using the following notation for an IP
+
              The following specific algorithm has been suggested:
−
            address:
 
  
−
                { <Network-number>, <Host-number> }
+
              o    Associate a "reroute timer" with each gateway
 +
                  pointed to by the route cache.  Initialize the
 +
                  timer to a value Tr, which must be small enough to
 +
                  allow detection of a dead gateway before transport
 +
                  connections time out.
  
−
            or
+
              o    Positive advice would reset the reroute timer to
−
                { <Network-number>, <Subnet-number>, <Host-number> }
+
                  Tr.  Negative advice would reduce or zero the
 +
                  reroute timer.
  
−
            and the notation "-1" for a field that contains all 1 bits.
+
              o    Whenever the IP layer used a particular gateway to
−
            This notation is not intended to imply that the 1-bits in an
+
                  route a datagram, it would check the corresponding
−
            address mask need be contiguous.
+
                  reroute timer. If the timer had expired (reached
 +
                  zero), the IP layer would send a ping to the
 +
                  gateway, followed immediately by the datagram.
  
−
            (a)  { 0, 0 }
+
              o    The ping (ICMP Echo) would be sent again if
 +
                  necessary, up to N times. If no ping reply was
 +
                  received in N tries, the gateway would be assumed
 +
                  to have failed, and a new first-hop gateway would
 +
                  be chosen for all cache entries pointing to the
 +
                  failed gateway.
  
−
                This host on this network.  MUST NOT be sent, except as
+
RFC1122                      INTERNET LAYER                October 1989
−
                a source address as part of an initialization procedure
 
−
                by which the host learns its own IP address.
 
  
−
                See also Section 3.3.6 for a non-standard use of {0,0}.
+
              Note that the size of Tr is inversely related to the
 +
              amount of advice available. Tr should be large enough
 +
              to insure that:
  
−
            (b) { 0, <Host-number> }
+
              *    Any pinging will be at a low level (e.g., <10%) of
 +
                  all packets sent to a gateway from the host, AND
  
−
                Specified host on this network. It MUST NOT be sent,
+
              *    pinging is infrequent (e.g., every 3 minutes)
−
                except as a source address as part of an initialization
 
−
                procedure by which the host learns its full IP address.
 
  
−
            (c) { -1, -1 }
+
              Since the recommended algorithm is concerned with the
 +
              gateways pointed to by route cache entries, rather than
 +
              the cache entries themselves, a two level data
 +
              structure (perhaps coordinated with ARP or similar
 +
              caches) may be desirable for implementing a route
 +
              cache.
  
−
                Limited broadcast.  It MUST NOT be used as a source
+
      3.3.1.5 New Gateway Selection
−
                address.
 
  
−
                A datagram with this destination address will be
+
        If the failed gateway is not the current default, the IP
−
                received by every host on the connected physical
+
        layer can immediately switch to a default gateway.  If it is
−
                network but will not be forwarded outside that network.
+
        the current default that failed, the IP layer MUST select a
 +
        different default gateway (assuming more than one default is
 +
        known) for the failed route and for establishing new routes.
  
−
            (d)  { <Network-number>, -1 }
+
        DISCUSSION:
 +
              When a gateway does fail, the other gateways on the
 +
              connected network will learn of the failure through
 +
              some inter-gateway routing protocol.  However, this
 +
              will not happen instantaneously, since gateway routing
 +
              protocols typically have a settling time of 30-60
 +
              seconds.  If the host switches to an alternative
 +
              gateway before the gateways have agreed on the failure,
 +
              the new target gateway will probably forward the
 +
              datagram to the failed gateway and send a Redirect back
 +
              to the host pointing to the failed gateway (!).  The
 +
              result is likely to be a rapid oscillation in the
 +
              contents of the host's route cache during the gateway
 +
              settling period. It has been proposed that the dead-
 +
              gateway logic should include some hysteresis mechanism
 +
              to prevent such oscillations.  However, experience has
 +
              not shown any harm from such oscillations, since
 +
              service cannot be restored to the host until the
 +
              gateways' routing information does settle down.
  
−
                Directed broadcast to the specified network.  It MUST
+
        IMPLEMENTATION:
−
                NOT be used as a source address.
+
              One implementation technique for choosing a new default
 +
              gateway is to simply round-robin among the default
 +
              gateways in the host's list.  Another is to rank the
  
−
            (e)  { <Network-number>, <Subnet-number>, -1 }
+
RFC1122                      INTERNET LAYER                October 1989
  
−
                Directed broadcast to the specified subnet.  It MUST
+
              gateways in priority order, and when the current
−
                NOT be used as a source address.
+
              default gateway is not the highest priority one, to
 +
              "ping" the higher-priority gateways slowly to detect
 +
              when they return to service.  This pinging can be at a
 +
              very low rate, e.g., 0.005 per second.
  
 +
      3.3.1.6  Initialization
  
 +
        The following information MUST be configurable:
  
−
Internet Engineering Task Force                                [Page 30]
+
        (1)  IP address(es).
  
 +
        (2)  Address mask(s).
  
 +
        (3)  A list of default gateways, with a preference level.
  
 +
        A manual method of entering this configuration data MUST be
 +
        provided.  In addition, a variety of methods can be used to
 +
        determine this information dynamically; see the section on
 +
        "Host Initialization" in [INTRO:1].
  
−
RFC1122                      INTERNET LAYER                October 1989
+
        DISCUSSION:
 +
              Some host implementations use "wiretapping" of gateway
 +
              protocols on a broadcast network to learn what gateways
 +
              exist.  A standard method for default gateway discovery
 +
              is under development.
  
 +
  3.3.2  Reassembly
  
−
            (f)  { <Network-number>, -1, -1 }
+
      The IP layer MUST implement reassembly of IP datagrams.
  
−
                Directed broadcast to all subnets of the specified
+
      We designate the largest datagram size that can be reassembled
−
                subnetted network.  It MUST NOT be used as a source
+
      by EMTU_R ("Effective MTU to receive"); this is sometimes
−
                address.
+
      called the "reassembly buffer size".  EMTU_R MUST be greater
 +
      than or equal to 576, SHOULD be either configurable or
 +
      indefinite, and SHOULD be greater than or equal to the MTU of
 +
      the connected network(s).
  
−
            (g)  { 127, <any> }
+
      DISCUSSION:
 +
          A fixed EMTU_R limit should not be built into the code
 +
          because some application layer protocols require EMTU_R
 +
          values larger than 576.
  
−
                Internal host loopback address.  Addresses of this form
+
      IMPLEMENTATION:
−
                MUST NOT appear outside a host.
+
          An implementation may use a contiguous reassembly buffer
 +
          for each datagram, or it may use a more complex data
 +
          structure that places no definite limit on the reassembled
 +
          datagram size; in the latter case, EMTU_R is said to be
  
−
            The <Network-number> is administratively assigned so that
+
RFC1122                      INTERNET LAYER                October 1989
−
            its value will be unique in the entire world.
 
  
−
            IP addresses are not permitted to have the value 0 or -1 for
+
          "indefinite".
−
            any of the <Host-number>, <Network-number>, or <Subnet-
 
−
            number> fields (except in the special cases listed above).
 
−
            This implies that each of these fields will be at least two
 
−
            bits long.
 
  
−
            For further discussion of broadcast addresses, see Section
+
          Logically, reassembly is performed by simply copying each
−
            3.3.6.
+
          fragment into the packet buffer at the proper offset.
 +
          Note that fragments may overlap if successive
 +
          retransmissions use different packetizing but the same
 +
          reassembly Id.
  
−
            A host MUST support the subnet extensions to IP [IP:3]. As
+
          The tricky part of reassembly is the bookkeeping to
−
            a result, there will be an address mask of the form:
+
          determine when all bytes of the datagram have been
−
            {-1, -1, 0} associated with each of the host's local IP
+
          reassembled.  We recommend Clark's algorithm [IP:10] that
−
            addresses; see Sections 3.2.2.9 and 3.3.1.1.
+
          requires no additional data space for the bookkeeping.
 +
          However, note that, contrary to [IP:10], the first
 +
          fragment header needs to be saved for inclusion in a
 +
          possible ICMP Time Exceeded (Reassembly Timeout) message.
  
−
            When a host sends any datagram, the IP source address MUST
+
      There MUST be a mechanism by which the transport layer can
−
            be one of its own IP addresses (but not a broadcast or
+
      learn MMS_R, the maximum message size that can be received and
−
            multicast address).
+
      reassembled in an IP datagram (see GET_MAXSIZES calls in
 +
      Section 3.4). If EMTU_R is not indefinite, then the value of
 +
      MMS_R is given by:
  
−
            A host MUST silently discard an incoming datagram that is
+
        MMS_R = EMTU_R - 20
−
            not destined for the host.  An incoming datagram is destined
 
−
            for the host if the datagram's destination address field is:
 
  
−
            (1)  (one of) the host's IP address(es); or
+
      since 20 is the minimum size of an IP header.
  
−
            (2) an IP broadcast address valid for the connected
+
      There MUST be a reassembly timeout.  The reassembly timeout
−
                network; or
+
      value SHOULD be a fixed value, not set from the remaining TTL.
 +
      It is recommended that the value lie between 60 seconds and 120
 +
      seconds. If this timeout expires, the partially-reassembled
 +
      datagram MUST be discarded and an ICMP Time Exceeded message
 +
      sent to the source host (if fragment zero has been received).
  
−
            (3)  the address for a multicast group of which the host is
+
      DISCUSSION:
−
                a member on the incoming physical interface.
+
          The IP specification says that the reassembly timeout
 +
          should be the remaining TTL from the IP header, but this
 +
          does not work well because gateways generally treat TTL as
 +
          a simple hop count rather than an elapsed time.  If the
 +
          reassembly timeout is too small, datagrams will be
 +
          discarded unnecessarily, and communication may fail.  The
 +
          timeout needs to be at least as large as the typical
 +
          maximum delay across the Internet.  A realistic minimum
 +
          reassembly timeout would be 60 seconds.
  
−
            For most purposes, a datagram addressed to a broadcast or
+
          It has been suggested that a cache might be kept of
−
            multicast destination is processed as if it had been
+
          round-trip times measured by transport protocols for
−
            addressed to one of the host's IP addresses; we use the term
+
          various destinations, and that these values might be used
−
            "specific-destination address" for the equivalent local IP
+
          to dynamically determine a reasonable reassembly timeout
  
 +
RFC1122                      INTERNET LAYER                October 1989
  
 +
          value.  Further investigation of this approach is
 +
          required.
  
−
Internet Engineering Task Force                                [Page 31]
+
          If the reassembly timeout is set too high, buffer
 +
          resources in the receiving host will be tied up too long,
 +
          and the MSL (Maximum Segment Lifetime) [TCP:1] will be
 +
          larger than necessary.  The MSL controls the maximum rate
 +
          at which fragmented datagrams can be sent using distinct
 +
          values of the 16-bit Ident field; a larger MSL lowers the
 +
          maximum rate.  The TCP specification [TCP:1] arbitrarily
 +
          assumes a value of 2 minutes for MSL.  This sets an upper
 +
          limit on a reasonable reassembly timeout value.
  
 +
  3.3.3  Fragmentation
  
 +
      Optionally, the IP layer MAY implement a mechanism to fragment
 +
      outgoing datagrams intentionally.
  
 +
      We designate by EMTU_S ("Effective MTU for sending") the
 +
      maximum IP datagram size that may be sent, for a particular
 +
      combination of IP source and destination addresses and perhaps
 +
      TOS.
  
−
RFC1122                      INTERNET LAYER                October 1989
+
      A host MUST implement a mechanism to allow the transport layer
 +
      to learn MMS_S, the maximum transport-layer message size that
 +
      may be sent for a given {source, destination, TOS} triplet (see
 +
      GET_MAXSIZES call in Section 3.4).  If no local fragmentation
 +
      is performed, the value of MMS_S will be:
  
 +
        MMS_S = EMTU_S - <IP header size>
  
−
            address of the host.  The specific-destination address is
+
      and EMTU_S must be less than or equal to the MTU of the network
−
            defined to be the destination address in the IP header
+
      interface corresponding to the source address of the datagram.
−
            unless the header contains a broadcast or multicast address,
+
      Note that <IP header size> in this equation will be 20, unless
−
            in which case the specific-destination is an IP address
+
      the IP reserves space to insert IP options for its own purposes
−
            assigned to the physical interface on which the datagram
+
      in addition to any options inserted by the transport layer.
−
            arrived.
 
  
−
            A host MUST silently discard an incoming datagram containing
+
      A host that does not implement local fragmentation MUST ensure
−
            an IP source address that is invalid by the rules of this
+
      that the transport layer (for TCP) or the application layer
−
            section.  This validation could be done in either the IP
+
      (for UDP) obtains MMS_S from the IP layer and does not send a
−
            layer or by each protocol in the transport layer.
+
      datagram exceeding MMS_S in size.
  
−
            DISCUSSION:
+
      It is generally desirable to avoid local fragmentation and to
−
                A mis-addressed datagram might be caused by a link-
+
      choose EMTU_S low enough to avoid fragmentation in any gateway
−
                layer broadcast of a unicast datagram or by a gateway
+
      along the path.  In the absence of actual knowledge of the
−
                or host that is confused or mis-configured.
+
      minimum MTU along the path, the IP layer SHOULD use
 +
      EMTU_S <= 576 whenever the destination address is not on a
 +
      connected network, and otherwise use the connected network's
  
−
                An architectural goal for Internet hosts was to allow
+
RFC1122                      INTERNET LAYER                October 1989
−
                IP addresses to be featureless 32-bit numbers, avoiding
 
−
                algorithms that required a knowledge of the IP address
 
−
                format.  Otherwise, any future change in the format or
 
−
                interpretation of IP addresses will require host
 
−
                software changes.  However, validation of broadcast and
 
−
                multicast addresses violates this goal; a few other
 
−
                violations are described elsewhere in this document.
 
  
−
                Implementers should be aware that applications
+
      MTU.
−
                depending upon the all-subnets directed broadcast
 
−
                address (f) may be unusable on some networks.  All-
 
−
                subnets broadcast is not widely implemented in vendor
 
−
                gateways at present, and even when it is implemented, a
 
−
                particular network administration may disable it in the
 
−
                gateway configuration.
 
  
−
        3.2.1.4  Fragmentation and Reassembly: RFC-791 Section 3.2
+
      The MTU of each physical interface MUST be configurable.
  
−
            The Internet model requires that every host support
+
      A host IP layer implementation MAY have a configuration flag
−
            reassembly.  See Sections 3.3.2 and 3.3.3 for the
+
      "All-Subnets-MTU", indicating that the MTU of the connected
−
            requirements on fragmentation and reassembly.
+
      network is to be used for destinations on different subnets
 +
      within the same network, but not for other networks.  Thus,
 +
      this flag causes the network class mask, rather than the subnet
 +
      address mask, to be used to choose an EMTU_S. For a multihomed
 +
      host, an "All-Subnets-MTU" flag is needed for each network
 +
      interface.
  
−
        3.2.1.5  Identification: RFC-791 Section 3.2
+
      DISCUSSION:
 +
          Picking the correct datagram size to use when sending data
 +
          is a complex topic [IP:9].
  
−
            When sending an identical copy of an earlier datagram, a
+
          (a)  In general, no host is required to accept an IP
−
            host MAY optionally retain the same Identification field in
+
                datagram larger than 576 bytes (including header and
−
            the copy.
+
                data), so a host must not send a larger datagram
 +
                without explicit knowledge or prior arrangement with
 +
                the destination host.  Thus, MMS_S is only an upper
 +
                bound on the datagram size that a transport protocol
 +
                may send; even when MMS_S exceeds 556, the transport
 +
                layer must limit its messages to 556 bytes in the
 +
                absence of other knowledge about the destination
 +
                host.
  
 +
          (b)  Some transport protocols (e.g., TCP) provide a way to
 +
                explicitly inform the sender about the largest
 +
                datagram the other end can receive and reassemble
 +
                [IP:7].  There is no corresponding mechanism in the
 +
                IP layer.
  
 +
                A transport protocol that assumes an EMTU_R larger
 +
                than 576 (see Section 3.3.2), can send a datagram of
 +
                this larger size to another host that implements the
 +
                same protocol.
  
 +
          (c)  Hosts should ideally limit their EMTU_S for a given
 +
                destination to the minimum MTU of all the networks
 +
                along the path, to avoid any fragmentation.  IP
 +
                fragmentation, while formally correct, can create a
 +
                serious transport protocol performance problem,
 +
                because loss of a single fragment means all the
 +
                fragments in the segment must be retransmitted
 +
                [IP:9].
  
 +
RFC1122                      INTERNET LAYER                October 1989
  
 +
          Since nearly all networks in the Internet currently
 +
          support an MTU of 576 or greater, we strongly recommend
 +
          the use of 576 for datagrams sent to non-local networks.
  
−
Internet Engineering Task Force                                [Page 32]
+
          It has been suggested that a host could determine the MTU
 +
          over a given path by sending a zero-offset datagram
 +
          fragment and waiting for the receiver to time out the
 +
          reassembly (which cannot complete!) and return an ICMP
 +
          Time Exceeded message.  This message would include the
 +
          largest remaining fragment header in its body.  More
 +
          direct mechanisms are being experimented with, but have
 +
          not yet been adopted (see e.g., RFC-1063).
  
 +
  3.3.4  Local Multihoming
  
 +
      3.3.4.1  Introduction
  
 +
        A multihomed host has multiple IP addresses, which we may
 +
        think of as "logical interfaces".  These logical interfaces
 +
        may be associated with one or more physical interfaces, and
 +
        these physical interfaces may be connected to the same or
 +
        different networks.
  
−
RFC1122                      INTERNET LAYER                October 1989
+
        Here are some important cases of multihoming:
  
 +
        (a)  Multiple Logical Networks
  
−
            DISCUSSION:
+
              The Internet architects envisioned that each physical
−
                Some Internet protocol experts have maintained that
+
              network would have a single unique IP network (or
−
                when a host sends an identical copy of an earlier
+
              subnet) number.  However, LAN administrators have
−
                datagram, the new copy should contain the same
+
              sometimes found it useful to violate this assumption,
−
                Identification value as the original.  There are two
+
              operating a LAN with multiple logical networks per
−
                suggested advantages:  (1) if the datagrams are
+
              physical connected network.
−
                fragmented and some of the fragments are lost, the
 
−
                receiver may be able to reconstruct a complete datagram
 
−
                from fragments of the original and the copies; (2) a
 
−
                congested gateway might use the IP Identification field
 
−
                (and Fragment Offset) to discard duplicate datagrams
 
−
                from the queue.
 
  
−
                However, the observed patterns of datagram loss in the
+
              If a host connected to such a physical network is
−
                Internet do not favor the probability of retransmitted
+
              configured to handle traffic for each of N different
−
                fragments filling reassembly gaps, while other
+
              logical networks, then the host will have N logical
−
                mechanisms (e.g., TCP repacketizing upon
+
              interfaces.  These could share a single physical
−
                retransmission) tend to prevent retransmission of an
+
              interface, or might use N physical interfaces to the
−
                identical datagram [IP:9].  Therefore, we believe that
+
              same network.
−
                retransmitting the same Identification field is not
 
−
                useful.  Also, a connectionless transport protocol like
 
−
                UDP would require the cooperation of the application
 
−
                programs to retain the same Identification value in
 
−
                identical datagrams.
 
  
−
         3.2.1.6 Type-of-Service: RFC-791 Section 3.2
+
         (b) Multiple Logical Hosts
  
−
            The "Type-of-Service" byte in the IP header is divided into
+
              When a host has multiple IP addresses that all have the
−
            two sections:  the Precedence field (high-order 3 bits), and
+
              same <Network-number> part (and the same <Subnet-
−
            a field that is customarily called "Type-of-Service" or
+
              number> part, if any), the logical interfaces are known
−
            "TOS" (low-order 5 bits).  In this document, all references
+
              as "logical hosts".  These logical interfaces might
−
            to "TOS" or the "TOS field" refer to the low-order 5 bits
+
              share a single physical interface or might use separate
−
            only.
 
  
−
            The Precedence field is intended for Department of Defense
+
RFC1122                      INTERNET LAYER                October 1989
−
            applications of the Internet protocols.  The use of non-zero
 
−
            values in this field is outside the scope of this document
 
−
            and the IP standard specification.  Vendors should consult
 
−
            the Defense Communication Agency (DCA) for guidance on the
 
−
            IP Precedence field and its implications for other protocol
 
−
            layers.  However, vendors should note that the use of
 
−
            precedence will most likely require that its value be passed
 
−
            between protocol layers in just the same way as the TOS
 
−
            field is passed.
 
  
−
            The IP layer MUST provide a means for the transport layer to
+
              physical interfaces to the same physical network.
−
            set the TOS field of every datagram that is sent; the
 
−
            default is all zero bits. The IP layer SHOULD pass received
 
  
 +
        (c)  Simple Multihoming
  
 +
              In this case, each logical interface is mapped into a
 +
              separate physical interface and each physical interface
 +
              is connected to a different physical network.  The term
 +
              "multihoming" was originally applied only to this case,
 +
              but it is now applied more generally.
  
−
Internet Engineering Task Force                                [Page 33]
+
              A host with embedded gateway functionality will
 +
              typically fall into the simple multihoming case.  Note,
 +
              however, that a host may be simply multihomed without
 +
              containing an embedded gateway, i.e., without
 +
              forwarding datagrams from one connected network to
 +
              another.
  
 +
              This case presents the most difficult routing problems.
 +
              The choice of interface (i.e., the choice of first-hop
 +
              network) may significantly affect performance or even
 +
              reachability of remote parts of the Internet.
  
 +
        Finally, we note another possibility that is NOT
 +
        multihoming:  one logical interface may be bound to multiple
 +
        physical interfaces, in order to increase the reliability or
 +
        throughput between directly connected machines by providing
 +
        alternative physical paths between them.  For instance, two
 +
        systems might be connected by multiple point-to-point links.
 +
        We call this "link-layer multiplexing".  With link-layer
 +
        multiplexing, the protocols above the link layer are unaware
 +
        that multiple physical interfaces are present; the link-
 +
        layer device driver is responsible for multiplexing and
 +
        routing packets across the physical interfaces.
  
 +
        In the Internet protocol architecture, a transport protocol
 +
        instance ("entity") has no address of its own, but instead
 +
        uses a single Internet Protocol (IP) address.  This has
 +
        implications for the IP, transport, and application layers,
 +
        and for the interfaces between them.  In particular, the
 +
        application software may have to be aware of the multiple IP
 +
        addresses of a multihomed host; in other cases, the choice
 +
        can be made within the network software.
  
−
RFC1122                      INTERNET LAYER                October 1989
+
      3.3.4.2  Multihoming Requirements
  
 +
        The following general rules apply to the selection of an IP
 +
        source address for sending a datagram from a multihomed
  
−
            TOS values up to the transport layer.
+
RFC1122                      INTERNET LAYER                October 1989
  
−
            The particular link-layer mappings of TOS contained in RFC-
+
        host.
−
            795 SHOULD NOT be implemented.
 
  
−
            DISCUSSION:
+
        (1)  If the datagram is sent in response to a received
−
                While the TOS field has been little used in the past,
+
              datagram, the source address for the response SHOULD be
−
                it is expected to play an increasing role in the near
+
              the specific-destination address of the request.  See
−
                future.  The TOS field is expected to be used to
+
              Sections 4.1.3.5 and 4.2.3.7 and the "General Issues"
−
                control two aspects of gateway operations: routing and
+
              section of [INTRO:1] for more specific requirements on
−
                queueing algorithms. See Section 2 of [INTRO:1] for
+
              higher layers.
−
                the requirements on application programs to specify TOS
 
−
                values.
 
  
−
                The TOS field may also be mapped into link-layer
+
              Otherwise, a source address must be selected.
−
                service selectors.  This has been applied to provide
 
−
                effective sharing of serial lines by different classes
 
−
                of TCP traffic, for example.  However, the mappings
 
−
                suggested in RFC-795 for networks that were included in
 
−
                the Internet as of 1981 are now obsolete.
 
  
−
         3.2.1.7 Time-to-Live: RFC-791 Section 3.2
+
         (2) An application MUST be able to explicitly specify the
 +
              source address for initiating a connection or a
 +
              request.
  
−
            A host MUST NOT send a datagram with a Time-to-Live (TTL)
+
        (3)  In the absence of such a specification, the networking
−
            value of zero.
+
              software MUST choose a source address.  Rules for this
 +
              choice are described below.
  
−
            A host MUST NOT discard a datagram just because it was
+
        There are two key requirement issues related to multihoming:
−
            received with TTL less than 2.
 
  
−
            The IP layer MUST provide a means for the transport layer to
+
        (A)  A host MAY silently discard an incoming datagram whose
−
            set the TTL field of every datagram that is sent.  When a
+
              destination address does not correspond to the physical
−
            fixed TTL value is used, it MUST be configurable.  The
+
              interface through which it is received.
−
            current suggested value will be published in the "Assigned
 
−
            Numbers" RFC.
 
  
−
            DISCUSSION:
+
        (B)  A host MAY restrict itself to sending (non-source-
−
                The TTL field has two functions: limit the lifetime of
+
              routed) IP datagrams only through the physical
−
                TCP segments (see RFC-793 [TCP:1], p. 28), and
+
              interface that corresponds to the IP source address of
−
                terminate Internet routing loops. Although TTL is a
+
              the datagrams.
−
                time in seconds, it also has some attributes of a hop-
 
−
                count, since each gateway is required to reduce the TTL
 
−
                field by at least one.
 
  
−
                The intent is that TTL expiration will cause a datagram
+
        DISCUSSION:
−
                to be discarded by a gateway but not by the destination
+
              Internet host implementors have used two different
−
                host; however, hosts that act as gateways by forwarding
+
              conceptual models for multihoming, briefly summarized
−
                datagrams must follow the gateway rules for TTL.
+
              in the following discussion.  This document takes no
 +
              stand on which model is preferred; each seems to have a
 +
              place.  This ambivalence is reflected in the issues (A)
 +
              and (B) being optional.
  
 +
              o    Strong ES Model
  
 +
                  The Strong ES (End System, i.e., host) model
 +
                  emphasizes the host/gateway (ES/IS) distinction,
 +
                  and would therefore substitute MUST for MAY in
 +
                  issues (A) and (B) above.  It tends to model a
 +
                  multihomed host as a set of logical hosts within
 +
                  the same physical host.
  
 +
RFC1122                      INTERNET LAYER                October 1989
  
−
Internet Engineering Task Force                                [Page 34]
+
                  With respect to (A), proponents of the Strong ES
 +
                  model note that automatic Internet routing
 +
                  mechanisms could not route a datagram to a
 +
                  physical interface that did not correspond to the
 +
                  destination address.
  
 +
                  Under the Strong ES model, the route computation
 +
                  for an outgoing datagram is the mapping:
  
 +
                      route(src IP addr, dest IP addr, TOS)
 +
                                                    -> gateway
  
 +
                  Here the source address is included as a parameter
 +
                  in order to select a gateway that is directly
 +
                  reachable on the corresponding physical interface.
 +
                  Note that this model logically requires that in
 +
                  general there be at least one default gateway, and
 +
                  preferably multiple defaults, for each IP source
 +
                  address.
  
−
RFC1122                      INTERNET LAYER                October 1989
+
              o    Weak ES Model
  
 +
                  This view de-emphasizes the ES/IS distinction, and
 +
                  would therefore substitute MUST NOT for MAY in
 +
                  issues (A) and (B).  This model may be the more
 +
                  natural one for hosts that wiretap gateway routing
 +
                  protocols, and is necessary for hosts that have
 +
                  embedded gateway functionality.
  
−
                A higher-layer protocol may want to set the TTL in
+
                  The Weak ES Model may cause the Redirect mechanism
−
                order to implement an "expanding scope" search for some
+
                  to fail.  If a datagram is sent out a physical
−
                Internet resource.  This is used by some diagnostic
+
                  interface that does not correspond to the
−
                tools, and is expected to be useful for locating the
+
                  destination address, the first-hop gateway will
−
                "nearest" server of a given class using IP
+
                  not realize when it needs to send a Redirect.  On
−
                multicasting, for example.  A particular transport
+
                  the other hand, if the host has embedded gateway
−
                protocol may also want to specify its own TTL bound on
+
                  functionality, then it has routing information
−
                maximum datagram lifetime.
+
                  without listening to Redirects.
  
−
                A fixed value must be at least big enough for the
+
                  In the Weak ES model, the route computation for an
−
                Internet "diameter," i.e., the longest possible path.
+
                  outgoing datagram is the mapping:
−
                A reasonable value is about twice the diameter, to
 
−
                allow for continued Internet growth.
 
  
−
        3.2.1.8  Options: RFC-791 Section 3.2
+
                      route(dest IP addr, TOS) -> gateway, interface
  
−
            There MUST be a means for the transport layer to specify IP
+
RFC1122                      INTERNET LAYER                October 1989
−
            options to be included in transmitted IP datagrams (see
 
−
            Section 3.4).
 
  
−
            All IP options (except NOP or END-OF-LIST) received in
+
      3.3.4.3 Choosing a Source Address
−
            datagrams MUST be passed to the transport layer (or to ICMP
 
−
            processing when the datagram is an ICMP message).  The IP
 
−
            and transport layer MUST each interpret those IP options
 
−
            that they understand and silently ignore the others.
 
  
−
            Later sections of this document discuss specific IP option
+
        DISCUSSION:
−
            support required by each of ICMP, TCP, and UDP.
+
              When it sends an initial connection request (e.g., a
 +
              TCP "SYN" segment) or a datagram service request (e.g.,
 +
              a UDP-based query), the transport layer on a multihomed
 +
              host needs to know which source address to use. If the
 +
              application does not specify it, the transport layer
 +
              must ask the IP layer to perform the conceptual
 +
              mapping:
  
−
            DISCUSSION:
+
                  GET_SRCADDR(remote IP addr, TOS)
−
                Passing all received IP options to the transport layer
+
                                            -> local IP address
−
                is a deliberate "violation of strict layering" that is
 
−
                designed to ease the introduction of new transport-
 
−
                relevant IP options in the future.  Each layer must
 
−
                pick out any options that are relevant to its own
 
−
                processing and ignore the rest.  For this purpose,
 
−
                every IP option except NOP and END-OF-LIST will include
 
−
                a specification of its own length.
 
  
−
                This document does not define the order in which a
+
              Here TOS is the Type-of-Service value (see Section
−
                receiver must process multiple options in the same IP
+
              3.2.1.6), and the result is the desired source address.
−
                header. Hosts sending multiple options must be aware
+
              The following rules are suggested for implementing this
−
                that this introduces an ambiguity in the meaning of
+
              mapping:
−
                certain options when combined with a source-route
 
−
                option.
 
  
−
            IMPLEMENTATION:
+
              (a)  If the remote Internet address lies on one of the
−
                The IP layer must not crash as the result of an option
+
                  (sub-) nets to which the host is directly
 +
                  connected, a corresponding source address may be
 +
                  chosen, unless the corresponding interface is
 +
                  known to be down.
  
 +
              (b)  The route cache may be consulted, to see if there
 +
                  is an active route to the specified destination
 +
                  network through any network interface; if so, a
 +
                  local IP address corresponding to that interface
 +
                  may be chosen.
  
 +
              (c)  The table of static routes, if any (see Section
 +
                  3.3.1.2) may be similarly consulted.
  
−
Internet Engineering Task Force                                [Page 35]
+
              (d)  The default gateways may be consulted.  If these
−
 
+
                  gateways are assigned to different interfaces, the
 +
                  interface corresponding to the gateway with the
 +
                  highest preference may be chosen.
  
 +
              In the future, there may be a defined way for a
 +
              multihomed host to ask the gateways on all connected
 +
              networks for advice about the best network to use for a
 +
              given destination.
  
 +
        IMPLEMENTATION:
 +
              It will be noted that this process is essentially the
 +
              same as datagram routing (see Section 3.3.1), and
 +
              therefore hosts may be able to combine the
  
 
RFC1122                      INTERNET LAYER                October 1989
 
RFC1122                      INTERNET LAYER                October 1989
  
 +
              implementation of the two functions.
  
−
                length that is outside the possible range.  For
+
  3.3.5 Source Route Forwarding
−
                example, erroneous option lengths have been observed to
 
−
                put some IP implementations into infinite loops.
 
  
−
            Here are the requirements for specific IP options:
+
      Subject to restrictions given below, a host MAY be able to act
 +
      as an intermediate hop in a source route, forwarding a source-
 +
      routed datagram to the next specified hop.
  
 +
      However, in performing this gateway-like function, the host
 +
      MUST obey all the relevant rules for a gateway forwarding
 +
      source-routed datagrams [INTRO:2].  This includes the following
 +
      specific provisions, which override the corresponding host
 +
      provisions given earlier in this document:
  
−
            (a)  Security Option
+
      (A)  TTL (ref. Section 3.2.1.7)
  
−
                Some environments require the Security option in every
+
          The TTL field MUST be decremented and the datagram perhaps
−
                datagram; such a requirement is outside the scope of
+
          discarded as specified for a gateway in [INTRO:2].
−
                this document and the IP standard specification.  Note,
 
−
                however, that the security options described in RFC-791
 
−
                and RFC-1038 are obsolete.  For DoD applications,
 
−
                vendors should consult [IP:8] for guidance.
 
  
 +
      (B)  ICMP Destination Unreachable (ref. Section 3.2.2.1)
  
−
            (b)  Stream Identifier Option
+
          A host MUST be able to generate Destination Unreachable
 +
          messages with the following codes:
  
−
                This option is obsolete; it SHOULD NOT be sent, and it
+
          4    (Fragmentation Required but DF Set) when a source-
−
                MUST be silently ignored if received.
+
                routed datagram cannot be fragmented to fit into the
 +
                target network;
  
 +
          5    (Source Route Failed) when a source-routed datagram
 +
                cannot be forwarded, e.g., because of a routing
 +
                problem or because the next hop of a strict source
 +
                route is not on a connected network.
  
−
            (c)  Source Route Options
+
      (C)  IP Source Address (ref. Section 3.2.1.3)
  
−
                A host MUST support originating a source route and MUST
+
          A source-routed datagram being forwarded MAY (and normally
−
                be able to act as the final destination of a source
+
          will) have a source address that is not one of the IP
−
                route.
+
          addresses of the forwarding host.
  
−
                If host receives a datagram containing a completed
+
      (D) Record Route Option (ref. Section 3.2.1.8d)
−
                source route (i.e., the pointer points beyond the last
 
−
                field), the datagram has reached its final destination;
 
−
                the option as received (the recorded route) MUST be
 
−
                passed up to the transport layer (or to ICMP message
 
−
                processing). This recorded route will be reversed and
 
−
                used to form a return source route for reply datagrams
 
−
                (see discussion of IP Options in Section 4). When a
 
−
                return source route is built, it MUST be correctly
 
−
                formed even if the recorded route included the source
 
−
                host (see case (B) in the discussion below).
 
  
−
                An IP header containing more than one Source Route
+
          A host that is forwarding a source-routed datagram
−
                option MUST NOT be sent; the effect on routing of
+
          containing a Record Route option MUST update that option,
−
                multiple Source Route options is implementation-
+
          if it has room.
−
                specific.
 
  
−
                Section 3.3.5 presents the rules for a host acting as
+
      (E)  Timestamp Option (ref. Section 3.2.1.8e)
−
                an intermediate hop in a source route, i.e., forwarding
 
  
 +
          A host that is forwarding a source-routed datagram
  
 +
RFC1122                      INTERNET LAYER                October 1989
  
−
Internet Engineering Task Force                                [Page 36]
+
          containing a Timestamp Option MUST add the current
 +
          timestamp to that option, according to the rules for this
 +
          option.
  
 +
      To define the rules restricting host forwarding of source-
 +
      routed datagrams, we use the term "local source-routing" if the
 +
      next hop will be through the same physical interface through
 +
      which the datagram arrived; otherwise, it is "non-local
 +
      source-routing".
  
 +
      o    A host is permitted to perform local source-routing
 +
          without restriction.
  
 +
      o    A host that supports non-local source-routing MUST have a
 +
          configurable switch to disable forwarding, and this switch
 +
          MUST default to disabled.
  
−
RFC1122                      INTERNET LAYER                October 1989
+
      o    The host MUST satisfy all gateway requirements for
 +
          configurable policy filters [INTRO:2] restricting non-
 +
          local forwarding.
  
 +
      If a host receives a datagram with an incomplete source route
 +
      but does not forward it for some reason, the host SHOULD return
 +
      an ICMP Destination Unreachable (code 5, Source Route Failed)
 +
      message, unless the datagram was itself an ICMP error message.
  
−
                a source-routed datagram.
+
  3.3.6  Broadcasts
  
−
                DISCUSSION:
+
      Section 3.2.1.3 defined the four standard IP broadcast address
−
                      If a source-routed datagram is fragmented, each
+
      forms:
−
                      fragment will contain a copy of the source route.
 
−
                      Since the processing of IP options (including a
 
−
                      source route) must precede reassembly, the
 
−
                      original datagram will not be reassembled until
 
−
                      the final destination is reached.
 
  
−
                      Suppose a source routed datagram is to be routed
+
        Limited Broadcast:  {-1, -1}
−
                      from host S to host D via gateways G1, G2, ... Gn.
 
−
                      There was an ambiguity in the specification over
 
−
                      whether the source route option in a datagram sent
 
−
                      out by S should be (A) or (B):
 
  
−
                          (A):  {>>G2, G3, ... Gn, D}     <--- CORRECT
+
        Directed Broadcast:  {<Network-number>,-1}
  
−
                          (B): {S, >>G2, G3, ... Gn, D}  <---- WRONG
+
        Subnet Directed Broadcast:
 +
                          {<Network-number>,<Subnet-number>,-1}
  
−
                      (where >> represents the pointer).  If (A) is
+
        All-Subnets Directed Broadcast: {<Network-number>,-1,-1}
−
                      sent, the datagram received at D will contain the
 
−
                      option: {G1, G2, ... Gn >>}, with S and D as the
 
−
                      IP source and destination addresses.  If (B) were
 
−
                      sent, the datagram received at D would again
 
−
                      contain S and D as the same IP source and
 
−
                      destination addresses, but the option would be:
 
−
                      {S, G1, ...Gn >>}; i.e., the originating host
 
−
                      would be the first hop in the route.
 
  
 +
      A host MUST recognize any of these forms in the destination
 +
      address of an incoming datagram.
  
−
            (d) Record Route Option
+
      There is a class of hosts* that use non-standard broadcast
 +
      address forms, substituting 0 for -1. All hosts SHOULD
 +
_________________________
 +
*4.2BSD Unix and its derivatives, but not 4.3BSD.
  
−
                Implementation of originating and processing the Record
+
RFC1122                      INTERNET LAYER                October 1989
−
                Route option is OPTIONAL.
 
  
 +
      recognize and accept any of these non-standard broadcast
 +
      addresses as the destination address of an incoming datagram.
 +
      A host MAY optionally have a configuration option to choose the
 +
      0 or the -1 form of broadcast address, for each physical
 +
      interface, but this option SHOULD default to the standard (-1)
 +
      form.
  
−
            (e)  Timestamp Option
+
      When a host sends a datagram to a link-layer broadcast address,
 +
      the IP destination address MUST be a legal IP broadcast or IP
 +
      multicast address.
  
−
                Implementation of originating and processing the
+
      A host SHOULD silently discard a datagram that is received via
−
                Timestamp option is OPTIONAL. If it is implemented,
+
      a link-layer broadcast (see Section 2.4) but does not specify
−
                the following rules apply:
+
      an IP multicast or broadcast destination address.
  
−
                o    The originating host MUST record a timestamp in a
+
      Hosts SHOULD use the Limited Broadcast address to broadcast to
−
                      Timestamp option whose Internet address fields are
+
      a connected network.
−
                      not pre-specified or whose first pre-specified
 
−
                      address is the host's interface address.
 
  
 +
      DISCUSSION:
 +
          Using the Limited Broadcast address instead of a Directed
 +
          Broadcast address may improve system robustness.  Problems
 +
          are often caused by machines that do not understand the
 +
          plethora of broadcast addresses (see Section 3.2.1.3), or
 +
          that may have different ideas about which broadcast
 +
          addresses are in use.  The prime example of the latter is
 +
          machines that do not understand subnetting but are
 +
          attached to a subnetted net.  Sending a Subnet Broadcast
 +
          for the connected network will confuse those machines,
 +
          which will see it as a message to some other host.
  
 +
          There has been discussion on whether a datagram addressed
 +
          to the Limited Broadcast address ought to be sent from all
 +
          the interfaces of a multihomed host.  This specification
 +
          takes no stand on the issue.
  
 +
  3.3.7  IP Multicasting
  
−
Internet Engineering Task Force                                [Page 37]
+
      A host SHOULD support local IP multicasting on all connected
 +
      networks for which a mapping from Class D IP addresses to
 +
      link-layer addresses has been specified (see below).  Support
 +
      for local IP multicasting includes sending multicast datagrams,
 +
      joining multicast groups and receiving multicast datagrams, and
 +
      leaving multicast groups.  This implies support for all of
 +
      [IP:4] except the IGMP protocol itself, which is OPTIONAL.
  
 +
RFC1122                      INTERNET LAYER                October 1989
  
 +
      DISCUSSION:
 +
          IGMP provides gateways that are capable of multicast
 +
          routing with the information required to support IP
 +
          multicasting across multiple networks.  At this time,
 +
          multicast-routing gateways are in the experimental stage
 +
          and are not widely available.  For hosts that are not
 +
          connected to networks with multicast-routing gateways or
 +
          that do not need to receive multicast datagrams
 +
          originating on other networks, IGMP serves no purpose and
 +
          is therefore optional for now.  However, the rest of
 +
          [IP:4] is currently recommended for the purpose of
 +
          providing IP-layer access to local network multicast
 +
          addressing, as a preferable alternative to local broadcast
 +
          addressing.  It is expected that IGMP will become
 +
          recommended at some future date, when multicast-routing
 +
          gateways have become more widely available.
  
 +
      If IGMP is not implemented, a host SHOULD still join the "all-
 +
      hosts" group (224.0.0.1) when the IP layer is initialized and
 +
      remain a member for as long as the IP layer is active.
  
−
RFC1122                      INTERNET LAYER                October 1989
+
      DISCUSSION:
 +
          Joining the "all-hosts" group will support strictly local
 +
          uses of multicasting, e.g., a gateway discovery protocol,
 +
          even if IGMP is not implemented.
  
 +
      The mapping of IP Class D addresses to local addresses is
 +
      currently specified for the following types of networks:
  
−
                o    The destination host MUST (if possible) add the
+
      o    Ethernet/IEEE 802.3, as defined in [IP:4].
−
                      current timestamp to a Timestamp option before
 
−
                      passing the option to the transport layer or to
 
−
                      ICMP for processing.
 
  
−
                o    A timestamp value MUST follow the rules given in
+
      o    Any network that supports broadcast but not multicast,
−
                      Section 3.2.2.8 for the ICMP Timestamp message.
+
          addressing: all IP Class D addresses map to the local
 +
          broadcast address.
  
 +
      o    Any type of point-to-point link (e.g., SLIP or HDLC
 +
          links): no mapping required.  All IP multicast datagrams
 +
          are sent as-is, inside the local framing.
  
−
       3.2.2 Internet Control Message Protocol -- ICMP
+
       Mappings for other types of networks will be specified in the
 +
      future.
  
−
        ICMP messages are grouped into two classes.
+
      A host SHOULD provide a way for higher-layer protocols or
 +
      applications to determine which of the host's connected
 +
      network(s) support IP multicast addressing.
  
−
        *
+
RFC1122                      INTERNET LAYER                October 1989
−
              ICMP error messages:
 
  
−
              Destination Unreachable  (see Section 3.2.2.1)
+
  3.3.8  Error Reporting
−
              Redirect                  (see Section 3.2.2.2)
 
−
              Source Quench            (see Section 3.2.2.3)
 
−
              Time Exceeded            (see Section 3.2.2.4)
 
−
              Parameter Problem        (see Section 3.2.2.5)
 
  
 +
      Wherever practical, hosts MUST return ICMP error datagrams on
 +
      detection of an error, except in those cases where returning an
 +
      ICMP error message is specifically prohibited.
  
−
        *
+
      DISCUSSION:
−
              ICMP query messages:
+
          A common phenomenon in datagram networks is the "black
−
 
+
          hole disease": datagrams are sent out, but nothing comes
−
                Echo                    (see Section 3.2.2.6)
+
          back. Without any error datagrams, it is difficult for
−
                Information              (see Section 3.2.2.7)
+
          the user to figure out what the problem is.
−
                Timestamp                (see Section 3.2.2.8)
 
−
                Address Mask            (see Section 3.2.2.9)
 
−
 
 
−
 
 
−
        If an ICMP message of unknown type is received, it MUST be
 
−
        silently discarded.
 
  
−
        Every ICMP error message includes the Internet header and at
+
3.4  INTERNET/TRANSPORT LAYER INTERFACE
−
        least the first 8 data octets of the datagram that triggered
 
−
        the error; more than 8 octets MAY be sent; this header and data
 
−
        MUST be unchanged from the received datagram.
 
  
−
        In those cases where the Internet layer is required to pass an
+
  The interface between the IP layer and the transport layer MUST
−
        ICMP error message to the transport layer, the IP protocol
+
  provide full access to all the mechanisms of the IP layer,
−
        number MUST be extracted from the original header and used to
+
  including options, Type-of-Service, and Time-to-Live.  The
−
        select the appropriate transport protocol entity to handle the
+
  transport layer MUST either have mechanisms to set these interface
−
        error.
+
  parameters, or provide a path to pass them through from an
 +
  application, or both.
  
−
        An ICMP error message SHOULD be sent with normal (i.e., zero)
+
  DISCUSSION:
−
        TOS bits.
+
        Applications are urged to make use of these mechanisms where
 +
        applicable, even when the mechanisms are not currently
 +
        effective in the Internet (e.g., TOS).  This will allow these
 +
        mechanisms to be immediately useful when they do become
 +
        effective, without a large amount of retrofitting of host
 +
        software.
  
 +
  We now describe a conceptual interface between the transport layer
 +
  and the IP layer, as a set of procedure calls.  This is an
 +
  extension of the information in Section 3.3 of RFC-791 [IP:1].
  
 +
  *    Send Datagram
  
−
Internet Engineering Task Force                                [Page 38]
+
            SEND(src, dst, prot, TOS, TTL, BufPTR, len, Id, DF, opt
 +
                  => result )
  
 +
        where the parameters are defined in RFC-791.  Passing an Id
 +
        parameter is optional; see Section 3.2.1.5.
  
 +
  *    Receive Datagram
  
 +
            RECV(BufPTR, prot
 +
                  => result, src, dst, SpecDest, TOS, len, opt)
  
 
RFC1122                      INTERNET LAYER                October 1989
 
RFC1122                      INTERNET LAYER                October 1989
  
 +
        All the parameters are defined in RFC-791, except for:
  
−
        An ICMP error message MUST NOT be sent as the result of
+
            SpecDest = specific-destination address of datagram
−
        receiving:
+
                        (defined in Section 3.2.1.3)
−
 
 
−
        *    an ICMP error message, or
 
−
 
 
−
        *    a datagram destined to an IP broadcast or IP multicast
 
−
              address, or
 
−
 
 
−
        *    a datagram sent as a link-layer broadcast, or
 
−
 
 
−
        *    a non-initial fragment, or
 
  
−
        *    a datagram whose source address does not define a single
+
        The result parameter dst contains the datagram's destination
−
              host -- e.g., a zero address, a loopback address, a
+
        address. Since this may be a broadcast or multicast address,
−
              broadcast address, a multicast address, or a Class E
+
        the SpecDest parameter (not shown in RFC-791) MUST be passed.
−
              address.
+
        The parameter opt contains all the IP options received in the
 +
        datagram; these MUST also be passed to the transport layer.
  
−
        NOTE: THESE RESTRICTIONS TAKE PRECEDENCE OVER ANY REQUIREMENT
+
  *    Select Source Address
−
        ELSEWHERE IN THIS DOCUMENT FOR SENDING ICMP ERROR MESSAGES.
 
  
−
        DISCUSSION:
+
            GET_SRCADDR(remote, TOS) -> local
−
              These rules will prevent the "broadcast storms" that have
 
−
              resulted from hosts returning ICMP error messages in
 
−
              response to broadcast datagrams. For example, a broadcast
 
−
              UDP segment to a non-existent port could trigger a flood
 
−
              of ICMP Destination Unreachable datagrams from all
 
−
              machines that do not have a client for that destination
 
−
              port.  On a large Ethernet, the resulting collisions can
 
−
              render the network useless for a second or more.
 
  
−
              Every datagram that is broadcast on the connected network
+
            remote = remote IP address
−
              should have a valid IP broadcast address as its IP
+
            TOS = Type-of-Service
−
              destination (see Section 3.3.6).  However, some hosts
+
            local = local IP address
−
              violate this rule.  To be certain to detect broadcast
 
−
              datagrams, therefore, hosts are required to check for a
 
−
              link-layer broadcast as well as an IP-layer broadcast
 
−
              address.
 
  
−
        IMPLEMENTATION:
+
        See Section 3.3.4.3.
−
              This requires that the link layer inform the IP layer when
 
−
              a link-layer broadcast datagram has been received; see
 
−
              Section 2.4.
 
  
−
        3.2.2.1  Destination Unreachable: RFC-792
+
  *    Find Maximum Datagram Sizes
  
−
            The following additional codes are hereby defined:
+
            GET_MAXSIZES(local, remote, TOS) -> MMS_R, MMS_S
  
−
                    6 = destination network unknown
+
            MMS_R = maximum receive transport-message size.
 +
            MMS_S = maximum send transport-message size.
 +
            (local, remote, TOS defined above)
  
 +
        See Sections 3.3.2 and 3.3.3.
  
 +
  *    Advice on Delivery Success
  
−
Internet Engineering Task Force                                [Page 39]
+
            ADVISE_DELIVPROB(sense, local, remote, TOS)
  
 +
        Here the parameter sense is a 1-bit flag indicating whether
 +
        positive or negative advice is being given; see the
 +
        discussion in Section 3.3.1.4. The other parameters were
 +
        defined earlier.
  
 +
  *    Send ICMP Message
  
 +
            SEND_ICMP(src, dst, TOS, TTL, BufPTR, len, Id, DF, opt)
 +
                  -> result
  
 
RFC1122                      INTERNET LAYER                October 1989
 
RFC1122                      INTERNET LAYER                October 1989
  
 +
            (Parameters defined in RFC-791).
  
−
                    7 = destination host unknown
+
        Passing an Id parameter is optional; see Section 3.2.1.5.
−
 
+
        The transport layer MUST be able to send certain ICMP
−
                    8 = source host isolated
+
        messages:  Port Unreachable or any of the query-type
−
 
+
        messages.  This function could be considered to be a special
−
                    9 = communication with destination network
+
        case of the SEND() call, of course; we describe it separately
−
                            administratively prohibited
+
        for clarity.
−
 
 
−
                  10 = communication with destination host
 
−
                            administratively prohibited
 
−
 
 
−
                  11 = network unreachable for type of service
 
−
 
 
−
                  12 = host unreachable for type of service
 
−
 
 
−
            A host SHOULD generate Destination Unreachable messages with
 
−
            code:
 
−
 
 
−
            2    (Protocol Unreachable), when the designated transport
 
−
                protocol is not supported; or
 
  
−
            3   (Port Unreachable), when the designated transport
+
   *    Receive ICMP Message
−
                protocol (e.g., UDP) is unable to demultiplex the
 
−
                datagram but has no protocol mechanism to inform the
 
−
                sender.
 
  
−
            A Destination Unreachable message that is received MUST be
+
            RECV_ICMP(BufPTR ) -> result, src, dst, len, opt
−
            reported to the transport layer.  The transport layer SHOULD
 
−
            use the information appropriately; for example, see Sections
 
−
            4.1.3.3, 4.2.3.9, and 4.2.4 below.  A transport protocol
 
−
            that has its own mechanism for notifying the sender that a
 
−
            port is unreachable (e.g., TCP, which sends RST segments)
 
−
            MUST nevertheless accept an ICMP Port Unreachable for the
 
−
            same purpose.
 
  
−
            A Destination Unreachable message that is received with code
+
            (Parameters defined in RFC-791).
−
            0 (Net), 1 (Host), or 5 (Bad Source Route) may result from a
 
−
            routing transient and MUST therefore be interpreted as only
 
−
            a hint, not proof, that the specified destination is
 
−
            unreachable [IP:11].  For example, it MUST NOT be used as
 
−
            proof of a dead gateway (see Section 3.3.1).
 
  
−
        3.2.2.2  Redirect: RFC-792
+
        The IP layer MUST pass certain ICMP messages up to the
 +
        appropriate transport-layer routine. This function could be
 +
        considered to be a special case of the RECV() call, of
 +
        course; we describe it separately for clarity.
  
−
            A host SHOULD NOT send an ICMP Redirect message; Redirects
+
        For an ICMP error message, the data that is passed up MUST
−
            are to be sent only by gateways.
+
        include the original Internet header plus all the octets of
 +
        the original message that are included in the ICMP message.
 +
        This data will be used by the transport layer to locate the
 +
        connection state information, if any.
  
−
            A host receiving a Redirect message MUST update its routing
+
        In particular, the following ICMP messages are to be passed
−
            information accordingly.  Every host MUST be prepared to
+
        up:
  
 +
        o    Destination Unreachable
  
 +
        o    Source Quench
  
−
Internet Engineering Task Force                                [Page 40]
+
        o    Echo Reply (to ICMP user interface, unless the Echo
 +
            Request originated in the IP layer)
  
 +
        o    Timestamp Reply (to ICMP user interface)
  
 +
        o    Time Exceeded
  
 +
  DISCUSSION:
 +
        In the future, there may be additions to this interface to
 +
        pass path data (see Section 3.3.1.3) between the IP and
 +
        transport layers.
  
 
RFC1122                      INTERNET LAYER                October 1989
 
RFC1122                      INTERNET LAYER                October 1989
  
 +
3.5  INTERNET LAYER REQUIREMENTS SUMMARY
  
−
            accept both Host and Network Redirects and to process them
+
                                              |        | | | |S| |
−
            as described in Section 3.3.1.2 below.
+
                                              |        | | | |H| |F
−
 
+
                                              |        | | | |O|M|o
−
            A Redirect message SHOULD be silently discarded if the new
+
                                              |        | |S| |U|U|o
−
            gateway address it specifies is not on the same connected
+
                                              |        | |H| |L|S|t
−
            (sub-) net through which the Redirect arrived [INTRO:2,
+
                                              |        |M|O| |D|T|n
−
            Appendix A], or if the source of the Redirect is not the
+
                                              |        |U|U|M| | |o
−
            current first-hop gateway for the specified destination (see
+
                                              |        |S|L|A|N|N|t
−
            Section 3.3.1).
+
                                              |        |T|D|Y|O|O|t
−
 
+
FEATURE                                          |SECTION | | | |T|T|e
−
        3.2.2.3  Source Quench: RFC-792
+
-------------------------------------------------|--------|-|-|-|-|-|--
−
 
+
                                              |        | | | | | |
−
            A host MAY send a Source Quench message if it is
+
Implement IP and ICMP                            |3.1    |x| | | | |
−
            approaching, or has reached, the point at which it is forced
+
Handle remote multihoming in application layer  |3.1    |x| | | | |
−
            to discard incoming datagrams due to a shortage of
+
Support local multihoming                        |3.1     | | |x| | |
−
            reassembly buffers or other resources. See Section 2.2.3 of
+
Meet gateway specs if forward datagrams          |3.1    |x| | | | |
−
            [INTRO:2] for suggestions on when to send Source Quench.
+
Configuration switch for embedded gateway       |3.1    |x| | | | |1
−
 
+
Config switch default to non-gateway          |3.1    |x| | | | |1
−
            If a Source Quench message is received, the IP layer MUST
+
Auto-config based on number of interfaces    |3.1    | | | | |x|1
−
            report it to the transport layer (or ICMP processing). In
+
Able to log discarded datagrams                  |3.1    | |x| | | |
−
            general, the transport or application layer SHOULD implement
+
Record in counter                            |3.1    | |x| | | |
−
            a mechanism to respond to Source Quench for any protocol
+
                                              |        | | | | | |
−
            that can send a sequence of datagrams to the same
+
Silently discard Version != 4                    |3.2.1.1 |x| | | | |
−
            destination and which can reasonably be expected to maintain
+
Verify IP checksum, silently discard bad dgram  |3.2.1.2 |x| | | | |
−
            enough state information to make this feasible. See Section
+
Addressing:                                     |        | | | | | |
−
            4 for the handling of Source Quench by TCP and UDP.
+
  Subnet addressing (RFC-950)                    |3.2.1.3 |x| | | | |
−
 
+
  Src address must be host's own IP address      |3.2.1.3 |x| | | | |
−
            DISCUSSION:
+
  Silently discard datagram with bad dest addr  |3.2.1.3 |x| | | | |
−
                A Source Quench may be generated by the target host or
+
  Silently discard datagram with bad src addr    |3.2.1.3 |x| | | | |
−
                by some gateway in the path of a datagram. The host
+
Support reassembly                               |3.2.1.4 |x| | | | |
−
                receiving a Source Quench should throttle itself back
+
Retain same Id field in identical datagram      |3.2.1.5 | | |x| | |
−
                for a period of time, then gradually increase the
+
                                              |        | | | | | |
−
                transmission rate again. The mechanism to respond to
+
TOS:                                             |        | | | | | |
−
                Source Quench may be in the transport layer (for
+
  Allow transport layer to set TOS              |3.2.1.6 |x| | | | |
−
                connection-oriented protocols like TCP) or in the
+
  Pass received TOS up to transport layer       |3.2.1.6 | |x| | | |
−
                application layer (for protocols that are built on top
+
  Use RFC-795 link-layer mappings for TOS        |3.2.1.6 | | | |x| |
−
                of UDP).
+
TTL:                                             |        | | | | | |
−
 
+
  Send packet with TTL of 0                      |3.2.1.7 | | | | |x|
−
                A mechanism has been proposed [IP:14] to make the IP
+
  Discard received packets with TTL < 2          |3.2.1.7 | | | | |x|
−
                layer respond directly to Source Quench by controlling
+
  Allow transport layer to set TTL              |3.2.1.7 |x| | | | |
−
                the rate at which datagrams are sent, however, this
+
  Fixed TTL is configurable                      |3.2.1.7 |x| | | | |
−
                proposal is currently experimental and not currently
+
                                              |        | | | | | |
−
                recommended.
+
IP Options:                                     |        | | | | | |
−
 
+
  Allow transport layer to send IP options      |3.2.1.8 |x| | | | |
−
        3.2.2.4  Time Exceeded: RFC-792
+
  Pass all IP options rcvd to higher layer       |3.2.1.8 |x| | | | |
−
 
 
−
            An incoming Time Exceeded message MUST be passed to the
 
−
            transport layer.
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 41]
 
−
 
 
−
 
 
−
 
 
  
 
RFC1122                      INTERNET LAYER                October 1989
 
RFC1122                      INTERNET LAYER                October 1989
  
 +
  IP layer silently ignore unknown options      |3.2.1.8 |x| | | | |
 +
  Security option                                |3.2.1.8a| | |x| | |
 +
  Send Stream Identifier option                  |3.2.1.8b| | | |x| |
 +
  Silently ignore Stream Identifer option        |3.2.1.8b|x| | | | |
 +
  Record Route option                            |3.2.1.8d| | |x| | |
 +
  Timestamp option                              |3.2.1.8e| | |x| | |
 +
Source Route Option:                            |        | | | | | |
 +
  Originate & terminate Source Route options    |3.2.1.8c|x| | | | |
 +
  Datagram with completed SR passed up to TL    |3.2.1.8c|x| | | | |
 +
  Build correct (non-redundant) return route    |3.2.1.8c|x| | | | |
 +
  Send multiple SR options in one header        |3.2.1.8c| | | | |x|
 +
                                              |        | | | | | |
 +
ICMP:                                            |        | | | | | |
 +
  Silently discard ICMP msg with unknown type    |3.2.2  |x| | | | |
 +
  Include more than 8 octets of orig datagram    |3.2.2  | | |x| | |
 +
  Included octets same as received          |3.2.2  |x| | | | |
 +
  Demux ICMP Error to transport protocol        |3.2.2  |x| | | | |
 +
  Send ICMP error message with TOS=0            |3.2.2  | |x| | | |
 +
  Send ICMP error message for:                  |        | | | | | |
 +
- ICMP error msg                              |3.2.2  | | | | |x|
 +
- IP b'cast or IP m'cast                      |3.2.2  | | | | |x|
 +
- Link-layer b'cast                          |3.2.2  | | | | |x|
 +
- Non-initial fragment                        |3.2.2  | | | | |x|
 +
- Datagram with non-unique src address        |3.2.2  | | | | |x|
 +
  Return ICMP error msgs (when not prohibited)  |3.3.8  |x| | | | |
 +
                                              |        | | | | | |
 +
  Dest Unreachable:                              |        | | | | | |
 +
Generate Dest Unreachable (code 2/3)        |3.2.2.1 | |x| | | |
 +
Pass ICMP Dest Unreachable to higher layer  |3.2.2.1 |x| | | | |
 +
Higher layer act on Dest Unreach            |3.2.2.1 | |x| | | |
 +
  Interpret Dest Unreach as only hint        |3.2.2.1 |x| | | | |
 +
  Redirect:                                      |        | | | | | |
 +
Host send Redirect                          |3.2.2.2 | | | |x| |
 +
Update route cache when recv Redirect        |3.2.2.2 |x| | | | |
 +
Handle both Host and Net Redirects          |3.2.2.2 |x| | | | |
 +
Discard illegal Redirect                    |3.2.2.2 | |x| | | |
 +
  Source Quench:                                |        | | | | | |
 +
Send Source Quench if buffering exceeded    |3.2.2.3 | | |x| | |
 +
Pass Source Quench to higher layer          |3.2.2.3 |x| | | | |
 +
Higher layer act on Source Quench            |3.2.2.3 | |x| | | |
 +
  Time Exceeded: pass to higher layer            |3.2.2.4 |x| | | | |
 +
  Parameter Problem:                            |        | | | | | |
 +
Send Parameter Problem messages              |3.2.2.5 | |x| | | |
 +
Pass Parameter Problem to higher layer      |3.2.2.5 |x| | | | |
 +
Report Parameter Problem to user            |3.2.2.5 | | |x| | |
 +
                                              |        | | | | | |
 +
  ICMP Echo Request or Reply:                    |        | | | | | |
 +
Echo server and Echo client                  |3.2.2.6 |x| | | | |
  
−
            DISCUSSION:
+
RFC1122                      INTERNET LAYER                October 1989
−
                A gateway will send a Time Exceeded Code 0 (In Transit)
 
−
                message when it discards a datagram due to an expired
 
−
                TTL field.  This indicates either a gateway routing
 
−
                loop or too small an initial TTL value.
 
−
 
 
−
                A host may receive a Time Exceeded Code 1 (Reassembly
 
−
                Timeout) message from a destination host that has timed
 
−
                out and discarded an incomplete datagram; see Section
 
−
                3.3.2 below.  In the future, receipt of this message
 
−
                might be part of some "MTU discovery" procedure, to
 
−
                discover the maximum datagram size that can be sent on
 
−
                the path without fragmentation.
 
−
 
 
−
        3.2.2.5  Parameter Problem: RFC-792
 
−
 
 
−
            A host SHOULD generate Parameter Problem messages.  An
 
−
            incoming Parameter Problem message MUST be passed to the
 
−
            transport layer, and it MAY be reported to the user.
 
−
 
 
−
            DISCUSSION:
 
−
                The ICMP Parameter Problem message is sent to the
 
−
                source host for any problem not specifically covered by
 
−
                another ICMP message.  Receipt of a Parameter Problem
 
−
                message generally indicates some local or remote
 
−
                implementation error.
 
−
 
 
−
            A new variant on the Parameter Problem message is hereby
 
−
            defined:
 
−
              Code 1 = required option is missing.
 
−
 
 
−
            DISCUSSION:
 
−
                This variant is currently in use in the military
 
−
                community for a missing security option.
 
−
 
 
−
        3.2.2.6  Echo Request/Reply: RFC-792
 
−
 
 
−
            Every host MUST implement an ICMP Echo server function that
 
−
            receives Echo Requests and sends corresponding Echo Replies.
 
−
            A host SHOULD also implement an application-layer interface
 
−
            for sending an Echo Request and receiving an Echo Reply, for
 
−
            diagnostic purposes.
 
−
 
 
−
            An ICMP Echo Request destined to an IP broadcast or IP
 
−
            multicast address MAY be silently discarded.
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 42]
 
−
 
 
−
 
 
  
 +
Echo client                                  |3.2.2.6 | |x| | | |
 +
Discard Echo Request to broadcast address    |3.2.2.6 | | |x| | |
 +
Discard Echo Request to multicast address    |3.2.2.6 | | |x| | |
 +
Use specific-dest addr as Echo Reply src    |3.2.2.6 |x| | | | |
 +
Send same data in Echo Reply                |3.2.2.6 |x| | | | |
 +
Pass Echo Reply to higher layer              |3.2.2.6 |x| | | | |
 +
Reflect Record Route, Time Stamp options    |3.2.2.6 | |x| | | |
 +
Reverse and reflect Source Route option      |3.2.2.6 |x| | | | |
 +
                                              |        | | | | | |
 +
  ICMP Information Request or Reply:            |3.2.2.7 | | | |x| |
 +
  ICMP Timestamp and Timestamp Reply:            |3.2.2.8 | | |x| | |
 +
Minimize delay variability                  |3.2.2.8 | |x| | | |1
 +
Silently discard b'cast Timestamp            |3.2.2.8 | | |x| | |1
 +
Silently discard m'cast Timestamp            |3.2.2.8 | | |x| | |1
 +
Use specific-dest addr as TS Reply src      |3.2.2.8 |x| | | | |1
 +
Reflect Record Route, Time Stamp options    |3.2.2.6 | |x| | | |1
 +
Reverse and reflect Source Route option      |3.2.2.8 |x| | | | |1
 +
Pass Timestamp Reply to higher layer        |3.2.2.8 |x| | | | |1
 +
Obey rules for "standard value"              |3.2.2.8 |x| | | | |1
 +
                                              |        | | | | | |
 +
  ICMP Address Mask Request and Reply:          |        | | | | | |
 +
Addr Mask source configurable                |3.2.2.9 |x| | | | |
 +
Support static configuration of addr mask    |3.2.2.9 |x| | | | |
 +
Get addr mask dynamically during booting    |3.2.2.9 | | |x| | |
 +
Get addr via ICMP Addr Mask Request/Reply    |3.2.2.9 | | |x| | |
 +
  Retransmit Addr Mask Req if no Reply      |3.2.2.9 |x| | | | |3
 +
  Assume default mask if no Reply            |3.2.2.9 | |x| | | |3
 +
  Update address mask from first Reply only  |3.2.2.9 |x| | | | |3
 +
Reasonableness check on Addr Mask            |3.2.2.9 | |x| | | |
 +
Send unauthorized Addr Mask Reply msgs      |3.2.2.9 | | | | |x|
 +
  Explicitly configured to be agent          |3.2.2.9 |x| | | | |
 +
Static config=> Addr-Mask-Authoritative flag |3.2.2.9 | |x| | | |
 +
  Broadcast Addr Mask Reply when init.      |3.2.2.9 |x| | | | |3
 +
                                              |        | | | | | |
 +
ROUTING OUTBOUND DATAGRAMS:                      |        | | | | | |
 +
  Use address mask in local/remote decision      |3.3.1.1 |x| | | | |
 +
  Operate with no gateways on conn network      |3.3.1.1 |x| | | | |
 +
  Maintain "route cache" of next-hop gateways    |3.3.1.2 |x| | | | |
 +
  Treat Host and Net Redirect the same          |3.3.1.2 | |x| | | |
 +
  If no cache entry, use default gateway        |3.3.1.2 |x| | | | |
 +
Support multiple default gateways            |3.3.1.2 |x| | | | |
 +
  Provide table of static routes                |3.3.1.2 | | |x| | |
 +
Flag: route overridable by Redirects        |3.3.1.2 | | |x| | |
 +
  Key route cache on host, not net address      |3.3.1.3 | | |x| | |
 +
  Include TOS in route cache                    |3.3.1.3 | |x| | | |
 +
                                              |        | | | | | |
 +
  Able to detect failure of next-hop gateway    |3.3.1.4 |x| | | | |
 +
  Assume route is good forever                  |3.3.1.4 | | | |x| |
  
 
RFC1122                      INTERNET LAYER                October 1989
 
RFC1122                      INTERNET LAYER                October 1989
  
 +
  Ping gateways continuously                    |3.3.1.4 | | | | |x|
 +
  Ping only when traffic being sent              |3.3.1.4 |x| | | | |
 +
  Ping only when no positive indication          |3.3.1.4 |x| | | | |
 +
  Higher and lower layers give advice            |3.3.1.4 | |x| | | |
 +
  Switch from failed default g'way to another    |3.3.1.5 |x| | | | |
 +
  Manual method of entering config info          |3.3.1.6 |x| | | | |
 +
                                              |        | | | | | |
 +
REASSEMBLY and FRAGMENTATION:                    |        | | | | | |
 +
  Able to reassemble incoming datagrams          |3.3.2  |x| | | | |
 +
At least 576 byte datagrams                  |3.3.2  |x| | | | |
 +
EMTU_R configurable or indefinite            |3.3.2  | |x| | | |
 +
  Transport layer able to learn MMS_R            |3.3.2  |x| | | | |
 +
  Send ICMP Time Exceeded on reassembly timeout  |3.3.2  |x| | | | |
 +
Fixed reassembly timeout value              |3.3.2  | |x| | | |
 +
                                              |        | | | | | |
 +
  Pass MMS_S to higher layers                    |3.3.3  |x| | | | |
 +
  Local fragmentation of outgoing packets        |3.3.3  | | |x| | |
 +
  Else don't send bigger than MMS_S          |3.3.3  |x| | | | |
 +
  Send max 576 to off-net destination            |3.3.3  | |x| | | |
 +
  All-Subnets-MTU configuration flag            |3.3.3  | | |x| | |
 +
                                              |        | | | | | |
 +
MULTIHOMING:                                    |        | | | | | |
 +
  Reply with same addr as spec-dest addr        |3.3.4.2 | |x| | | |
 +
  Allow application to choose local IP addr      |3.3.4.2 |x| | | | |
 +
  Silently discard d'gram in "wrong" interface  |3.3.4.2 | | |x| | |
 +
  Only send d'gram through "right" interface    |3.3.4.2 | | |x| | |4
 +
                                              |        | | | | | |
 +
SOURCE-ROUTE FORWARDING:                        |        | | | | | |
 +
  Forward datagram with Source Route option      |3.3.5  | | |x| | |1
 +
Obey corresponding gateway rules            |3.3.5  |x| | | | |1
 +
  Update TTL by gateway rules                |3.3.5  |x| | | | |1
 +
  Able to generate ICMP err code 4, 5        |3.3.5  |x| | | | |1
 +
  IP src addr not local host                |3.3.5  | | |x| | |1
 +
  Update Timestamp, Record Route options    |3.3.5  |x| | | | |1
 +
Configurable switch for non-local SRing      |3.3.5  |x| | | | |1
 +
  Defaults to OFF                            |3.3.5  |x| | | | |1
 +
Satisfy gwy access rules for non-local SRing |3.3.5  |x| | | | |1
 +
If not forward, send Dest Unreach (cd 5)    |3.3.5  | |x| | | |2
 +
                                              |        | | | | | |
 +
BROADCAST:                                      |        | | | | | |
 +
  Broadcast addr as IP source addr              |3.2.1.3 | | | | |x|
 +
  Receive 0 or -1 broadcast formats OK          |3.3.6  | |x| | | |
 +
  Config'ble option to send 0 or -1 b'cast      |3.3.6  | | |x| | |
 +
Default to -1 broadcast                      |3.3.6  | |x| | | |
 +
  Recognize all broadcast address formats        |3.3.6  |x| | | | |
 +
  Use IP b'cast/m'cast addr in link-layer b'cast |3.3.6  |x| | | | |
 +
  Silently discard link-layer-only b'cast dg's  |3.3.6  | |x| | | |
 +
  Use Limited Broadcast addr for connected net  |3.3.6  | |x| | | |
  
−
            DISCUSSION:
+
RFC1122                      INTERNET LAYER                October 1989
−
                This neutral provision results from a passionate debate
 
−
                between those who feel that ICMP Echo to a broadcast
 
−
                address provides a valuable diagnostic capability and
 
−
                those who feel that misuse of this feature can too
 
−
                easily create packet storms.
 
−
 
 
−
            The IP source address in an ICMP Echo Reply MUST be the same
 
−
            as the specific-destination address (defined in Section
 
−
            3.2.1.3) of the corresponding ICMP Echo Request message.
 
−
 
 
−
            Data received in an ICMP Echo Request MUST be entirely
 
−
            included in the resulting Echo Reply.  However, if sending
 
−
            the Echo Reply requires intentional fragmentation that is
 
−
            not implemented, the datagram MUST be truncated to maximum
 
−
            transmission size (see Section 3.3.3) and sent.
 
−
 
 
−
            Echo Reply messages MUST be passed to the ICMP user
 
−
            interface, unless the corresponding Echo Request originated
 
−
            in the IP layer.
 
−
 
 
−
            If a Record Route and/or Time Stamp option is received in an
 
−
            ICMP Echo Request, this option (these options) SHOULD be
 
−
            updated to include the current host and included in the IP
 
−
            header of the Echo Reply message, without "truncation".
 
−
            Thus, the recorded route will be for the entire round trip.
 
−
 
 
−
            If a Source Route option is received in an ICMP Echo
 
−
            Request, the return route MUST be reversed and used as a
 
−
            Source Route option for the Echo Reply message.
 
  
−
        3.2.2.7 Information Request/Reply: RFC-792
+
                                              |        | | | | | |
−
 
+
MULTICAST:                                      |        | | | | | |
−
            A host SHOULD NOT implement these messages.
+
  Support local IP multicasting (RFC-1112)      |3.3.7   | |x| | | |
−
 
+
  Support IGMP (RFC-1112)                        |3.3.7  | | |x| | |
−
            DISCUSSION:
+
  Join all-hosts group at startup                |3.3.7  | |x| | | |
−
                The Information Request/Reply pair was intended to
+
  Higher layers learn i'face m'cast capability  |3.3.7  | |x| | | |
−
                support self-configuring systems such as diskless
+
                                              |        | | | | | |
−
                workstations, to allow them to discover their IP
+
INTERFACE:                                       |        | | | | | |
−
                network numbers at boot time. However, the RARP and
+
  Allow transport layer to use all IP mechanisms |3.4    |x| | | | |
−
                BOOTP protocols provide better mechanisms for a host to
+
  Pass interface ident up to transport layer    |3.4    |x| | | | |
−
                discover its own IP address.
+
  Pass all IP options up to transport layer      |3.4    |x| | | | |
−
 
+
  Transport layer can send certain ICMP messages |3.4    |x| | | | |
−
        3.2.2.8 Timestamp and Timestamp Reply: RFC-792
+
  Pass spec'd ICMP messages up to transp. layer  |3.4    |x| | | | |
 +
  Include IP hdr+8 octets or more from orig. |3.4    |x| | | | |
 +
  Able to leap tall buildings at a single bound |3.5    | |x| | | |
  
−
            A host MAY implement Timestamp and Timestamp Reply.  If they
+
Footnotes:
−
            are implemented, the following rules MUST be followed.
 
  
 +
(1)  Only if feature is implemented.
  
 +
(2)  This requirement is overruled if datagram is an ICMP error message.
  
 +
(3)  Only if feature is implemented and is configured "on".
  
−
Internet Engineering Task Force                                [Page 43]
+
(4)  Unless has embedded gateway functionality or is source routed.
  
 +
RFC1122                  TRANSPORT LAYER -- UDP            October 1989
  
 +
== TRANSPORT PROTOCOLS ==
  
 +
4.1  USER DATAGRAM PROTOCOL -- UDP
  
−
RFC1122                      INTERNET LAYER                October 1989
+
  4.1.1  INTRODUCTION
  
 +
      The User Datagram Protocol UDP [UDP:1] offers only a minimal
 +
      transport service -- non-guaranteed datagram delivery -- and
 +
      gives applications direct access to the datagram service of the
 +
      IP layer.  UDP is used by applications that do not require the
 +
      level of service of TCP or that wish to use communications
 +
      services (e.g., multicast or broadcast delivery) not available
 +
      from TCP.
  
−
            o    The ICMP Timestamp server function returns a Timestamp
+
      UDP is almost a null protocol; the only services it provides
−
                Reply to every Timestamp message that is received.  If
+
      over IP are checksumming of data and multiplexing by port
−
                this function is implemented, it SHOULD be designed for
+
      number.  Therefore, an application program running over UDP
−
                minimum variability in delay (e.g., implemented in the
+
      must deal directly with end-to-end communication problems that
−
                kernel to avoid delay in scheduling a user process).
+
      a connection-oriented protocol would have handled -- e.g.,
 +
      retransmission for reliable delivery, packetization and
 +
      reassembly, flow control, congestion avoidance, etc., when
 +
      these are required.  The fairly complex coupling between IP and
 +
      TCP will be mirrored in the coupling between UDP and many
 +
      applications using UDP.
  
−
            The following cases for Timestamp are to be handled
+
  4.1.2  PROTOCOL WALK-THROUGH
−
            according to the corresponding rules for ICMP Echo:
 
  
−
            o    An ICMP Timestamp Request message to an IP broadcast or
+
      There are no known errors in the specification of UDP.
−
                IP multicast address MAY be silently discarded.
 
  
−
            o   The IP source address in an ICMP Timestamp Reply MUST
+
   4.1.3  SPECIFIC ISSUES
−
                be the same as the specific-destination address of the
 
−
                corresponding Timestamp Request message.
 
  
−
            o    If a Source-route option is received in an ICMP Echo
+
      4.1.3.1  Ports
−
                Request, the return route MUST be reversed and used as
 
−
                a Source Route option for the Timestamp Reply message.
 
  
−
            o    If a Record Route and/or Timestamp option is received
+
        UDP well-known ports follow the same rules as TCP well-known
−
                in a Timestamp Request, this (these) option(s) SHOULD
+
        ports; see Section 4.2.2.1 below.
−
                be updated to include the current host and included in
 
−
                the IP header of the Timestamp Reply message.
 
  
−
            o    Incoming Timestamp Reply messages MUST be passed up to
+
        If a datagram arrives addressed to a UDP port for which
−
                the ICMP user interface.
+
        there is no pending LISTEN call, UDP SHOULD send an ICMP
 +
        Port Unreachable message.
  
−
            The preferred form for a timestamp value (the "standard
+
      4.1.3.2 IP Options
−
            value") is in units of milliseconds since midnight Universal
 
−
            Time. However, it may be difficult to provide this value
 
−
            with millisecond resolution. For example, many systems use
 
−
            clocks that update only at line frequency, 50 or 60 times
 
−
            per second.  Therefore, some latitude is allowed in a
 
−
            "standard value":
 
  
−
            (a)  A "standard value" MUST be updated at least 15 times
+
        UDP MUST pass any IP option that it receives from the IP
−
                per second (i.e., at most the six low-order bits of the
+
        layer transparently to the application layer.
−
                value may be undefined).
 
  
−
            (b)  The accuracy of a "standard value" MUST approximate
+
        An application MUST be able to specify IP options to be sent
−
                that of operator-set CPU clocks, i.e., correct within a
+
        in its UDP datagrams, and UDP MUST pass these options to the
−
                few minutes.
+
        IP layer.
  
 +
RFC1122                  TRANSPORT LAYER -- UDP            October 1989
  
 +
        DISCUSSION:
 +
              At present, the only options that need be passed
 +
              through UDP are Source Route, Record Route, and Time
 +
              Stamp.  However, new options may be defined in the
 +
              future, and UDP need not and should not make any
 +
              assumptions about the format or content of options it
 +
              passes to or from the application; an exception to this
 +
              might be an IP-layer security option.
  
 +
              An application based on UDP will need to obtain a
 +
              source route from a request datagram and supply a
 +
              reversed route for sending the corresponding reply.
  
 +
      4.1.3.3  ICMP Messages
  
 +
        UDP MUST pass to the application layer all ICMP error
 +
        messages that it receives from the IP layer.  Conceptually
 +
        at least, this may be accomplished with an upcall to the
 +
        ERROR_REPORT routine (see Section 4.2.4.1).
  
 +
        DISCUSSION:
 +
              Note that ICMP error messages resulting from sending a
 +
              UDP datagram are received asynchronously.  A UDP-based
 +
              application that wants to receive ICMP error messages
 +
              is responsible for maintaining the state necessary to
 +
              demultiplex these messages when they arrive; for
 +
              example, the application may keep a pending receive
 +
              operation for this purpose.  The application is also
 +
              responsible to avoid confusion from a delayed ICMP
 +
              error message resulting from an earlier use of the same
 +
              port(s).
  
 +
      4.1.3.4  UDP Checksums
  
−
Internet Engineering Task Force                                [Page 44]
+
        A host MUST implement the facility to generate and validate
 +
        UDP checksums.  An application MAY optionally be able to
 +
        control whether a UDP checksum will be generated, but it
 +
        MUST default to checksumming on.
  
 +
        If a UDP datagram is received with a checksum that is non-
 +
        zero and invalid, UDP MUST silently discard the datagram.
 +
        An application MAY optionally be able to control whether UDP
 +
        datagrams without checksums should be discarded or passed to
 +
        the application.
  
 +
        DISCUSSION:
 +
              Some applications that normally run only across local
 +
              area networks have chosen to turn off UDP checksums for
  
 +
RFC1122                  TRANSPORT LAYER -- UDP            October 1989
  
−
RFC1122                      INTERNET LAYER                October 1989
+
              efficiency.  As a result, numerous cases of undetected
 +
              errors have been reported.  The advisability of ever
 +
              turning off UDP checksumming is very controversial.
  
 +
        IMPLEMENTATION:
 +
              There is a common implementation error in UDP
 +
              checksums.  Unlike the TCP checksum, the UDP checksum
 +
              is optional; the value zero is transmitted in the
 +
              checksum field of a UDP header to indicate the absence
 +
              of a checksum.  If the transmitter really calculates a
 +
              UDP checksum of zero, it must transmit the checksum as
 +
              all 1's (65535).  No special action is required at the
 +
              receiver, since zero and 65535 are equivalent in 1's
 +
              complement arithmetic.
  
−
        3.2.2.9 Address Mask Request/Reply: RFC-950
+
      4.1.3.5 UDP Multihoming
  
−
            A host MUST support the first, and MAY implement all three,
+
        When a UDP datagram is received, its specific-destination
−
            of the following methods for determining the address mask(s)
+
        address MUST be passed up to the application layer.
−
            corresponding to its IP address(es):
 
  
−
            (1)  static configuration information;
+
        An application program MUST be able to specify the IP source
 +
        address to be used for sending a UDP datagram or to leave it
 +
        unspecified (in which case the networking software will
 +
        choose an appropriate source address). There SHOULD be a
 +
        way to communicate the chosen source address up to the
 +
        application layer (e.g, so that the application can later
 +
        receive a reply datagram only from the corresponding
 +
        interface).
  
−
            (2)  obtaining the address mask(s) dynamically as a side-
+
        DISCUSSION:
−
                effect of the system initialization process (see
+
              A request/response application that uses UDP should use
−
                [INTRO:1]); and
+
              a source address for the response that is the same as
 +
              the specific destination address of the request.  See
 +
              the "General Issues" section of [INTRO:1].
  
−
            (3) sending ICMP Address Mask Request(s) and receiving ICMP
+
      4.1.3.6 Invalid Addresses
−
                Address Mask Reply(s).
 
  
−
            The choice of method to be used in a particular host MUST be
+
        A UDP datagram received with an invalid IP source address
−
            configurable.
+
        (e.g., a broadcast or multicast address) must be discarded
 +
        by UDP or by the IP layer (see Section 3.2.1.3).
  
−
            When method (3), the use of Address Mask messages, is
+
        When a host sends a UDP datagram, the source address MUST be
−
            enabled, then:
+
        (one of) the IP address(es) of the host.
  
−
            (a)  When it initializes, the host MUST broadcast an Address
+
  4.1.4 UDP/APPLICATION LAYER INTERFACE
−
                Mask Request message on the connected network
 
−
                corresponding to the IP address.  It MUST retransmit
 
−
                this message a small number of times if it does not
 
−
                receive an immediate Address Mask Reply.
 
  
−
            (b)  Until it has received an Address Mask Reply, the host
+
      The application interface to UDP MUST provide the full services
−
                SHOULD assume a mask appropriate for the address class
+
      of the IP/transport interface described in Section 3.4 of this
−
                of the IP address, i.e., assume that the connected
 
−
                network is not subnetted.
 
  
−
             (c)  The first Address Mask Reply message received MUST be
+
RFC1122                  TRANSPORT LAYER -- UDP             October 1989
−
                used to set the address mask corresponding to the
 
−
                particular local IP address.  This is true even if the
 
−
                first Address Mask Reply message is "unsolicited", in
 
−
                which case it will have been broadcast and may arrive
 
−
                after the host has ceased to retransmit Address Mask
 
−
                Requests.  Once the mask has been set by an Address
 
−
                Mask Reply, later Address Mask Reply messages MUST be
 
−
                (silently) ignored.
 
  
−
            Conversely, if Address Mask messages are disabled, then no
+
      document.  Thus, an application using UDP needs the functions
−
            ICMP Address Mask Requests will be sent, and any ICMP
+
      of the GET_SRCADDR(), GET_MAXSIZES(), ADVISE_DELIVPROB(), and
−
            Address Mask Replies received for that local IP address MUST
+
      RECV_ICMP() calls described in Section 3.4.  For example,
−
            be (silently) ignored.
+
      GET_MAXSIZES() can be used to learn the effective maximum UDP
−
 
+
      maximum datagram size for a particular {interface,remote
−
            A host SHOULD make some reasonableness check on any address
+
      host,TOS} triplet.
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 45]
 
  
 +
      An application-layer program MUST be able to set the TTL and
 +
      TOS values as well as IP options for sending a UDP datagram,
 +
      and these values must be passed transparently to the IP layer.
 +
      UDP MAY pass the received TOS up to the application layer.
  
 +
  4.1.5  UDP REQUIREMENTS SUMMARY
  
 +
                                              |        | | | |S| |
 +
                                              |        | | | |H| |F
 +
                                              |        | | | |O|M|o
 +
                                              |        | |S| |U|U|o
 +
                                              |        | |H| |L|S|t
 +
                                              |        |M|O| |D|T|n
 +
                                              |        |U|U|M| | |o
 +
                                              |        |S|L|A|N|N|t
 +
                                              |        |T|D|Y|O|O|t
 +
FEATURE                                          |SECTION | | | |T|T|e
 +
-------------------------------------------------|--------|-|-|-|-|-|--
 +
                                              |        | | | | | |
 +
UDP                                          |        | | | | | |
 +
-------------------------------------------------|--------|-|-|-|-|-|--
 +
                                              |        | | | | | |
 +
UDP send Port Unreachable                        |4.1.3.1 | |x| | | |
 +
                                              |        | | | | | |
 +
IP Options in UDP                                |        | | | | | |
 +
- Pass rcv'd IP options to applic layer        |4.1.3.2 |x| | | | |
 +
- Applic layer can specify IP options in Send  |4.1.3.2 |x| | | | |
 +
- UDP passes IP options down to IP layer        |4.1.3.2 |x| | | | |
 +
                                              |        | | | | | |
 +
Pass ICMP msgs up to applic layer                |4.1.3.3 |x| | | | |
 +
                                              |        | | | | | |
 +
UDP checksums:                                  |        | | | | | |
 +
- Able to generate/check checksum              |4.1.3.4 |x| | | | |
 +
- Silently discard bad checksum                |4.1.3.4 |x| | | | |
 +
- Sender Option to not generate checksum        |4.1.3.4 | | |x| | |
 +
- Default is to checksum                      |4.1.3.4 |x| | | | |
 +
- Receiver Option to require checksum          |4.1.3.4 | | |x| | |
 +
                                              |        | | | | | |
 +
UDP Multihoming                                  |        | | | | | |
 +
- Pass spec-dest addr to application            |4.1.3.5 |x| | | | |
  
−
RFC1122                     INTERNET LAYER                 October 1989
+
RFC1122                 TRANSPORT LAYER -- UDP            October 1989
  
 +
- Applic layer can specify Local IP addr        |4.1.3.5 |x| | | | |
 +
- Applic layer specify wild Local IP addr      |4.1.3.5 |x| | | | |
 +
- Applic layer notified of Local IP addr used  |4.1.3.5 | |x| | | |
 +
                                              |        | | | | | |
 +
Bad IP src addr silently discarded by UDP/IP    |4.1.3.6 |x| | | | |
 +
Only send valid IP source address                |4.1.3.6 |x| | | | |
 +
UDP Application Interface Services              |        | | | | | |
 +
Full IP interface of 3.4 for application        |4.1.4  |x| | | | |
 +
- Able to spec TTL, TOS, IP opts when send dg  |4.1.4  |x| | | | |
 +
- Pass received TOS up to applic layer          |4.1.4  | | |x| | |
  
−
             mask it installs; see IMPLEMENTATION section below.
+
RFC1122                  TRANSPORT LAYER -- TCP             October 1989
  
−
            A system MUST NOT send an Address Mask Reply unless it is an
+
4.2 TRANSMISSION CONTROL PROTOCOL -- TCP
−
            authoritative agent for address masks.  An authoritative
 
−
            agent may be a host or a gateway, but it MUST be explicitly
 
−
            configured as a address mask agent.  Receiving an address
 
−
            mask via an Address Mask Reply does not give the receiver
 
−
            authority and MUST NOT be used as the basis for issuing
 
−
            Address Mask Replies.
 
  
−
            With a statically configured address mask, there SHOULD be
+
  4.2.1  INTRODUCTION
−
            an additional configuration flag that determines whether the
 
−
            host is to act as an authoritative agent for this mask,
 
−
            i.e., whether it will answer Address Mask Request messages
 
−
            using this mask.
 
  
−
            If it is configured as an agent, the host MUST broadcast an
+
      The Transmission Control Protocol TCP [TCP:1] is the primary
−
            Address Mask Reply for the mask on the appropriate interface
+
      virtual-circuit transport protocol for the Internet suite.  TCP
−
            when it initializes.
+
      provides reliable, in-sequence delivery of a full-duplex stream
 +
      of octets (8-bit bytes).  TCP is used by those applications
 +
      needing reliable, connection-oriented transport service, e.g.,
 +
      mail (SMTP), file transfer (FTP), and virtual terminal service
 +
      (Telnet); requirements for these application-layer protocols
 +
      are described in [INTRO:1].
  
−
            See "System Initialization" in [INTRO:1] for more
+
  4.2.2  PROTOCOL WALK-THROUGH
−
            information about the use of Address Mask Request/Reply
 
−
            messages.
 
  
−
            DISCUSSION
+
      4.2.2.1 Well-Known Ports: RFC-793 Section 2.7
−
                Hosts that casually send Address Mask Replies with
 
−
                invalid address masks have often been a serious
 
−
                nuisance.  To prevent this, Address Mask Replies ought
 
−
                to be sent only by authoritative agents that have been
 
−
                selected by explicit administrative action.
 
  
−
                When an authoritative agent receives an Address Mask
+
        DISCUSSION:
−
                Request message, it will send a unicast Address Mask
+
              TCP reserves port numbers in the range 0-255 for
−
                Reply to the source IP address.  If the network part of
+
              "well-known" ports, used to access services that are
−
                this address is zero (see (a) and (b) in 3.2.1.3), the
+
              standardized across the Internet.  The remainder of the
−
                Reply will be broadcast.
+
              port space can be freely allocated to application
 +
              processes.  Current well-known port definitions are
 +
              listed in the RFC entitled "Assigned Numbers"
 +
              [INTRO:6]. A prerequisite for defining a new well-
 +
              known port is an RFC documenting the proposed service
 +
              in enough detail to allow new implementations.
  
−
                Getting no reply to its Address Mask Request messages,
+
              Some systems extend this notion by adding a third
−
                a host will assume there is no agent and use an
+
              subdivision of the TCP port space: reserved ports,
−
                unsubnetted mask, but the agent may be only temporarily
+
              which are generally used for operating-system-specific
−
                unreachable.  An agent will broadcast an unsolicited
+
              services.  For example, reserved ports might fall
−
                Address Mask Reply whenever it initializes, in order to
+
              between 256 and some system-dependent upper limit.
−
                update the masks of all hosts that have initialized in
+
              Some systems further choose to protect well-known and
−
                the meantime.
+
              reserved ports by permitting only privileged users to
 +
              open TCP connections with those port values.  This is
 +
              perfectly reasonable as long as the host does not
 +
              assume that all hosts protect their low-numbered ports
 +
              in this manner.
  
−
            IMPLEMENTATION:
+
      4.2.2.2  Use of Push: RFC-793 Section 2.8
−
                The following reasonableness check on an address mask
 
−
                is suggested: the mask is not all 1 bits, and it is
 
  
 +
        When an application issues a series of SEND calls without
 +
        setting the PUSH flag, the TCP MAY aggregate the data
 +
        internally without sending it.  Similarly, when a series of
 +
        segments is received without the PSH bit, a TCP MAY queue
 +
        the data internally without passing it to the receiving
 +
        application.
  
 +
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
  
−
Internet Engineering Task Force                                [Page 46]
+
        The PSH bit is not a record marker and is independent of
 +
        segment boundaries.  The transmitter SHOULD collapse
 +
        successive PSH bits when it packetizes data, to send the
 +
        largest possible segment.
  
 +
        A TCP MAY implement PUSH flags on SEND calls.  If PUSH flags
 +
        are not implemented, then the sending TCP: (1) must not
 +
        buffer data indefinitely, and (2) MUST set the PSH bit in
 +
        the last buffered segment (i.e., when there is no more
 +
        queued data to be sent).
  
 +
        The discussion in RFC-793 on pages 48, 50, and 74
 +
        erroneously implies that a received PSH flag must be passed
 +
        to the application layer.  Passing a received PSH flag to
 +
        the application layer is now OPTIONAL.
  
 +
        An application program is logically required to set the PUSH
 +
        flag in a SEND call whenever it needs to force delivery of
 +
        the data to avoid a communication deadlock.  However, a TCP
 +
        SHOULD send a maximum-sized segment whenever possible, to
 +
        improve performance (see Section 4.2.3.4).
  
−
RFC1122                      INTERNET LAYER                October 1989
+
        DISCUSSION:
 +
              When the PUSH flag is not implemented on SEND calls,
 +
              i.e., when the application/TCP interface uses a pure
 +
              streaming model, responsibility for aggregating any
 +
              tiny data fragments to form reasonable sized segments
 +
              is partially borne by the application layer.
  
 +
              Generally, an interactive application protocol must set
 +
              the PUSH flag at least in the last SEND call in each
 +
              command or response sequence.  A bulk transfer protocol
 +
              like FTP should set the PUSH flag on the last segment
 +
              of a file or when necessary to prevent buffer deadlock.
  
−
                either zero or else the 8 highest-order bits are on.
+
              At the receiver, the PSH bit forces buffered data to be
 +
              delivered to the application (even if less than a full
 +
              buffer has been received). Conversely, the lack of a
 +
              PSH bit can be used to avoid unnecessary wakeup calls
 +
              to the application process; this can be an important
 +
              performance optimization for large timesharing hosts.
 +
              Passing the PSH bit to the receiving application allows
 +
              an analogous optimization within the application.
  
−
       3.2.3  Internet Group Management Protocol IGMP
+
       4.2.2.3  Window Size: RFC-793 Section 3.1
  
−
         IGMP [IP:4] is a protocol used between hosts and gateways on a
+
         The window size MUST be treated as an unsigned number, or
−
        single network to establish hosts' membership in particular
+
         else large window sizes will appear like negative windows
−
        multicast groups.  The gateways use this information, in
 
−
        conjunction with a multicast routing protocol, to support IP
 
−
         multicasting across the Internet.
 
  
−
        At this time, implementation of IGMP is OPTIONAL; see Section
+
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
−
        3.3.7 for more information.  Without IGMP, a host can still
 
−
        participate in multicasting local to its connected networks.
 
  
−
  3.3 SPECIFIC ISSUES
+
        and TCP will not work.  It is RECOMMENDED that
 +
        implementations reserve 32-bit fields for the send and
 +
        receive window sizes in the connection record and do all
 +
        window computations with 32 bits.
  
−
      3.3.1 Routing Outbound Datagrams
+
        DISCUSSION:
 +
              It is known that the window field in the TCP header is
 +
              too small for high-speed, long-delay paths.
 +
              Experimental TCP options have been defined to extend
 +
              the window size; see for example [TCP:11].  In
 +
              anticipation of the adoption of such an extension, TCP
 +
              implementors should treat windows as 32 bits.
  
−
        The IP layer chooses the correct next hop for each datagram it
+
      4.2.2.4 Urgent Pointer: RFC-793 Section 3.1
−
        sends.  If the destination is on a connected network, the
 
−
        datagram is sent directly to the destination host; otherwise,
 
−
        it has to be routed to a gateway on a connected network.
 
  
−
         3.3.1.1 Local/Remote Decision
+
         The second sentence is in error: the urgent pointer points
 +
        to the sequence number of the LAST octet (not LAST+1) in a
 +
        sequence of urgent data.  The description on page 56 (last
 +
        sentence) is correct.
  
−
            To decide if the destination is on a connected network, the
+
        A TCP MUST support a sequence of urgent data of any length.
−
            following algorithm MUST be used [see IP:3]:
 
  
−
            (a) The address mask (particular to a local IP address for
+
        A TCP MUST inform the application layer asynchronously
−
                a multihomed host) is a 32-bit mask that selects the
+
        whenever it receives an Urgent pointer and there was
−
                network number and subnet number fields of the
+
        previously no pending urgent data, or whenever the Urgent
−
                corresponding IP address.
+
        pointer advances in the data stream. There MUST be a way
 +
        for the application to learn how much urgent data remains to
 +
        be read from the connection, or at least to determine
 +
        whether or not more urgent data remains to be read.
  
−
            (b)  If the IP destination address bits extracted by the
+
        DISCUSSION:
−
                address mask match the IP source address bits extracted
+
              Although the Urgent mechanism may be used for any
−
                by the same mask, then the destination is on the
+
              application, it is normally used to send "interrupt"-
−
                corresponding connected network, and the datagram is to
+
              type commands to a Telnet program (see "Using Telnet
−
                be transmitted directly to the destination host.
+
              Synch Sequence" section in [INTRO:1]).
  
−
            (c)  If not, then the destination is accessible only through
+
              The asynchronous or "out-of-band" notification will
−
                a gateway.  Selection of a gateway is described below
+
              allow the application to go into "urgent mode", reading
−
                (3.3.1.2).
+
              data from the TCP connection.  This allows control
 +
              commands to be sent to an application whose normal
 +
              input buffers are full of unprocessed data.
  
−
            A special-case destination address is handled as follows:
+
        IMPLEMENTATION:
 +
              The generic ERROR-REPORT() upcall described in Section
 +
              4.2.4.1 is a possible mechanism for informing the
 +
              application of the arrival of urgent data.
  
−
             *    For a limited broadcast or a multicast address, simply
+
RFC1122                  TRANSPORT LAYER -- TCP             October 1989
−
                pass the datagram to the link layer for the appropriate
 
−
                interface.
 
  
 +
      4.2.2.5  TCP Options: RFC-793 Section 3.1
  
 +
        A TCP MUST be able to receive a TCP option in any segment.
 +
        A TCP MUST ignore without error any TCP option it does not
 +
        implement, assuming that the option has a length field (all
 +
        TCP options defined in the future will have length fields).
 +
        TCP MUST be prepared to handle an illegal option length
 +
        (e.g., zero) without crashing; a suggested procedure is to
 +
        reset the connection and log the reason.
  
−
Internet Engineering Task Force                                [Page 47]
+
      4.2.2.6  Maximum Segment Size Option: RFC-793 Section 3.1
  
 +
        TCP MUST implement both sending and receiving the Maximum
 +
        Segment Size option [TCP:4].
  
 +
        TCP SHOULD send an MSS (Maximum Segment Size) option in
 +
        every SYN segment when its receive MSS differs from the
 +
        default 536, and MAY send it always.
  
 +
        If an MSS option is not received at connection setup, TCP
 +
        MUST assume a default send MSS of 536 (576-40) [TCP:4].
  
−
RFC1122                      INTERNET LAYER                October 1989
+
        The maximum size of a segment that TCP really sends, the
 +
        "effective send MSS," MUST be the smaller of the send MSS
 +
        (which reflects the available reassembly buffer size at the
 +
        remote host) and the largest size permitted by the IP layer:
  
 +
            Eff.snd.MSS =
  
−
            *    For a (network or subnet) directed broadcast, the
+
              min(SendMSS+20, MMS_S) - TCPhdrsize - IPoptionsize
−
                datagram can use the standard routing algorithms.
 
  
−
            The host IP layer MUST operate correctly in a minimal
+
        where:
−
            network environment, and in particular, when there are no
 
−
            gateways.  For example, if the IP layer of a host insists on
 
−
            finding at least one gateway to initialize, the host will be
 
−
            unable to operate on a single isolated broadcast net.
 
  
−
         3.3.1.2  Gateway Selection
+
         *    SendMSS is the MSS value received from the remote host,
 +
              or the default 536 if no MSS option is received.
  
−
            To efficiently route a series of datagrams to the same
+
        *    MMS_S is the maximum size for a transport-layer message
−
            destination, the source host MUST keep a "route cache" of
+
              that TCP may send.
−
            mappings to next-hop gateways.  A host uses the following
 
−
            basic algorithm on this cache to route a datagram; this
 
−
            algorithm is designed to put the primary routing burden on
 
−
            the gateways [IP:11].
 
  
−
            (a)  If the route cache contains no information for a
+
        *    TCPhdrsize is the size of the TCP header; this is
−
                particular destination, the host chooses a "default"
+
              normally 20, but may be larger if TCP options are to be
−
                gateway and sends the datagram to it.  It also builds a
+
              sent.
−
                corresponding Route Cache entry.
 
  
−
            (b)  If that gateway is not the best next hop to the
+
        *    IPoptionsize is the size of any IP options that TCP
−
                destination, the gateway will forward the datagram to
+
              will pass to the IP layer with the current message.
−
                the best next-hop gateway and return an ICMP Redirect
 
−
                message to the source host.
 
  
−
            (c)  When it receives a Redirect, the host updates the
+
        The MSS value to be sent in an MSS option must be less than
−
                next-hop gateway in the appropriate route cache entry,
 
−
                so later datagrams to the same destination will go
 
−
                directly to the best gateway.
 
  
−
             Since the subnet mask appropriate to the destination address
+
RFC1122                  TRANSPORT LAYER -- TCP             October 1989
−
            is generally not known, a Network Redirect message SHOULD be
 
−
            treated identically to a Host Redirect message; i.e., the
 
−
            cache entry for the destination host (only) would be updated
 
−
            (or created, if an entry for that host did not exist) for
 
−
            the new gateway.
 
  
−
            DISCUSSION:
+
        or equal to:
−
                This recommendation is to protect against gateways that
 
−
                erroneously send Network Redirects for a subnetted
 
−
                network, in violation of the gateway requirements
 
−
                [INTRO:2].
 
  
−
             When there is no route cache entry for the destination host
+
             MMS_R - 20
−
            address (and the destination is not on the connected
 
  
 +
        where MMS_R is the maximum size for a transport-layer
 +
        message that can be received (and reassembled).  TCP obtains
 +
        MMS_R and MMS_S from the IP layer; see the generic call
 +
        GET_MAXSIZES in Section 3.4.
  
 +
        DISCUSSION:
 +
              The choice of TCP segment size has a strong effect on
 +
              performance.  Larger segments increase throughput by
 +
              amortizing header size and per-datagram processing
 +
              overhead over more data bytes; however, if the packet
 +
              is so large that it causes IP fragmentation, efficiency
 +
              drops sharply if any fragments are lost [IP:9].
  
−
Internet Engineering Task Force                                [Page 48]
+
              Some TCP implementations send an MSS option only if the
 +
              destination host is on a non-connected network.
 +
              However, in general the TCP layer may not have the
 +
              appropriate information to make this decision, so it is
 +
              preferable to leave to the IP layer the task of
 +
              determining a suitable MTU for the Internet path.  We
 +
              therefore recommend that TCP always send the option (if
 +
              not 536) and that the IP layer determine MMS_R as
 +
              specified in 3.3.3 and 3.4.  A proposed IP-layer
 +
              mechanism to measure the MTU would then modify the IP
 +
              layer without changing TCP.
  
 +
      4.2.2.7  TCP Checksum: RFC-793 Section 3.1
  
 +
        Unlike the UDP checksum (see Section 4.1.3.4), the TCP
 +
        checksum is never optional.  The sender MUST generate it and
 +
        the receiver MUST check it.
  
 +
      4.2.2.8  TCP Connection State Diagram: RFC-793 Section 3.2,
 +
        page 23
  
−
RFC1122                      INTERNET LAYER                October 1989
+
        There are several problems with this diagram:
  
 +
        (a)  The arrow from SYN-SENT to SYN-RCVD should be labeled
 +
              with "snd SYN,ACK", to agree with the text on page 68
 +
              and with Figure 8.
  
−
            network), the IP layer MUST pick a gateway from its list of
+
        (b) There could be an arrow from SYN-RCVD state to LISTEN
−
            "default" gateways.  The IP layer MUST support multiple
+
              state, conditioned on receiving a RST after a passive
−
            default gateways.
+
              open (see text page 70).
  
−
             As an extra feature, a host IP layer MAY implement a table
+
RFC1122                  TRANSPORT LAYER -- TCP             October 1989
−
            of "static routes".  Each such static route MAY include a
 
−
            flag specifying whether it may be overridden by ICMP
 
−
            Redirects.
 
  
−
            DISCUSSION:
+
        (c)  It is possible to go directly from FIN-WAIT-1 to the
−
                A host generally needs to know at least one default
+
              TIME-WAIT state (see page 75 of the spec).
−
                gateway to get started.  This information can be
 
−
                obtained from a configuration file or else from the
 
−
                host startup sequence, e.g., the BOOTP protocol (see
 
−
                [INTRO:1]).
 
  
−
                It has been suggested that a host can augment its list
+
      4.2.2.9 Initial Sequence Number Selection: RFC-793 Section
−
                of default gateways by recording any new gateways it
+
        3.3, page 27
−
                learns about. For example, it can record every gateway
 
−
                to which it is ever redirected.  Such a feature, while
 
−
                possibly useful in some circumstances, may cause
 
−
                problems in other cases (e.g., gateways are not all
 
−
                equal), and it is not recommended.
 
  
−
                A static route is typically a particular preset mapping
+
        A TCP MUST use the specified clock-driven selection of
−
                from destination host or network into a particular
+
        initial sequence numbers.
−
                next-hop gateway; it might also depend on the Type-of-
 
−
                Service (see next section).  Static routes would be set
 
−
                up by system administrators to override the normal
 
−
                automatic routing mechanism, to handle exceptional
 
−
                situations.  However, any static routing information is
 
−
                a potential source of failure as configurations change
 
−
                or equipment fails.
 
  
−
        3.3.1.3 Route Cache
+
      4.2.2.10  Simultaneous Open Attempts: RFC-793 Section 3.4, page
 +
        32
  
−
            Each route cache entry needs to include the following
+
        There is an error in Figure 8: the packet on line 7 should
−
            fields:
+
        be identical to the packet on line 5.
  
−
            (1)  Local IP address (for a multihomed host)
+
        A TCP MUST support simultaneous open attempts.
  
−
            (2)  Destination IP address
+
        DISCUSSION:
 +
              It sometimes surprises implementors that if two
 +
              applications attempt to simultaneously connect to each
 +
              other, only one connection is generated instead of two.
 +
              This was an intentional design decision; don't try to
 +
              "fix" it.
  
−
            (3) Type(s)-of-Service
+
      4.2.2.11 Recovery from Old Duplicate SYN: RFC-793 Section 3.4,
 +
        page 33
  
−
            (4)  Next-hop gateway IP address
+
        Note that a TCP implementation MUST keep track of whether a
 +
        connection has reached SYN_RCVD state as the result of a
 +
        passive OPEN or an active OPEN.
  
−
            Field (2) MAY be the full IP address of the destination
+
      4.2.2.12  RST Segment: RFC-793 Section 3.4
  
 +
        A TCP SHOULD allow a received RST segment to include data.
  
 +
        DISCUSSION
 +
              It has been suggested that a RST segment could contain
 +
              ASCII text that encoded and explained the cause of the
 +
              RST.  No standard has yet been established for such
 +
              data.
  
−
Internet Engineering Task Force                                [Page 49]
+
      4.2.2.13  Closing a Connection: RFC-793 Section 3.5
  
 +
        A TCP connection may terminate in two ways: (1) the normal
 +
        TCP close sequence using a FIN handshake, and (2) an "abort"
 +
        in which one or more RST segments are sent and the
 +
        connection state is immediately discarded.  If a TCP
  
 +
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
  
 +
        connection is closed by the remote site, the local
 +
        application MUST be informed whether it closed normally or
 +
        was aborted.
  
−
RFC1122                      INTERNET LAYER                October 1989
+
        The normal TCP close sequence delivers buffered data
 +
        reliably in both directions.  Since the two directions of a
 +
        TCP connection are closed independently, it is possible for
 +
        a connection to be "half closed," i.e., closed in only one
 +
        direction, and a host is permitted to continue sending data
 +
        in the open direction on a half-closed connection.
  
 +
        A host MAY implement a "half-duplex" TCP close sequence, so
 +
        that an application that has called CLOSE cannot continue to
 +
        read data from the connection.  If such a host issues a
 +
        CLOSE call while received data is still pending in TCP, or
 +
        if new data is received after CLOSE is called, its TCP
 +
        SHOULD send a RST to show that data was lost.
  
−
            host, or only the destination network number.  Field (3),
+
        When a connection is closed actively, it MUST linger in
−
            the TOS, SHOULD be included.
+
        TIME-WAIT state for a time 2xMSL (Maximum Segment Lifetime).
 +
        However, it MAY accept a new SYN from the remote TCP to
 +
        reopen the connection directly from TIME-WAIT state, if it:
  
−
            See Section 3.3.4.2 for a discussion of the implications of
+
        (1)  assigns its initial sequence number for the new
−
            multihoming for the lookup procedure in this cache.
+
              connection to be larger than the largest sequence
 +
              number it used on the previous connection incarnation,
 +
              and
  
−
            DISCUSSION:
+
        (2)  returns to TIME-WAIT state if the SYN turns out to be
−
                Including the Type-of-Service field in the route cache
+
              an old duplicate.
−
                and considering it in the host route algorithm will
 
−
                provide the necessary mechanism for the future when
 
−
                Type-of-Service routing is commonly used in the
 
−
                Internet.  See Section 3.2.1.6.
 
  
−
                Each route cache entry defines the endpoints of an
+
        DISCUSSION:
−
                Internet path.  Although the connecting path may change
+
              TCP's full-duplex data-preserving close is a feature
−
                dynamically in an arbitrary way, the transmission
+
              that is not included in the analogous ISO transport
−
                characteristics of the path tend to remain
+
              protocol TP4.
−
                approximately constant over a time period longer than a
 
−
                single typical host-host transport connection.
 
−
                Therefore, a route cache entry is a natural place to
 
−
                cache data on the properties of the path.  Examples of
 
−
                such properties might be the maximum unfragmented
 
−
                datagram size (see Section 3.3.3), or the average
 
−
                round-trip delay measured by a transport protocol.
 
−
                This data will generally be both gathered and used by a
 
−
                higher layer protocol, e.g., by TCP, or by an
 
−
                application using UDP.  Experiments are currently in
 
−
                progress on caching path properties in this manner.
 
  
−
                There is no consensus on whether the route cache should
+
              Some systems have not implemented half-closed
−
                be keyed on destination host addresses alone, or allow
+
              connections, presumably because they do not fit into
−
                both host and network addresses.  Those who favor the
+
              the I/O model of their particular operating system.  On
−
                use of only host addresses argue that:
+
              these systems, once an application has called CLOSE, it
 +
              can no longer read input data from the connection; this
 +
              is referred to as a "half-duplex" TCP close sequence.
  
−
                (1)  As required in Section 3.3.1.2, Redirect messages
+
              The graceful close algorithm of TCP requires that the
−
                      will generally result in entries keyed on
+
              connection state remain defined on (at least)  one end
−
                      destination host addresses; the simplest and most
+
              of the connection, for a timeout period of 2xMSL, i.e.,
−
                      general scheme would be to use host addresses
+
              4 minutes. During this period, the (remote socket,
−
                      always.
 
  
−
                 (2)  The IP layer may not always know the address mask
+
RFC1122                 TRANSPORT LAYER -- TCP            October 1989
−
                      for a network address in a complex subnetted
 
−
                      environment.
 
  
−
                (3) The use of only host addresses allows the
+
              local socket) pair that defines the connection is busy
−
                      destination address to be used as a pure 32-bit
+
              and cannot be reused.  To shorten the time that a given
−
                      number, which may allow the Internet architecture
+
              port pair is tied up, some TCPs allow a new SYN to be
−
                      to be more easily extended in the future without
+
              accepted in TIME-WAIT state.
  
 +
      4.2.2.14  Data Communication: RFC-793 Section 3.7, page 40
  
 +
        Since RFC-793 was written, there has been extensive work on
 +
        TCP algorithms to achieve efficient data communication.
 +
        Later sections of the present document describe required and
 +
        recommended TCP algorithms to determine when to send data
 +
        (Section 4.2.3.4), when to send an acknowledgment (Section
 +
        4.2.3.2), and when to update the window (Section 4.2.3.3).
  
−
Internet Engineering Task Force                                [Page 50]
+
        DISCUSSION:
 +
              One important performance issue is "Silly Window
 +
              Syndrome" or "SWS" [TCP:5], a stable pattern of small
 +
              incremental window movements resulting in extremely
 +
              poor TCP performance.  Algorithms to avoid SWS are
 +
              described below for both the sending side (Section
 +
              4.2.3.4) and the receiving side (Section 4.2.3.3).
  
 +
              In brief, SWS is caused by the receiver advancing the
 +
              right window edge whenever it has any new buffer space
 +
              available to receive data and by the sender using any
 +
              incremental window, no matter how small, to send more
 +
              data [TCP:5].  The result can be a stable pattern of
 +
              sending tiny data segments, even though both sender and
 +
              receiver have a large total buffer space for the
 +
              connection.  SWS can only occur during the transmission
 +
              of a large amount of data; if the connection goes
 +
              quiescent, the problem will disappear.  It is caused by
 +
              typical straightforward implementation of window
 +
              management, but the sender and receiver algorithms
 +
              given below will avoid it.
  
 +
              Another important TCP performance issue is that some
 +
              applications, especially remote login to character-at-
 +
              a-time hosts, tend to send streams of one-octet data
 +
              segments.  To avoid deadlocks, every TCP SEND call from
 +
              such applications must be "pushed", either explicitly
 +
              by the application or else implicitly by TCP.  The
 +
              result may be a stream of TCP segments that contain one
 +
              data octet each, which makes very inefficient use of
 +
              the Internet and contributes to Internet congestion.
 +
              The Nagle Algorithm described in Section 4.2.3.4
 +
              provides a simple and effective solution to this
 +
              problem.  It does have the effect of clumping
  
 +
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
  
−
RFC1122                      INTERNET LAYER                October 1989
+
              characters over Telnet connections; this may initially
 +
              surprise users accustomed to single-character echo, but
 +
              user acceptance has not been a problem.
  
 +
              Note that the Nagle algorithm and the send SWS
 +
              avoidance algorithm play complementary roles in
 +
              improving performance.  The Nagle algorithm discourages
 +
              sending tiny segments when the data to be sent
 +
              increases in small increments, while the SWS avoidance
 +
              algorithm discourages small segments resulting from the
 +
              right window edge advancing in small increments.
  
−
                      any change to the hosts.
+
              A careless implementation can send two or more
 +
              acknowledgment segments per data segment received.  For
 +
              example, suppose the receiver acknowledges every data
 +
              segment immediately.  When the application program
 +
              subsequently consumes the data and increases the
 +
              available receive buffer space again, the receiver may
 +
              send a second acknowledgment segment to update the
 +
              window at the sender.  The extreme case occurs with
 +
              single-character segments on TCP connections using the
 +
              Telnet protocol for remote login service.  Some
 +
              implementations have been observed in which each
 +
              incoming 1-character segment generates three return
 +
              segments: (1) the acknowledgment, (2) a one byte
 +
              increase in the window, and (3) the echoed character,
 +
              respectively.
  
−
                The opposing view is that allowing a mixture of
+
      4.2.2.15  Retransmission Timeout: RFC-793 Section 3.7, page 41
−
                destination hosts and networks in the route cache:
 
  
−
                (1)  Saves memory space.
+
        The algorithm suggested in RFC-793 for calculating the
 +
        retransmission timeout is now known to be inadequate; see
 +
        Section 4.2.3.1 below.
  
−
                (2)  Leads to a simpler data structure, easily
+
        Recent work by Jacobson [TCP:7] on Internet congestion and
−
                      combining the cache with the tables of default and
+
        TCP retransmission stability has produced a transmission
−
                      static routes (see below).
+
        algorithm combining "slow start" with "congestion
 +
        avoidance".  A TCP MUST implement this algorithm.
  
−
                (3)  Provides a more useful place to cache path
+
        If a retransmitted packet is identical to the original
−
                      properties, as discussed earlier.
+
        packet (which implies not only that the data boundaries have
 +
        not changed, but also that the window and acknowledgment
 +
        fields of the header have not changed), then the same IP
 +
        Identification field MAY be used (see Section 3.2.1.5).
  
 +
        IMPLEMENTATION:
 +
              Some TCP implementors have chosen to "packetize" the
 +
              data stream, i.e., to pick segment boundaries when
  
−
             IMPLEMENTATION:
+
RFC1122                  TRANSPORT LAYER -- TCP             October 1989
−
                The cache needs to be large enough to include entries
 
−
                for the maximum number of destination hosts that may be
 
−
                in use at one time.
 
  
−
                A route cache entry may also include control
+
              segments are originally sent and to queue these
−
                information used to choose an entry for replacement.
+
              segments in a "retransmission queue" until they are
−
                This might take the form of a "recently used" bit, a
+
              acknowledged.  Another design (which may be simpler) is
−
                use count, or a last-used timestamp, for example.  It
+
              to defer packetizing until each time data is
−
                is recommended that it include the time of last
+
              transmitted or retransmitted, so there will be no
−
                modification of the entry, for diagnostic purposes.
+
              segment retransmission queue.
  
−
                An implementation may wish to reduce the overhead of
+
              In an implementation with a segment retransmission
−
                scanning the route cache for every datagram to be
+
              queue, TCP performance may be enhanced by repacketizing
−
                transmitted.  This may be accomplished with a hash
+
              the segments awaiting acknowledgment when the first
−
                table to speed the lookup, or by giving a connection-
+
              retransmission timeout occurs.  That is, the
−
                oriented transport protocol a "hint" or temporary
+
              outstanding segments that fitted would be combined into
−
                handle on the appropriate cache entry, to be passed to
+
              one maximum-sized segment, with a new IP Identification
−
                the IP layer with each subsequent datagram.
+
              value.  The TCP would then retain this combined segment
 +
              in the retransmit queue until it was acknowledged.
 +
              However, if the first two segments in the
 +
              retransmission queue totalled more than one maximum-
 +
              sized segment, the TCP would retransmit only the first
 +
              segment using the original IP Identification field.
  
−
                Although we have described the route cache, the lists
+
      4.2.2.16  Managing the Window: RFC-793 Section 3.7, page 41
−
                of default gateways, and a table of static routes as
 
−
                conceptually distinct, in practice they may be combined
 
−
                into a single "routing table" data structure.
 
  
−
         3.3.1.4  Dead Gateway Detection
+
         A TCP receiver SHOULD NOT shrink the window, i.e., move the
 +
        right window edge to the left.  However, a sending TCP MUST
 +
        be robust against window shrinking, which may cause the
 +
        "useable window" (see Section 4.2.3.4) to become negative.
  
−
            The IP layer MUST be able to detect the failure of a "next-
+
        If this happens, the sender SHOULD NOT send new data, but
−
            hop" gateway that is listed in its route cache and to choose
+
        SHOULD retransmit normally the old unacknowledged data
−
            an alternate gateway (see Section 3.3.1.5).
+
        between SND.UNA and SND.UNA+SND.WND.  The sender MAY also
 +
        retransmit old data beyond SND.UNA+SND.WND, but SHOULD NOT
 +
        time out the connection if data beyond the right window edge
 +
        is not acknowledged.  If the window shrinks to zero, the TCP
 +
        MUST probe it in the standard way (see next Section).
  
−
            Dead gateway detection is covered in some detail in RFC-816
+
        DISCUSSION:
−
            [IP:11]. Experience to date has not produced a complete
+
              Many TCP implementations become confused if the window
 +
              shrinks from the right after data has been sent into a
 +
              larger window. Note that TCP has a heuristic to select
 +
              the latest window update despite possible datagram
 +
              reordering; as a result, it may ignore a window update
 +
              with a smaller window than previously offered if
 +
              neither the sequence number nor the acknowledgment
 +
              number is increased.
  
 +
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
  
 +
      4.2.2.17  Probing Zero Windows: RFC-793 Section 3.7, page 42
  
−
Internet Engineering Task Force                                [Page 51]
+
        Probing of zero (offered) windows MUST be supported.
  
 +
        A TCP MAY keep its offered receive window closed
 +
        indefinitely.  As long as the receiving TCP continues to
 +
        send acknowledgments in response to the probe segments, the
 +
        sending TCP MUST allow the connection to stay open.
  
 +
        DISCUSSION:
 +
              It is extremely important to remember that ACK
 +
              (acknowledgment) segments that contain no data are not
 +
              reliably transmitted by TCP.  If zero window probing is
 +
              not supported, a connection may hang forever when an
 +
              ACK segment that re-opens the window is lost.
  
 +
              The delay in opening a zero window generally occurs
 +
              when the receiving application stops taking data from
 +
              its TCP.  For example, consider a printer daemon
 +
              application, stopped because the printer ran out of
 +
              paper.
  
−
RFC1122                      INTERNET LAYER                October 1989
+
        The transmitting host SHOULD send the first zero-window
 +
        probe when a zero window has existed for the retransmission
 +
        timeout period (see Section 4.2.2.15), and SHOULD increase
 +
        exponentially the interval between successive probes.
  
 +
        DISCUSSION:
 +
              This procedure minimizes delay if the zero-window
 +
              condition is due to a lost ACK segment containing a
 +
              window-opening update.  Exponential backoff is
 +
              recommended, possibly with some maximum interval not
 +
              specified here.  This procedure is similar to that of
 +
              the retransmission algorithm, and it may be possible to
 +
              combine the two procedures in the implementation.
  
−
            algorithm which is totally satisfactory, though it has
+
      4.2.2.18  Passive OPEN Calls:  RFC-793 Section 3.8
−
            identified several forbidden paths and promising techniques.
 
  
−
            *    A particular gateway SHOULD NOT be used indefinitely in
+
        Every passive OPEN call either creates a new connection
−
                the absence of positive indications that it is
+
        record in LISTEN state, or it returns an error; it MUST NOT
−
                functioning.
+
        affect any previously created connection record.
  
−
            *    Active probes such as "pinging" (i.e., using an ICMP
+
        A TCP that supports multiple concurrent users MUST provide
−
                Echo Request/Reply exchange) are expensive and scale
+
        an OPEN call that will functionally allow an application to
−
                poorly.  In particular, hosts MUST NOT actively check
+
        LISTEN on a port while a connection block with the same
−
                the status of a first-hop gateway by simply pinging the
+
        local port is in SYN-SENT or SYN-RECEIVED state.
−
                gateway continuously.
 
  
−
            *    Even when it is the only effective way to verify a
+
        DISCUSSION:
−
                gateway's status, pinging MUST be used only when
 
−
                traffic is being sent to the gateway and when there is
 
−
                no other positive indication to suggest that the
 
−
                gateway is functioning.
 
  
−
             *    To avoid pinging, the layers above and/or below the
+
RFC1122                  TRANSPORT LAYER -- TCP             October 1989
−
                Internet layer SHOULD be able to give "advice" on the
 
−
                status of route cache entries when either positive
 
−
                (gateway OK) or negative (gateway dead) information is
 
−
                available.
 
  
 +
              Some applications (e.g., SMTP servers) may need to
 +
              handle multiple connection attempts at about the same
 +
              time.  The probability of a connection attempt failing
 +
              is reduced by giving the application some means of
 +
              listening for a new connection at the same time that an
 +
              earlier connection attempt is going through the three-
 +
              way handshake.
  
−
            DISCUSSION:
+
        IMPLEMENTATION:
−
                If an implementation does not include an adequate
+
              Acceptable implementations of concurrent opens may
−
                mechanism for detecting a dead gateway and re-routing,
+
              permit multiple passive OPEN calls, or they may allow
−
                a gateway failure may cause datagrams to apparently
+
              "cloning" of LISTEN-state connections from a single
−
                vanish into a "black hole".  This failure can be
+
              passive OPEN call.
−
                extremely confusing for users and difficult for network
 
−
                personnel to debug.
 
  
−
                The dead-gateway detection mechanism must not cause
+
      4.2.2.19 Time to Live: RFC-793 Section 3.9, page 52
−
                unacceptable load on the host, on connected networks,
 
−
                or on first-hop gateway(s).  The exact constraints on
 
−
                the timeliness of dead gateway detection and on
 
−
                acceptable load may vary somewhat depending on the
 
−
                nature of the host's mission, but a host generally
 
−
                needs to detect a failed first-hop gateway quickly
 
−
                enough that transport-layer connections will not break
 
−
                before an alternate gateway can be selected.
 
  
−
                Passing advice from other layers of the protocol stack
+
        RFC-793 specified that TCP was to request the IP layer to
−
                complicates the interfaces between the layers, but it
+
        send TCP segments with TTL = 60.  This is obsolete; the TTL
−
                is the preferred approach to dead gateway detection.
+
        value used to send TCP segments MUST be configurable. See
−
                Advice can come from almost any part of the IP/TCP
+
        Section 3.2.1.7 for discussion.
  
 +
      4.2.2.20  Event Processing: RFC-793 Section 3.9
  
 +
        While it is not strictly required, a TCP SHOULD be capable
 +
        of queueing out-of-order TCP segments.  Change the "may" in
 +
        the last sentence of the first paragraph on page 70 to
 +
        "should".
  
−
Internet Engineering Task Force                                [Page 52]
+
        DISCUSSION:
 +
              Some small-host implementations have omitted segment
 +
              queueing because of limited buffer space.  This
 +
              omission may be expected to adversely affect TCP
 +
              throughput, since loss of a single segment causes all
 +
              later segments to appear to be "out of sequence".
  
 +
        In general, the processing of received segments MUST be
 +
        implemented to aggregate ACK segments whenever possible.
 +
        For example, if the TCP is processing a series of queued
 +
        segments, it MUST process them all before sending any ACK
 +
        segments.
  
 +
        Here are some detailed error corrections and notes on the
 +
        Event Processing section of RFC-793.
  
 +
        (a)  CLOSE Call, CLOSE-WAIT state, p. 61: enter LAST-ACK
 +
              state, not CLOSING.
  
−
RFC1122                      INTERNET LAYER                October 1989
+
        (b)  LISTEN state, check for SYN (pp. 65, 66): With a SYN
  
 +
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
  
−
                architecture, but it is expected to come primarily from
+
              bit, if the security/compartment or the precedence is
−
                the transport and link layers.  Here are some possible
+
              wrong for the segment, a reset is sent.  The wrong form
−
                sources for gateway advice:
+
              of reset is shown in the text; it should be:
  
−
                o    TCP or any connection-oriented transport protocol
+
                <SEQ=0><ACK=SEG.SEQ+SEG.LEN><CTL=RST,ACK>
−
                      should be able to give negative advice, e.g.,
 
−
                      triggered by excessive retransmissions.
 
  
−
                o    TCP may give positive advice when (new) data is
+
        (c) SYN-SENT state, Check for SYN, p. 68: When the
−
                      acknowledged. Even though the route may be
+
              connection enters ESTABLISHED state, the following
−
                      asymmetric, an ACK for new data proves that the
+
              variables must be set:
−
                      acknowleged data must have been transmitted
+
                SND.WND <- SEG.WND
−
                      successfully.
+
                SND.WL1 <- SEG.SEQ
 +
                SND.WL2 <- SEG.ACK
  
−
                o    An ICMP Redirect message from a particular gateway
+
        (d)  Check security and precedence, p. 71: The first heading
−
                      should be used as positive advice about that
+
              "ESTABLISHED STATE" should really be a list of all
−
                      gateway.
+
              states other than SYN-RECEIVED: ESTABLISHED, FIN-WAIT-
 +
              1, FIN-WAIT-2, CLOSE-WAIT, CLOSING, LAST-ACK, and
 +
              TIME-WAIT.
  
−
                o    Link-layer information that reliably detects and
+
        (e)  Check SYN bit, p. 71:  "In SYN-RECEIVED state and if
−
                      reports host failures (e.g., ARPANET Destination
+
              the connection was initiated with a passive OPEN, then
−
                      Dead messages) should be used as negative advice.
+
              return this connection to the LISTEN state and return.
 +
              Otherwise...".
  
−
                o    Failure to ARP or to re-validate ARP mappings may
+
        (f)  Check ACK field, SYN-RECEIVED state, p. 72: When the
−
                      be used as negative advice for the corresponding
+
              connection enters ESTABLISHED state, the variables
−
                      IP address.
+
              listed in (c) must be set.
  
−
                o    Packets arriving from a particular link-layer
+
        (g)  Check ACK field, ESTABLISHED state, p. 72: The ACK is a
−
                      address are evidence that the system at this
+
              duplicate if SEG.ACK =< SND.UNA (the = was omitted).
−
                      address is alive. However, turning this
+
              Similarly, the window should be updated if: SND.UNA =<
−
                      information into advice about gateways requires
+
              SEG.ACK =< SND.NXT.
−
                      mapping the link-layer address into an IP address,
 
−
                      and then checking that IP address against the
 
−
                      gateways pointed to by the route cache. This is
 
−
                      probably prohibitively inefficient.
 
  
−
                Note that positive advice that is given for every
+
        (h)  USER TIMEOUT, p. 77:
−
                datagram received may cause unacceptable overhead in
 
−
                the implementation.
 
  
−
                While advice might be passed using required arguments
+
              It would be better to notify the application of the
−
                in all interfaces to the IP layer, some transport and
+
              timeout rather than letting TCP force the connection
−
                application layer protocols cannot deduce the correct
+
              closed.  However, see also Section 4.2.3.5.
−
                advice.  These interfaces must therefore allow a
 
−
                neutral value for advice, since either always-positive
 
−
                or always-negative advice leads to incorrect behavior.
 
  
−
                There is another technique for dead gateway detection
+
      4.2.2.21  Acknowledging Queued Segments: RFC-793 Section 3.9
−
                that has been commonly used but is not recommended.
 
  
 +
        A TCP MAY send an ACK segment acknowledging RCV.NXT when a
 +
        valid segment arrives that is in the window but not at the
 +
        left window edge.
  
 +
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
  
−
Internet Engineering Task Force                                [Page 53]
+
        DISCUSSION:
 +
              RFC-793 (see page 74) was ambiguous about whether or
 +
              not an ACK segment should be sent when an out-of-order
 +
              segment was received, i.e., when SEG.SEQ was unequal to
 +
              RCV.NXT.
  
 +
              One reason for ACKing out-of-order segments might be to
 +
              support an experimental algorithm known as "fast
 +
              retransmit".  With this algorithm, the sender uses the
 +
              "redundant" ACK's to deduce that a segment has been
 +
              lost before the retransmission timer has expired.  It
 +
              counts the number of times an ACK has been received
 +
              with the same value of SEG.ACK and with the same right
 +
              window edge.  If more than a threshold number of such
 +
              ACK's is received, then the segment containing the
 +
              octets starting at SEG.ACK is assumed to have been lost
 +
              and is retransmitted, without awaiting a timeout.  The
 +
              threshold is chosen to compensate for the maximum
 +
              likely segment reordering in the Internet.  There is
 +
              not yet enough experience with the fast retransmit
 +
              algorithm to determine how useful it is.
  
 +
  4.2.3  SPECIFIC ISSUES
  
 +
      4.2.3.1  Retransmission Timeout Calculation
  
−
RFC1122                      INTERNET LAYER                October 1989
+
        A host TCP MUST implement Karn's algorithm and Jacobson's
 +
        algorithm for computing the retransmission timeout ("RTO").
  
 +
        o    Jacobson's algorithm for computing the smoothed round-
 +
              trip ("RTT") time incorporates a simple measure of the
 +
              variance [TCP:7].
  
−
                This technique depends upon the host passively
+
        o    Karn's algorithm for selecting RTT measurements ensures
−
                receiving ("wiretapping") the Interior Gateway Protocol
+
              that ambiguous round-trip times will not corrupt the
−
                (IGP) datagrams that the gateways are broadcasting to
+
              calculation of the smoothed round-trip time [TCP:6].
−
                each other.  This approach has the drawback that a host
 
−
                needs to recognize all the interior gateway protocols
 
−
                that gateways may use (see [INTRO:2]).  In addition, it
 
−
                only works on a broadcast network.
 
  
−
                At present, pinging (i.e., using ICMP Echo messages) is
+
        This implementation also MUST include "exponential backoff"
−
                the mechanism for gateway probing when absolutely
+
        for successive RTO values for the same segment.
−
                required. A successful ping guarantees that the
+
        Retransmission of SYN segments SHOULD use the same algorithm
−
                addressed interface and its associated machine are up,
+
        as data segments.
−
                but it does not guarantee that the machine is a gateway
 
−
                as opposed to a host. The normal inference is that if
 
−
                a Redirect or other evidence indicates that a machine
 
−
                was a gateway, successful pings will indicate that the
 
−
                machine is still up and hence still a gateway.
 
−
                However, since a host silently discards packets that a
 
−
                gateway would forward or redirect, this assumption
 
−
                could sometimes fail.  To avoid this problem, a new
 
−
                ICMP message under development will ask "are you a
 
−
                gateway?"
 
  
−
            IMPLEMENTATION:
+
        DISCUSSION:
−
                The following specific algorithm has been suggested:
+
              There were two known problems with the RTO calculations
 +
              specified in RFC-793.  First, the accurate measurement
 +
              of RTTs is difficult when there are retransmissions.
 +
              Second, the algorithm to compute the smoothed round-
 +
              trip time is inadequate [TCP:7], because it incorrectly
  
−
                 o    Associate a "reroute timer" with each gateway
+
RFC1122                 TRANSPORT LAYER -- TCP            October 1989
−
                      pointed to by the route cache.  Initialize the
 
−
                      timer to a value Tr, which must be small enough to
 
−
                      allow detection of a dead gateway before transport
 
−
                      connections time out.
 
  
−
                o    Positive advice would reset the reroute timer to
+
              assumed that the variance in RTT values would be small
−
                      Tr.  Negative advice would reduce or zero the
+
              and constant.  These problems were solved by Karn's and
−
                      reroute timer.
+
              Jacobson's algorithm, respectively.
  
−
                o    Whenever the IP layer used a particular gateway to
+
              The performance increase resulting from the use of
−
                      route a datagram, it would check the corresponding
+
              these improvements varies from noticeable to dramatic.
−
                      reroute timer.  If the timer had expired (reached
+
              Jacobson's algorithm for incorporating the measured RTT
−
                      zero), the IP layer would send a ping to the
+
              variance is especially important on a low-speed link,
−
                      gateway, followed immediately by the datagram.
+
              where the natural variation of packet sizes causes a
 +
              large variation in RTT.  One vendor found link
 +
              utilization on a 9.6kb line went from 10% to 90% as a
 +
              result of implementing Jacobson's variance algorithm in
 +
              TCP.
  
−
                o    The ping (ICMP Echo) would be sent again if
+
        The following values SHOULD be used to initialize the
−
                      necessary, up to N times.  If no ping reply was
+
        estimation parameters for a new connection:
−
                      received in N tries, the gateway would be assumed
 
−
                      to have failed, and a new first-hop gateway would
 
−
                      be chosen for all cache entries pointing to the
 
−
                      failed gateway.
 
  
 +
        (a)  RTT = 0 seconds.
  
 +
        (b)  RTO = 3 seconds.  (The smoothed variance is to be
 +
              initialized to the value that will result in this RTO).
  
−
Internet Engineering Task Force                                [Page 54]
+
        The recommended upper and lower bounds on the RTO are known
 +
        to be inadequate on large internets.  The lower bound SHOULD
 +
        be measured in fractions of a second (to accommodate high
 +
        speed LANs) and the upper bound should be 2*MSL, i.e., 240
 +
        seconds.
  
 +
        DISCUSSION:
 +
              Experience has shown that these initialization values
 +
              are reasonable, and that in any case the Karn and
 +
              Jacobson algorithms make TCP behavior reasonably
 +
              insensitive to the initial parameter choices.
  
 +
      4.2.3.2  When to Send an ACK Segment
  
 +
        A host that is receiving a stream of TCP data segments can
 +
        increase efficiency in both the Internet and the hosts by
 +
        sending fewer than one ACK (acknowledgment) segment per data
 +
        segment received; this is known as a "delayed ACK" [TCP:5].
  
−
RFC1122                      INTERNET LAYER                October 1989
+
        A TCP SHOULD implement a delayed ACK, but an ACK should not
 +
        be excessively delayed; in particular, the delay MUST be
 +
        less than 0.5 seconds, and in a stream of full-sized
 +
        segments there SHOULD be an ACK for at least every second
 +
        segment.
  
 +
        DISCUSSION:
  
−
                 Note that the size of Tr is inversely related to the
+
RFC1122                 TRANSPORT LAYER -- TCP            October 1989
−
                amount of advice available.  Tr should be large enough
 
−
                to insure that:
 
  
−
                *    Any pinging will be at a low level (e.g., <10%) of
+
              A delayed ACK gives the application an opportunity to
−
                      all packets sent to a gateway from the host, AND
+
              update the window and perhaps to send an immediate
 +
              response. In particular, in the case of character-mode
 +
              remote login, a delayed ACK can reduce the number of
 +
              segments sent by the server by a factor of 3 (ACK,
 +
              window update, and echo character all combined in one
 +
              segment).
  
−
                *    pinging is infrequent (e.g., every 3 minutes)
+
              In addition, on some large multi-user hosts, a delayed
 +
              ACK can substantially reduce protocol processing
 +
              overhead by reducing the total number of packets to be
 +
              processed [TCP:5]. However, excessive delays on ACK's
 +
              can disturb the round-trip timing and packet "clocking"
 +
              algorithms [TCP:7].
  
−
                Since the recommended algorithm is concerned with the
+
      4.2.3.3  When to Send a Window Update
−
                gateways pointed to by route cache entries, rather than
 
−
                the cache entries themselves, a two level data
 
−
                structure (perhaps coordinated with ARP or similar
 
−
                caches) may be desirable for implementing a route
 
−
                cache.
 
  
−
         3.3.1.5  New Gateway Selection
+
         A TCP MUST include a SWS avoidance algorithm in the receiver
 +
        [TCP:5].
  
−
            If the failed gateway is not the current default, the IP
+
        IMPLEMENTATION:
−
            layer can immediately switch to a default gateway.  If it is
+
              The receiver's SWS avoidance algorithm determines when
−
            the current default that failed, the IP layer MUST select a
+
              the right window edge may be advanced; this is
−
            different default gateway (assuming more than one default is
+
              customarily known as "updating the window".  This
−
            known) for the failed route and for establishing new routes.
+
              algorithm combines with the delayed ACK algorithm (see
 +
              Section 4.2.3.2) to determine when an ACK segment
 +
              containing the current window will really be sent to
 +
              the receiver.  We use the notation of RFC-793; see
 +
              Figures 4 and 5 in that document.
  
−
            DISCUSSION:
+
              The solution to receiver SWS is to avoid advancing the
−
                When a gateway does fail, the other gateways on the
+
              right window edge RCV.NXT+RCV.WND in small increments,
−
                connected network will learn of the failure through
+
              even if data is received from the network in small
−
                some inter-gateway routing protocol.  However, this
+
              segments.
−
                will not happen instantaneously, since gateway routing
 
−
                protocols typically have a settling time of 30-60
 
−
                seconds.  If the host switches to an alternative
 
−
                gateway before the gateways have agreed on the failure,
 
−
                the new target gateway will probably forward the
 
−
                datagram to the failed gateway and send a Redirect back
 
−
                to the host pointing to the failed gateway (!).  The
 
−
                result is likely to be a rapid oscillation in the
 
−
                contents of the host's route cache during the gateway
 
−
                settling period. It has been proposed that the dead-
 
−
                gateway logic should include some hysteresis mechanism
 
−
                to prevent such oscillations. However, experience has
 
−
                not shown any harm from such oscillations, since
 
−
                service cannot be restored to the host until the
 
−
                gateways' routing information does settle down.
 
  
−
            IMPLEMENTATION:
+
              Suppose the total receive buffer space is RCV.BUFF.  At
−
                One implementation technique for choosing a new default
+
              any given moment, RCV.USER octets of this total may be
−
                gateway is to simply round-robin among the default
+
              tied up with data that has been received and
−
                gateways in the host's list.  Another is to rank the
+
              acknowledged but which the user process has not yet
 +
              consumed.  When the connection is quiescent, RCV.WND =
 +
              RCV.BUFF and RCV.USER = 0.
  
 +
              Keeping the right window edge fixed as data arrives and
 +
              is acknowledged requires that the receiver offer less
 +
              than its full buffer space, i.e., the receiver must
 +
              specify a RCV.WND that keeps RCV.NXT+RCV.WND constant
 +
              as RCV.NXT increases.  Thus, the total buffer space
 +
              RCV.BUFF is generally divided into three parts:
  
 +
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
  
−
Internet Engineering Task Force                                [Page 55]
+
              |<------- RCV.BUFF ---------------->|
 +
                  1            2            3
 +
          ----|---------|------------------|------|----
 +
                    RCV.NXT              ^
 +
                                        (Fixed)
  
 +
          1 - RCV.USER =  data received but not yet consumed;
 +
          2 - RCV.WND =  space advertised to sender;
 +
          3 - Reduction = space available but not yet
 +
                          advertised.
  
 +
              The suggested SWS avoidance algorithm for the receiver
 +
              is to keep RCV.NXT+RCV.WND fixed until the reduction
 +
              satisfies:
  
 +
                  RCV.BUFF - RCV.USER - RCV.WND  >=
  
−
RFC1122                      INTERNET LAYER                October 1989
+
                          min( Fr * RCV.BUFF, Eff.snd.MSS )
  
 +
              where Fr is a fraction whose recommended value is 1/2,
 +
              and Eff.snd.MSS is the effective send MSS for the
 +
              connection (see Section 4.2.2.6).  When the inequality
 +
              is satisfied, RCV.WND is set to RCV.BUFF-RCV.USER.
  
−
                gateways in priority order, and when the current
+
              Note that the general effect of this algorithm is to
−
                default gateway is not the highest priority one, to
+
              advance RCV.WND in increments of Eff.snd.MSS (for
−
                "ping" the higher-priority gateways slowly to detect
+
              realistic receive buffers:  Eff.snd.MSS < RCV.BUFF/2).
−
                when they return to service. This pinging can be at a
+
              Note also that the receiver must use its own
−
                very low rate, e.g., 0.005 per second.
+
              Eff.snd.MSS, assuming it is the same as the sender's.
  
−
        3.3.1.6 Initialization
+
      4.2.3.4 When to Send Data
  
−
            The following information MUST be configurable:
+
        A TCP MUST include a SWS avoidance algorithm in the sender.
  
−
            (1) IP address(es).
+
        A TCP SHOULD implement the Nagle Algorithm [TCP:9] to
 +
        coalesce short segments.  However, there MUST be a way for
 +
        an application to disable the Nagle algorithm on an
 +
        individual connection. In all cases, sending data is also
 +
        subject to the limitation imposed by the Slow Start
 +
        algorithm (Section 4.2.2.15).
  
−
            (2)  Address mask(s).
+
        DISCUSSION:
 +
              The Nagle algorithm is generally as follows:
  
−
            (3) A list of default gateways, with a preference level.
+
                  If there is unacknowledged data (i.e., SND.NXT >
 +
                  SND.UNA), then the sending TCP buffers all user
  
−
             A manual method of entering this configuration data MUST be
+
RFC1122                  TRANSPORT LAYER -- TCP             October 1989
−
            provided.  In addition, a variety of methods can be used to
 
−
            determine this information dynamically; see the section on
 
−
            "Host Initialization" in [INTRO:1].
 
  
−
            DISCUSSION:
+
                  data (regardless of the PSH bit), until the
−
                Some host implementations use "wiretapping" of gateway
+
                  outstanding data has been acknowledged or until
−
                protocols on a broadcast network to learn what gateways
+
                  the TCP can send a full-sized segment (Eff.snd.MSS
−
                exist. A standard method for default gateway discovery
+
                  bytes; see Section 4.2.2.6).
−
                is under development.
 
  
−
      3.3.2  Reassembly
+
              Some applications (e.g., real-time display window
 +
              updates) require that the Nagle algorithm be turned
 +
              off, so small data segments can be streamed out at the
 +
              maximum rate.
  
−
         The IP layer MUST implement reassembly of IP datagrams.
+
         IMPLEMENTATION:
 +
              The sender's SWS avoidance algorithm is more difficult
 +
              than the receivers's, because the sender does not know
 +
              (directly) the receiver's total buffer space RCV.BUFF.
 +
              An approach which has been found to work well is for
 +
              the sender to calculate Max(SND.WND), the maximum send
 +
              window it has seen so far on the connection, and to use
 +
              this value as an estimate of RCV.BUFF.  Unfortunately,
 +
              this can only be an estimate; the receiver may at any
 +
              time reduce the size of RCV.BUFF.  To avoid a resulting
 +
              deadlock, it is necessary to have a timeout to force
 +
              transmission of data, overriding the SWS avoidance
 +
              algorithm.  In practice, this timeout should seldom
 +
              occur.
  
−
        We designate the largest datagram size that can be reassembled
+
              The "useable window" [TCP:5] is:
−
        by EMTU_R ("Effective MTU to receive"); this is sometimes
 
−
        called the "reassembly buffer size".  EMTU_R MUST be greater
 
−
        than or equal to 576, SHOULD be either configurable or
 
−
        indefinite, and SHOULD be greater than or equal to the MTU of
 
−
        the connected network(s).
 
  
−
        DISCUSSION:
+
                  U = SND.UNA + SND.WND - SND.NXT
−
              A fixed EMTU_R limit should not be built into the code
 
−
              because some application layer protocols require EMTU_R
 
−
              values larger than 576.
 
  
−
        IMPLEMENTATION:
+
               i.e., the offered window less the amount of data sent
−
               An implementation may use a contiguous reassembly buffer
+
               but not acknowledged.  If D is the amount of data
−
              for each datagram, or it may use a more complex data
+
               queued in the sending TCP but not yet sent, then the
−
               structure that places no definite limit on the reassembled
+
              following set of rules is recommended.
−
               datagram size; in the latter case, EMTU_R is said to be
 
  
 +
              Send data:
  
 +
              (1)  if a maximum-sized segment can be sent, i.e, if:
  
−
Internet Engineering Task Force                                [Page 56]
+
                        min(D,U) >= Eff.snd.MSS;
  
 +
              (2)  or if the data is pushed and all queued data can
 +
                  be sent now, i.e., if:
  
 +
                      [SND.NXT = SND.UNA and] PUSHED and D <= U
  
 +
                  (the bracketed condition is imposed by the Nagle
 +
                  algorithm);
  
−
RFC1122                     INTERNET LAYER                 October 1989
+
RFC1122                 TRANSPORT LAYER -- TCP            October 1989
  
 +
              (3)  or if at least a fraction Fs of the maximum window
 +
                  can be sent, i.e., if:
  
−
              "indefinite".
+
                      [SND.NXT = SND.UNA and]
  
−
              Logically, reassembly is performed by simply copying each
+
                              min(D.U) >= Fs * Max(SND.WND);
−
              fragment into the packet buffer at the proper offset.
 
−
              Note that fragments may overlap if successive
 
−
              retransmissions use different packetizing but the same
 
−
              reassembly Id.
 
  
−
               The tricky part of reassembly is the bookkeeping to
+
               (4) or if data is PUSHed and the override timeout
−
              determine when all bytes of the datagram have been
+
                  occurs.
−
              reassembled. We recommend Clark's algorithm [IP:10] that
 
−
              requires no additional data space for the bookkeeping.
 
−
              However, note that, contrary to [IP:10], the first
 
−
              fragment header needs to be saved for inclusion in a
 
−
              possible ICMP Time Exceeded (Reassembly Timeout) message.
 
  
−
        There MUST be a mechanism by which the transport layer can
+
              Here Fs is a fraction whose recommended value is 1/2.
−
        learn MMS_R, the maximum message size that can be received and
+
              The override timeout should be in the range 0.1 - 1.0
−
        reassembled in an IP datagram (see GET_MAXSIZES calls in
+
              seconds.  It may be convenient to combine this timer
−
        Section 3.4). If EMTU_R is not indefinite, then the value of
+
              with the timer used to probe zero windows (Section
−
        MMS_R is given by:
+
              4.2.2.17).
  
−
            MMS_R = EMTU_R - 20
+
              Finally, note that the SWS avoidance algorithm just
 +
              specified is to be used instead of the sender-side
 +
              algorithm contained in [TCP:5].
  
−
        since 20 is the minimum size of an IP header.
+
      4.2.3.5  TCP Connection Failures
  
−
         There MUST be a reassembly timeout.  The reassembly timeout
+
         Excessive retransmission of the same segment by TCP
−
         value SHOULD be a fixed value, not set from the remaining TTL.
+
         indicates some failure of the remote host or the Internet
−
        It is recommended that the value lie between 60 seconds and 120
+
         path.  This failure may be of short or long duration.  The
−
         seconds.  If this timeout expires, the partially-reassembled
+
         following procedure MUST be used to handle excessive
−
         datagram MUST be discarded and an ICMP Time Exceeded message
+
         retransmissions of data segments [IP:11]:
−
         sent to the source host (if fragment zero has been received).
 
  
−
         DISCUSSION:
+
         (a)  There are two thresholds R1 and R2 measuring the amount
−
               The IP specification says that the reassembly timeout
+
               of retransmission that has occurred for the same
−
               should be the remaining TTL from the IP header, but this
+
               segment.  R1 and R2 might be measured in time units or
−
               does not work well because gateways generally treat TTL as
+
               as a count of retransmissions.
−
              a simple hop count rather than an elapsed time.  If the
 
−
              reassembly timeout is too small, datagrams will be
 
−
              discarded unnecessarily, and communication may fail.  The
 
−
              timeout needs to be at least as large as the typical
 
−
              maximum delay across the Internet.  A realistic minimum
 
−
              reassembly timeout would be 60 seconds.
 
  
−
              It has been suggested that a cache might be kept of
+
        (b)  When the number of transmissions of the same segment
−
               round-trip times measured by transport protocols for
+
               reaches or exceeds threshold R1, pass negative advice
−
               various destinations, and that these values might be used
+
               (see Section 3.3.1.4) to the IP layer, to trigger
−
               to dynamically determine a reasonable reassembly timeout
+
               dead-gateway diagnosis.
  
 +
        (c)  When the number of transmissions of the same segment
 +
              reaches a threshold R2 greater than R1, close the
 +
              connection.
  
 +
        (d)  An application MUST be able to set the value for R2 for
 +
              a particular connection.  For example, an interactive
 +
              application might set R2 to "infinity," giving the user
 +
              control over when to disconnect.
  
−
Internet Engineering Task Force                                [Page 57]
+
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
  
 +
        (d)  TCP SHOULD inform the application of the delivery
 +
              problem (unless such information has been disabled by
 +
              the application; see Section 4.2.4.1), when R1 is
 +
              reached and before R2.  This will allow a remote login
 +
              (User Telnet) application program to inform the user,
 +
              for example.
  
 +
        The value of R1 SHOULD correspond to at least 3
 +
        retransmissions, at the current RTO.  The value of R2 SHOULD
 +
        correspond to at least 100 seconds.
  
 +
        An attempt to open a TCP connection could fail with
 +
        excessive retransmissions of the SYN segment or by receipt
 +
        of a RST segment or an ICMP Port Unreachable.  SYN
 +
        retransmissions MUST be handled in the general way just
 +
        described for data retransmissions, including notification
 +
        of the application layer.
  
−
RFC1122                      INTERNET LAYER                October 1989
+
        However, the values of R1 and R2 may be different for SYN
 +
        and data segments.  In particular, R2 for a SYN segment MUST
 +
        be set large enough to provide retransmission of the segment
 +
        for at least 3 minutes.  The application can close the
 +
        connection (i.e., give up on the open attempt) sooner, of
 +
        course.
  
 +
        DISCUSSION:
 +
              Some Internet paths have significant setup times, and
 +
              the number of such paths is likely to increase in the
 +
              future.
  
−
              value.  Further investigation of this approach is
+
      4.2.3.6 TCP Keep-Alives
−
              required.
 
  
−
              If the reassembly timeout is set too high, buffer
+
        Implementors MAY include "keep-alives" in their TCP
−
              resources in the receiving host will be tied up too long,
+
        implementations, although this practice is not universally
−
              and the MSL (Maximum Segment Lifetime) [TCP:1] will be
+
        accepted.  If keep-alives are included, the application MUST
−
              larger than necessary.  The MSL controls the maximum rate
+
        be able to turn them on or off for each TCP connection, and
−
              at which fragmented datagrams can be sent using distinct
+
        they MUST default to off.
−
              values of the 16-bit Ident field; a larger MSL lowers the
 
−
              maximum rate.  The TCP specification [TCP:1] arbitrarily
 
−
              assumes a value of 2 minutes for MSL.  This sets an upper
 
−
              limit on a reasonable reassembly timeout value.
 
  
−
      3.3.3  Fragmentation
+
        Keep-alive packets MUST only be sent when no data or
 +
        acknowledgement packets have been received for the
 +
        connection within an interval. This interval MUST be
 +
        configurable and MUST default to no less than two hours.
  
−
         Optionally, the IP layer MAY implement a mechanism to fragment
+
         It is extremely important to remember that ACK segments that
−
         outgoing datagrams intentionally.
+
        contain no data are not reliably transmitted by TCP.
 +
        Consequently, if a keep-alive mechanism is implemented it
 +
        MUST NOT interpret failure to respond to any specific probe
 +
         as a dead connection.
  
−
        We designate by EMTU_S ("Effective MTU for sending") the
+
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
−
        maximum IP datagram size that may be sent, for a particular
 
−
        combination of IP source and destination addresses and perhaps
 
−
        TOS.
 
  
−
         A host MUST implement a mechanism to allow the transport layer
+
         An implementation SHOULD send a keep-alive segment with no
−
         to learn MMS_S, the maximum transport-layer message size that
+
         data; however, it MAY be configurable to send a keep-alive
−
         may be sent for a given {source, destination, TOS} triplet (see
+
         segment containing one garbage octet, for compatibility with
−
         GET_MAXSIZES call in Section 3.4).  If no local fragmentation
+
         erroneous TCP implementations.
−
        is performed, the value of MMS_S will be:
 
  
−
            MMS_S = EMTU_S - <IP header size>
+
        DISCUSSION:
 +
              A "keep-alive" mechanism periodically probes the other
 +
              end of a connection when the connection is otherwise
 +
              idle, even when there is no data to be sent.  The TCP
 +
              specification does not include a keep-alive mechanism
 +
              because it could:  (1) cause perfectly good connections
 +
              to break during transient Internet failures; (2)
 +
              consume unnecessary bandwidth ("if no one is using the
 +
              connection, who cares if it is still good?"); and (3)
 +
              cost money for an Internet path that charges for
 +
              packets.
  
−
        and EMTU_S must be less than or equal to the MTU of the network
+
              Some TCP implementations, however, have included a
−
        interface corresponding to the source address of the datagram.
+
              keep-alive mechanism.  To confirm that an idle
−
        Note that <IP header size> in this equation will be 20, unless
+
              connection is still active, these implementations send
−
        the IP reserves space to insert IP options for its own purposes
+
              a probe segment designed to elicit a response from the
−
        in addition to any options inserted by the transport layer.
+
              peer TCP.  Such a segment generally contains SEG.SEQ =
 +
              SND.NXT-1 and may or may not contain one garbage octet
 +
              of data.  Note that on a quiet connection SND.NXT =
 +
              RCV.NXT, so that this SEG.SEQ will be outside the
 +
              window.  Therefore, the probe causes the receiver to
 +
              return an acknowledgment segment, confirming that the
 +
              connection is still live.  If the peer has dropped the
 +
              connection due to a network partition or a crash, it
 +
              will respond with a RST instead of an acknowledgment
 +
              segment.
  
−
        A host that does not implement local fragmentation MUST ensure
+
              Unfortunately, some misbehaved TCP implementations fail
−
        that the transport layer (for TCP) or the application layer
+
              to respond to a segment with SEG.SEQ = SND.NXT-1 unless
−
        (for UDP) obtains MMS_S from the IP layer and does not send a
+
              the segment contains data.  Alternatively, an
−
        datagram exceeding MMS_S in size.
+
              implementation could determine whether a peer responded
 +
              correctly to keep-alive packets with no garbage data
 +
              octet.
  
−
        It is generally desirable to avoid local fragmentation and to
+
              A TCP keep-alive mechanism should only be invoked in
−
        choose EMTU_S low enough to avoid fragmentation in any gateway
+
              server applications that might otherwise hang
−
        along the path.  In the absence of actual knowledge of the
+
              indefinitely and consume resources unnecessarily if a
−
        minimum MTU along the path, the IP layer SHOULD use
+
              client crashes or aborts a connection during a network
−
        EMTU_S <= 576 whenever the destination address is not on a
+
              failure.
−
        connected network, and otherwise use the connected network's
 
  
 +
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
  
 +
      4.2.3.7  TCP Multihoming
  
−
Internet Engineering Task Force                                [Page 58]
+
        If an application on a multihomed host does not specify the
 +
        local IP address when actively opening a TCP connection,
 +
        then the TCP MUST ask the IP layer to select a local IP
 +
        address before sending the (first) SYN.  See the function
 +
        GET_SRCADDR() in Section 3.4.
  
 +
        At all other times, a previous segment has either been sent
 +
        or received on this connection, and TCP MUST use the same
 +
        local address is used that was used in those previous
 +
        segments.
  
 +
      4.2.3.8  IP Options
  
 +
        When received options are passed up to TCP from the IP
 +
        layer, TCP MUST ignore options that it does not understand.
  
−
RFC1122                      INTERNET LAYER                October 1989
+
        A TCP MAY support the Time Stamp and Record Route options.
  
 +
        An application MUST be able to specify a source route when
 +
        it actively opens a TCP connection, and this MUST take
 +
        precedence over a source route received in a datagram.
  
−
         MTU.
+
         When a TCP connection is OPENed passively and a packet
 +
        arrives with a completed IP Source Route option (containing
 +
        a return route), TCP MUST save the return route and use it
 +
        for all segments sent on this connection.  If a different
 +
        source route arrives in a later segment, the later
 +
        definition SHOULD override the earlier one.
  
−
        The MTU of each physical interface MUST be configurable.
+
      4.2.3.9  ICMP Messages
  
−
         A host IP layer implementation MAY have a configuration flag
+
         TCP MUST act on an ICMP error message passed up from the IP
−
         "All-Subnets-MTU", indicating that the MTU of the connected
+
         layer, directing it to the connection that created the
−
        network is to be used for destinations on different subnets
+
         error.  The necessary demultiplexing information can be
−
         within the same network, but not for other networks.  Thus,
+
         found in the IP header contained within the ICMP message.
−
         this flag causes the network class mask, rather than the subnet
 
−
        address mask, to be used to choose an EMTU_S.  For a multihomed
 
−
        host, an "All-Subnets-MTU" flag is needed for each network
 
−
        interface.
 
  
−
         DISCUSSION:
+
         o    Source Quench
−
              Picking the correct datagram size to use when sending data
 
−
              is a complex topic [IP:9].
 
  
−
               (a)  In general, no host is required to accept an IP
+
               TCP MUST react to a Source Quench by slowing
−
                  datagram larger than 576 bytes (including header and
+
              transmission on the connection.  The RECOMMENDED
−
                  data), so a host must not send a larger datagram
+
              procedure is for a Source Quench to trigger a "slow
−
                  without explicit knowledge or prior arrangement with
+
              start," as if a retransmission timeout had occurred.
−
                  the destination host.  Thus, MMS_S is only an upper
 
−
                  bound on the datagram size that a transport protocol
 
−
                  may send; even when MMS_S exceeds 556, the transport
 
−
                  layer must limit its messages to 556 bytes in the
 
−
                  absence of other knowledge about the destination
 
−
                  host.
 
  
−
              (b)  Some transport protocols (e.g., TCP) provide a way to
+
        o    Destination Unreachable -- codes 0, 1, 5
−
                  explicitly inform the sender about the largest
 
−
                  datagram the other end can receive and reassemble
 
−
                  [IP:7].  There is no corresponding mechanism in the
 
−
                  IP layer.
 
  
−
                  A transport protocol that assumes an EMTU_R larger
+
              Since these Unreachable messages indicate soft error
−
                  than 576 (see Section 3.3.2), can send a datagram of
 
−
                  this larger size to another host that implements the
 
−
                  same protocol.
 
  
−
              (c)  Hosts should ideally limit their EMTU_S for a given
+
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
−
                  destination to the minimum MTU of all the networks
 
−
                  along the path, to avoid any fragmentation.  IP
 
−
                  fragmentation, while formally correct, can create a
 
−
                  serious transport protocol performance problem,
 
−
                  because loss of a single fragment means all the
 
−
                  fragments in the segment must be retransmitted
 
−
                  [IP:9].
 
  
 +
              conditions, TCP MUST NOT abort the connection, and it
 +
              SHOULD make the information available to the
 +
              application.
  
 +
              DISCUSSION:
 +
                  TCP could report the soft error condition directly
 +
                  to the application layer with an upcall to the
 +
                  ERROR_REPORT routine, or it could merely note the
 +
                  message and report it to the application only when
 +
                  and if the TCP connection times out.
  
 +
        o    Destination Unreachable -- codes 2-4
  
−
Internet Engineering Task Force                                [Page 59]
+
              These are hard error conditions, so TCP SHOULD abort
 +
              the connection.
  
 +
        o    Time Exceeded -- codes 0, 1
  
 +
              This should be handled the same way as Destination
 +
              Unreachable codes 0, 1, 5 (see above).
  
 +
        o    Parameter Problem
  
−
RFC1122                      INTERNET LAYER                October 1989
+
              This should be handled the same way as Destination
 +
              Unreachable codes 0, 1, 5 (see above).
  
 +
      4.2.3.10  Remote Address Validation
  
−
              Since nearly all networks in the Internet currently
+
        A TCP implementation MUST reject as an error a local OPEN
−
              support an MTU of 576 or greater, we strongly recommend
+
        call for an invalid remote IP address (e.g., a broadcast or
−
              the use of 576 for datagrams sent to non-local networks.
+
        multicast address).
  
−
              It has been suggested that a host could determine the MTU
+
        An incoming SYN with an invalid source address must be
−
              over a given path by sending a zero-offset datagram
+
        ignored either by TCP or by the IP layer (see Section
−
              fragment and waiting for the receiver to time out the
+
        3.2.1.3).
−
              reassembly (which cannot complete!) and return an ICMP
 
−
              Time Exceeded message. This message would include the
 
−
              largest remaining fragment header in its body.  More
 
−
              direct mechanisms are being experimented with, but have
 
−
              not yet been adopted (see e.g., RFC-1063).
 
  
−
      3.3.4  Local Multihoming
+
        A TCP implementation MUST silently discard an incoming SYN
 +
        segment that is addressed to a broadcast or multicast
 +
        address.
  
−
        3.3.4.1 Introduction
+
      4.2.3.11 TCP Traffic Patterns
  
−
            A multihomed host has multiple IP addresses, which we may
+
        IMPLEMENTATION:
−
            think of as "logical interfaces".  These logical interfaces
+
              The TCP protocol specification [TCP:1] gives the
−
            may be associated with one or more physical interfaces, and
+
              implementor much freedom in designing the algorithms
−
            these physical interfaces may be connected to the same or
+
              that control the message flow over the connection --
−
            different networks.
+
              packetizing, managing the window, sending
  
−
             Here are some important cases of multihoming:
+
RFC1122                  TRANSPORT LAYER -- TCP             October 1989
  
−
            (a) Multiple Logical Networks
+
              acknowledgments, etc.  These design decisions are
 +
              difficult because a TCP must adapt to a wide range of
 +
              traffic patterns. Experience has shown that a TCP
 +
              implementor needs to verify the design on two extreme
 +
              traffic patterns:
  
−
                The Internet architects envisioned that each physical
+
              o    Single-character Segments
−
                network would have a single unique IP network (or
 
−
                subnet) number.  However, LAN administrators have
 
−
                sometimes found it useful to violate this assumption,
 
−
                operating a LAN with multiple logical networks per
 
−
                physical connected network.
 
  
−
                If a host connected to such a physical network is
+
                  Even if the sender is using the Nagle Algorithm,
−
                configured to handle traffic for each of N different
+
                  when a TCP connection carries remote login traffic
−
                logical networks, then the host will have N logical
+
                  across a low-delay LAN the receiver will generally
−
                interfaces.  These could share a single physical
+
                  get a stream of single-character segments.  If
−
                interface, or might use N physical interfaces to the
+
                  remote terminal echo mode is in effect, the
−
                same network.
+
                  receiver's system will generally echo each
 +
                  character as it is received.
  
−
            (b)  Multiple Logical Hosts
+
              o    Bulk Transfer
  
−
                When a host has multiple IP addresses that all have the
+
                  When TCP is used for bulk transfer, the data
−
                same <Network-number> part (and the same <Subnet-
+
                  stream should be made up (almost) entirely of
−
                number> part, if any), the logical interfaces are known
+
                  segments of the size of the effective MSS.
−
                as "logical hosts". These logical interfaces might
+
                  Although TCP uses a sequence number space with
−
                share a single physical interface or might use separate
+
                  byte (octet) granularity, in bulk-transfer mode
 +
                  its operation should be as if TCP used a sequence
 +
                  space that counted only segments.
  
 +
              Experience has furthermore shown that a single TCP can
 +
              effectively and efficiently handle these two extremes.
  
 +
              The most important tool for verifying a new TCP
 +
              implementation is a packet trace program.  There is a
 +
              large volume of experience showing the importance of
 +
              tracing a variety of traffic patterns with other TCP
 +
              implementations and studying the results carefully.
  
−
Internet Engineering Task Force                                [Page 60]
+
      4.2.3.12  Efficiency
  
 +
        IMPLEMENTATION:
 +
              Extensive experience has led to the following
 +
              suggestions for efficient implementation of TCP:
  
 +
              (a)  Don't Copy Data
  
 +
                  In bulk data transfer, the primary CPU-intensive
 +
                  tasks are copying data from one place to another
 +
                  and checksumming the data.  It is vital to
 +
                  minimize the number of copies of TCP data.  Since
  
−
RFC1122                     INTERNET LAYER                 October 1989
+
RFC1122                 TRANSPORT LAYER -- TCP            October 1989
  
 +
                  the ultimate speed limitation may be fetching data
 +
                  across the memory bus, it may be useful to combine
 +
                  the copy with checksumming, doing both with a
 +
                  single memory fetch.
  
−
                physical interfaces to the same physical network.
+
              (b)  Hand-Craft the Checksum Routine
  
−
            (c) Simple Multihoming
+
                  A good TCP checksumming routine is typically two
 +
                  to five times faster than a simple and direct
 +
                  implementation of the definition. Great care and
 +
                  clever coding are often required and advisable to
 +
                  make the checksumming code "blazing fast".  See
 +
                  [TCP:10].
  
−
                In this case, each logical interface is mapped into a
+
              (c) Code for the Common Case
−
                separate physical interface and each physical interface
 
−
                is connected to a different physical network. The term
 
−
                "multihoming" was originally applied only to this case,
 
−
                but it is now applied more generally.
 
  
−
                A host with embedded gateway functionality will
+
                  TCP protocol processing can be complicated, but
−
                typically fall into the simple multihoming case.  Note,
+
                  for most segments there are only a few simple
−
                however, that a host may be simply multihomed without
+
                  decisions to be made.  Per-segment processing will
−
                containing an embedded gateway, i.e., without
+
                  be greatly speeded up by coding the main line to
−
                forwarding datagrams from one connected network to
+
                  minimize the number of decisions in the most
−
                another.
+
                  common case.
  
−
                This case presents the most difficult routing problems.
+
  4.2.4  TCP/APPLICATION LAYER INTERFACE
−
                The choice of interface (i.e., the choice of first-hop
 
−
                network) may significantly affect performance or even
 
−
                reachability of remote parts of the Internet.
 
  
 +
      4.2.4.1  Asynchronous Reports
  
−
            Finally, we note another possibility that is NOT
+
        There MUST be a mechanism for reporting soft TCP error
−
            multihoming:  one logical interface may be bound to multiple
+
        conditions to the application.  Generically, we assume this
−
            physical interfaces, in order to increase the reliability or
+
        takes the form of an application-supplied ERROR_REPORT
−
            throughput between directly connected machines by providing
+
        routine that may be upcalled [INTRO:7] asynchronously from
−
            alternative physical paths between them.  For instance, two
+
        the transport layer:
−
            systems might be connected by multiple point-to-point links.
 
−
            We call this "link-layer multiplexing".  With link-layer
 
−
            multiplexing, the protocols above the link layer are unaware
 
−
            that multiple physical interfaces are present; the link-
 
−
            layer device driver is responsible for multiplexing and
 
−
            routing packets across the physical interfaces.
 
  
−
             In the Internet protocol architecture, a transport protocol
+
             ERROR_REPORT(local connection name, reason, subreason)
−
            instance ("entity") has no address of its own, but instead
 
−
            uses a single Internet Protocol (IP) address.  This has
 
−
            implications for the IP, transport, and application layers,
 
−
            and for the interfaces between them.  In particular, the
 
−
            application software may have to be aware of the multiple IP
 
−
            addresses of a multihomed host; in other cases, the choice
 
−
            can be made within the network software.
 
  
−
         3.3.4.2 Multihoming Requirements
+
         The precise encoding of the reason and subreason parameters
 +
        is not specified here.  However, the conditions that are
 +
        reported asynchronously to the application MUST include:
  
−
            The following general rules apply to the selection of an IP
+
        *    ICMP error message arrived (see 4.2.3.9)
−
            source address for sending a datagram from a multihomed
 
  
 +
        *    Excessive retransmissions (see 4.2.3.5)
  
 +
        *    Urgent pointer advance (see 4.2.2.4).
  
−
Internet Engineering Task Force                                [Page 61]
+
        However, an application program that does not want to
 +
        receive such ERROR_REPORT calls SHOULD be able to
  
 +
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
  
 +
        effectively disable these calls.
  
 +
        DISCUSSION:
 +
              These error reports generally reflect soft errors that
 +
              can be ignored without harm by many applications.  It
 +
              has been suggested that these error report calls should
 +
              default to "disabled," but this is not required.
  
−
RFC1122                      INTERNET LAYER                October 1989
+
       4.2.4.2  Type-of-Service
−
 
 
−
 
 
−
            host.
 
−
 
 
−
            (1)  If the datagram is sent in response to a received
 
−
                datagram, the source address for the response SHOULD be
 
−
                the specific-destination address of the request.  See
 
−
                Sections 4.1.3.5 and 4.2.3.7 and the "General Issues"
 
−
                section of [INTRO:1] for more specific requirements on
 
−
                higher layers.
 
−
 
 
−
                Otherwise, a source address must be selected.
 
−
 
 
−
            (2)  An application MUST be able to explicitly specify the
 
−
                source address for initiating a connection or a
 
−
                request.
 
−
 
 
−
            (3)  In the absence of such a specification, the networking
 
−
                software MUST choose a source address.  Rules for this
 
−
                choice are described below.
 
−
 
 
−
 
 
−
            There are two key requirement issues related to multihoming:
 
−
 
 
−
            (A)  A host MAY silently discard an incoming datagram whose
 
−
                destination address does not correspond to the physical
 
−
                interface through which it is received.
 
−
 
 
−
            (B)  A host MAY restrict itself to sending (non-source-
 
−
                routed) IP datagrams only through the physical
 
−
                interface that corresponds to the IP source address of
 
−
                the datagrams.
 
−
 
 
−
 
 
−
            DISCUSSION:
 
−
                Internet host implementors have used two different
 
−
                conceptual models for multihoming, briefly summarized
 
−
                in the following discussion.  This document takes no
 
−
                stand on which model is preferred; each seems to have a
 
−
                place.  This ambivalence is reflected in the issues (A)
 
−
                and (B) being optional.
 
−
 
 
−
                o    Strong ES Model
 
−
 
 
−
                      The Strong ES (End System, i.e., host) model
 
−
                      emphasizes the host/gateway (ES/IS) distinction,
 
−
                      and would therefore substitute MUST for MAY in
 
−
                      issues (A) and (B) above.  It tends to model a
 
−
                      multihomed host as a set of logical hosts within
 
−
                      the same physical host.
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 62]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                      INTERNET LAYER                October 1989
 
−
 
 
−
 
 
−
                      With respect to (A), proponents of the Strong ES
 
−
                      model note that automatic Internet routing
 
−
                      mechanisms could not route a datagram to a
 
−
                      physical interface that did not correspond to the
 
−
                      destination address.
 
−
 
 
−
                      Under the Strong ES model, the route computation
 
−
                      for an outgoing datagram is the mapping:
 
−
 
 
−
                        route(src IP addr, dest IP addr, TOS)
 
−
                                                        -> gateway
 
−
 
 
−
                      Here the source address is included as a parameter
 
−
                      in order to select a gateway that is directly
 
−
                      reachable on the corresponding physical interface.
 
−
                      Note that this model logically requires that in
 
−
                      general there be at least one default gateway, and
 
−
                      preferably multiple defaults, for each IP source
 
−
                      address.
 
−
 
 
−
                o    Weak ES Model
 
−
 
 
−
                      This view de-emphasizes the ES/IS distinction, and
 
−
                      would therefore substitute MUST NOT for MAY in
 
−
                      issues (A) and (B).  This model may be the more
 
−
                      natural one for hosts that wiretap gateway routing
 
−
                      protocols, and is necessary for hosts that have
 
−
                      embedded gateway functionality.
 
−
 
 
−
                      The Weak ES Model may cause the Redirect mechanism
 
−
                      to fail.  If a datagram is sent out a physical
 
−
                      interface that does not correspond to the
 
−
                      destination address, the first-hop gateway will
 
−
                      not realize when it needs to send a Redirect.  On
 
−
                      the other hand, if the host has embedded gateway
 
−
                      functionality, then it has routing information
 
−
                      without listening to Redirects.
 
−
 
 
−
                      In the Weak ES model, the route computation for an
 
−
                      outgoing datagram is the mapping:
 
−
 
 
−
                        route(dest IP addr, TOS) -> gateway, interface
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 63]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                      INTERNET LAYER                October 1989
 
−
 
 
−
 
 
−
        3.3.4.3  Choosing a Source Address
 
−
 
 
−
            DISCUSSION:
 
−
                When it sends an initial connection request (e.g., a
 
−
                TCP "SYN" segment) or a datagram service request (e.g.,
 
−
                a UDP-based query), the transport layer on a multihomed
 
−
                host needs to know which source address to use.  If the
 
−
                application does not specify it, the transport layer
 
−
                must ask the IP layer to perform the conceptual
 
−
                mapping:
 
−
 
 
−
                    GET_SRCADDR(remote IP addr, TOS)
 
−
                                              -> local IP address
 
−
 
 
−
                Here TOS is the Type-of-Service value (see Section
 
−
                3.2.1.6), and the result is the desired source address.
 
−
                The following rules are suggested for implementing this
 
−
                mapping:
 
−
 
 
−
                (a)  If the remote Internet address lies on one of the
 
−
                      (sub-) nets to which the host is directly
 
−
                      connected, a corresponding source address may be
 
−
                      chosen, unless the corresponding interface is
 
−
                      known to be down.
 
−
 
 
−
                (b)  The route cache may be consulted, to see if there
 
−
                      is an active route to the specified destination
 
−
                      network through any network interface; if so, a
 
−
                      local IP address corresponding to that interface
 
−
                      may be chosen.
 
−
 
 
−
                (c)  The table of static routes, if any (see Section
 
−
                      3.3.1.2) may be similarly consulted.
 
−
 
 
−
                (d)  The default gateways may be consulted.  If these
 
−
                      gateways are assigned to different interfaces, the
 
−
                      interface corresponding to the gateway with the
 
−
                      highest preference may be chosen.
 
−
 
 
−
                In the future, there may be a defined way for a
 
−
                multihomed host to ask the gateways on all connected
 
−
                networks for advice about the best network to use for a
 
−
                given destination.
 
−
 
 
−
            IMPLEMENTATION:
 
−
                It will be noted that this process is essentially the
 
−
                same as datagram routing (see Section 3.3.1), and
 
−
                therefore hosts may be able to combine the
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 64]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                      INTERNET LAYER                October 1989
 
−
 
 
−
 
 
−
                implementation of the two functions.
 
−
 
 
−
      3.3.5  Source Route Forwarding
 
−
 
 
−
        Subject to restrictions given below, a host MAY be able to act
 
−
        as an intermediate hop in a source route, forwarding a source-
 
−
        routed datagram to the next specified hop.
 
−
 
 
−
        However, in performing this gateway-like function, the host
 
−
        MUST obey all the relevant rules for a gateway forwarding
 
−
        source-routed datagrams [INTRO:2].  This includes the following
 
−
        specific provisions, which override the corresponding host
 
−
        provisions given earlier in this document:
 
−
 
 
−
        (A)  TTL (ref. Section 3.2.1.7)
 
−
 
 
−
              The TTL field MUST be decremented and the datagram perhaps
 
−
              discarded as specified for a gateway in [INTRO:2].
 
−
 
 
−
        (B)  ICMP Destination Unreachable (ref. Section 3.2.2.1)
 
−
 
 
−
              A host MUST be able to generate Destination Unreachable
 
−
              messages with the following codes:
 
−
 
 
−
              4    (Fragmentation Required but DF Set) when a source-
 
−
                  routed datagram cannot be fragmented to fit into the
 
−
                  target network;
 
−
 
 
−
              5    (Source Route Failed) when a source-routed datagram
 
−
                  cannot be forwarded, e.g., because of a routing
 
−
                  problem or because the next hop of a strict source
 
−
                  route is not on a connected network.
 
−
 
 
−
        (C)  IP Source Address (ref. Section 3.2.1.3)
 
−
 
 
−
              A source-routed datagram being forwarded MAY (and normally
 
−
              will) have a source address that is not one of the IP
 
−
              addresses of the forwarding host.
 
−
 
 
−
        (D)  Record Route Option (ref. Section 3.2.1.8d)
 
−
 
 
−
              A host that is forwarding a source-routed datagram
 
−
              containing a Record Route option MUST update that option,
 
−
              if it has room.
 
−
 
 
−
        (E)  Timestamp Option (ref. Section 3.2.1.8e)
 
−
 
 
−
              A host that is forwarding a source-routed datagram
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 65]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                      INTERNET LAYER                October 1989
 
−
 
 
−
 
 
−
              containing a Timestamp Option MUST add the current
 
−
              timestamp to that option, according to the rules for this
 
−
              option.
 
−
 
 
−
        To define the rules restricting host forwarding of source-
 
−
        routed datagrams, we use the term "local source-routing" if the
 
−
        next hop will be through the same physical interface through
 
−
        which the datagram arrived; otherwise, it is "non-local
 
−
        source-routing".
 
−
 
 
−
        o    A host is permitted to perform local source-routing
 
−
              without restriction.
 
−
 
 
−
        o    A host that supports non-local source-routing MUST have a
 
−
              configurable switch to disable forwarding, and this switch
 
−
              MUST default to disabled.
 
−
 
 
−
        o    The host MUST satisfy all gateway requirements for
 
−
              configurable policy filters [INTRO:2] restricting non-
 
−
              local forwarding.
 
−
 
 
−
        If a host receives a datagram with an incomplete source route
 
−
        but does not forward it for some reason, the host SHOULD return
 
−
        an ICMP Destination Unreachable (code 5, Source Route Failed)
 
−
        message, unless the datagram was itself an ICMP error message.
 
−
 
 
−
       3.3.6  Broadcasts
 
−
 
 
−
        Section 3.2.1.3 defined the four standard IP broadcast address
 
−
        forms:
 
−
 
 
−
          Limited Broadcast:  {-1, -1}
 
−
 
 
−
          Directed Broadcast:  {<Network-number>,-1}
 
−
 
 
−
          Subnet Directed Broadcast:
 
−
                              {<Network-number>,<Subnet-number>,-1}
 
−
 
 
−
          All-Subnets Directed Broadcast: {<Network-number>,-1,-1}
 
−
 
 
−
        A host MUST recognize any of these forms in the destination
 
−
        address of an incoming datagram.
 
−
 
 
−
        There is a class of hosts* that use non-standard broadcast
 
−
        address forms, substituting 0 for -1.  All hosts SHOULD
 
−
_________________________
 
−
*4.2BSD Unix and its derivatives, but not 4.3BSD.
 
−
 
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 66]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                      INTERNET LAYER                October 1989
 
−
 
 
−
 
 
−
        recognize and accept any of these non-standard broadcast
 
−
        addresses as the destination address of an incoming datagram.
 
−
        A host MAY optionally have a configuration option to choose the
 
−
        0 or the -1 form of broadcast address, for each physical
 
−
        interface, but this option SHOULD default to the standard (-1)
 
−
        form.
 
−
 
 
−
        When a host sends a datagram to a link-layer broadcast address,
 
−
        the IP destination address MUST be a legal IP broadcast or IP
 
−
        multicast address.
 
−
 
 
−
        A host SHOULD silently discard a datagram that is received via
 
−
        a link-layer broadcast (see Section 2.4) but does not specify
 
−
        an IP multicast or broadcast destination address.
 
−
 
 
−
        Hosts SHOULD use the Limited Broadcast address to broadcast to
 
−
        a connected network.
 
−
 
 
−
 
 
−
        DISCUSSION:
 
−
              Using the Limited Broadcast address instead of a Directed
 
−
              Broadcast address may improve system robustness.  Problems
 
−
              are often caused by machines that do not understand the
 
−
              plethora of broadcast addresses (see Section 3.2.1.3), or
 
−
              that may have different ideas about which broadcast
 
−
              addresses are in use.  The prime example of the latter is
 
−
              machines that do not understand subnetting but are
 
−
              attached to a subnetted net.  Sending a Subnet Broadcast
 
−
              for the connected network will confuse those machines,
 
−
              which will see it as a message to some other host.
 
−
 
 
−
              There has been discussion on whether a datagram addressed
 
−
              to the Limited Broadcast address ought to be sent from all
 
−
              the interfaces of a multihomed host. This specification
 
−
              takes no stand on the issue.
 
−
 
 
−
      3.3.7  IP Multicasting
 
−
 
 
−
        A host SHOULD support local IP multicasting on all connected
 
−
        networks for which a mapping from Class D IP addresses to
 
−
        link-layer addresses has been specified (see below).  Support
 
−
        for local IP multicasting includes sending multicast datagrams,
 
−
        joining multicast groups and receiving multicast datagrams, and
 
−
        leaving multicast groups.  This implies support for all of
 
−
        [IP:4] except the IGMP protocol itself, which is OPTIONAL.
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 67]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                      INTERNET LAYER                October 1989
 
−
 
 
−
 
 
−
        DISCUSSION:
 
−
              IGMP provides gateways that are capable of multicast
 
−
              routing with the information required to support IP
 
−
              multicasting across multiple networks.  At this time,
 
−
              multicast-routing gateways are in the experimental stage
 
−
              and are not widely available.  For hosts that are not
 
−
              connected to networks with multicast-routing gateways or
 
−
              that do not need to receive multicast datagrams
 
−
              originating on other networks, IGMP serves no purpose and
 
−
              is therefore optional for now.  However, the rest of
 
−
              [IP:4] is currently recommended for the purpose of
 
−
              providing IP-layer access to local network multicast
 
−
              addressing, as a preferable alternative to local broadcast
 
−
              addressing.  It is expected that IGMP will become
 
−
              recommended at some future date, when multicast-routing
 
−
              gateways have become more widely available.
 
−
 
 
−
        If IGMP is not implemented, a host SHOULD still join the "all-
 
−
        hosts" group (224.0.0.1) when the IP layer is initialized and
 
−
        remain a member for as long as the IP layer is active.
 
−
 
 
−
        DISCUSSION:
 
−
              Joining the "all-hosts" group will support strictly local
 
−
              uses of multicasting, e.g., a gateway discovery protocol,
 
−
              even if IGMP is not implemented.
 
−
 
 
−
        The mapping of IP Class D addresses to local addresses is
 
−
        currently specified for the following types of networks:
 
−
 
 
−
        o    Ethernet/IEEE 802.3, as defined in [IP:4].
 
−
 
 
−
        o    Any network that supports broadcast but not multicast,
 
−
              addressing: all IP Class D addresses map to the local
 
−
              broadcast address.
 
−
 
 
−
        o    Any type of point-to-point link (e.g., SLIP or HDLC
 
−
              links): no mapping required.  All IP multicast datagrams
 
−
              are sent as-is, inside the local framing.
 
−
 
 
−
        Mappings for other types of networks will be specified in the
 
−
        future.
 
−
 
 
−
        A host SHOULD provide a way for higher-layer protocols or
 
−
        applications to determine which of the host's connected
 
−
        network(s) support IP multicast addressing.
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 68]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                      INTERNET LAYER                October 1989
 
−
 
 
−
 
 
−
      3.3.8  Error Reporting
 
−
 
 
−
        Wherever practical, hosts MUST return ICMP error datagrams on
 
−
        detection of an error, except in those cases where returning an
 
−
        ICMP error message is specifically prohibited.
 
−
 
 
−
        DISCUSSION:
 
−
              A common phenomenon in datagram networks is the "black
 
−
              hole disease": datagrams are sent out, but nothing comes
 
−
              back.  Without any error datagrams, it is difficult for
 
−
              the user to figure out what the problem is.
 
−
 
 
−
  3.4  INTERNET/TRANSPORT LAYER INTERFACE
 
−
 
 
−
      The interface between the IP layer and the transport layer MUST
 
−
      provide full access to all the mechanisms of the IP layer,
 
−
      including options, Type-of-Service, and Time-to-Live.  The
 
−
      transport layer MUST either have mechanisms to set these interface
 
−
      parameters, or provide a path to pass them through from an
 
−
      application, or both.
 
−
 
 
−
      DISCUSSION:
 
−
          Applications are urged to make use of these mechanisms where
 
−
          applicable, even when the mechanisms are not currently
 
−
          effective in the Internet (e.g., TOS).  This will allow these
 
−
          mechanisms to be immediately useful when they do become
 
−
          effective, without a large amount of retrofitting of host
 
−
          software.
 
−
 
 
−
      We now describe a conceptual interface between the transport layer
 
−
      and the IP layer, as a set of procedure calls.  This is an
 
−
      extension of the information in Section 3.3 of RFC-791 [IP:1].
 
−
 
 
−
 
 
−
      *    Send Datagram
 
−
 
 
−
                SEND(src, dst, prot, TOS, TTL, BufPTR, len, Id, DF, opt
 
−
                    => result )
 
−
 
 
−
          where the parameters are defined in RFC-791.  Passing an Id
 
−
          parameter is optional; see Section 3.2.1.5.
 
−
 
 
−
 
 
−
      *    Receive Datagram
 
−
 
 
−
                RECV(BufPTR, prot
 
−
                    => result, src, dst, SpecDest, TOS, len, opt)
 
−
 
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 69]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                      INTERNET LAYER                October 1989
 
−
 
 
−
 
 
−
          All the parameters are defined in RFC-791, except for:
 
−
 
 
−
                SpecDest = specific-destination address of datagram
 
−
                            (defined in Section 3.2.1.3)
 
−
 
 
−
          The result parameter dst contains the datagram's destination
 
−
          address.  Since this may be a broadcast or multicast address,
 
−
          the SpecDest parameter (not shown in RFC-791) MUST be passed.
 
−
          The parameter opt contains all the IP options received in the
 
−
          datagram; these MUST also be passed to the transport layer.
 
−
 
 
−
 
 
−
      *    Select Source Address
 
−
 
 
−
                GET_SRCADDR(remote, TOS)  -> local
 
−
 
 
−
                remote = remote IP address
 
−
                TOS = Type-of-Service
 
−
                local = local IP address
 
−
 
 
−
          See Section 3.3.4.3.
 
−
 
 
−
 
 
−
      *    Find Maximum Datagram Sizes
 
−
 
 
−
                GET_MAXSIZES(local, remote, TOS) -> MMS_R, MMS_S
 
−
 
 
−
                MMS_R = maximum receive transport-message size.
 
−
                MMS_S = maximum send transport-message size.
 
−
              (local, remote, TOS defined above)
 
−
 
 
−
          See Sections 3.3.2 and 3.3.3.
 
−
 
 
−
 
 
−
      *    Advice on Delivery Success
 
−
 
 
−
                ADVISE_DELIVPROB(sense, local, remote, TOS)
 
−
 
 
−
          Here the parameter sense is a 1-bit flag indicating whether
 
−
          positive or negative advice is being given; see the
 
−
          discussion in Section 3.3.1.4. The other parameters were
 
−
          defined earlier.
 
−
 
 
−
 
 
−
      *    Send ICMP Message
 
−
 
 
−
                SEND_ICMP(src, dst, TOS, TTL, BufPTR, len, Id, DF, opt)
 
−
                    -> result
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 70]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                      INTERNET LAYER                October 1989
 
−
 
 
−
 
 
−
                (Parameters defined in RFC-791).
 
−
 
 
−
          Passing an Id parameter is optional; see Section 3.2.1.5.
 
−
          The transport layer MUST be able to send certain ICMP
 
−
          messages:  Port Unreachable or any of the query-type
 
−
          messages.  This function could be considered to be a special
 
−
          case of the SEND() call, of course; we describe it separately
 
−
          for clarity.
 
−
 
 
−
 
 
−
      *    Receive ICMP Message
 
−
 
 
−
                RECV_ICMP(BufPTR ) -> result, src, dst, len, opt
 
−
 
 
−
                (Parameters defined in RFC-791).
 
−
 
 
−
          The IP layer MUST pass certain ICMP messages up to the
 
−
          appropriate transport-layer routine.  This function could be
 
−
          considered to be a special case of the RECV() call, of
 
−
          course; we describe it separately for clarity.
 
−
 
 
−
          For an ICMP error message, the data that is passed up MUST
 
−
          include the original Internet header plus all the octets of
 
−
          the original message that are included in the ICMP message.
 
−
          This data will be used by the transport layer to locate the
 
−
          connection state information, if any.
 
−
 
 
−
          In particular, the following ICMP messages are to be passed
 
−
          up:
 
−
 
 
−
          o    Destination Unreachable
 
−
 
 
−
          o    Source Quench
 
−
 
 
−
          o    Echo Reply (to ICMP user interface, unless the Echo
 
−
                Request originated in the IP layer)
 
−
 
 
−
          o    Timestamp Reply (to ICMP user interface)
 
−
 
 
−
          o    Time Exceeded
 
−
 
 
−
 
 
−
      DISCUSSION:
 
−
          In the future, there may be additions to this interface to
 
−
          pass path data (see Section 3.3.1.3) between the IP and
 
−
          transport layers.
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 71]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                      INTERNET LAYER                October 1989
 
−
 
 
−
 
 
−
  3.5  INTERNET LAYER REQUIREMENTS SUMMARY
 
−
 
 
−
 
 
−
                                                |        | | | |S| |
 
−
                                                |        | | | |H| |F
 
−
                                                |        | | | |O|M|o
 
−
                                                |        | |S| |U|U|o
 
−
                                                |        | |H| |L|S|t
 
−
                                                |        |M|O| |D|T|n
 
−
                                                |        |U|U|M| | |o
 
−
                                                |        |S|L|A|N|N|t
 
−
                                                |        |T|D|Y|O|O|t
 
−
FEATURE                                          |SECTION | | | |T|T|e
 
−
-------------------------------------------------|--------|-|-|-|-|-|--
 
−
                                                |        | | | | | |
 
−
Implement IP and ICMP                            |3.1    |x| | | | |
 
−
Handle remote multihoming in application layer  |3.1    |x| | | | |
 
−
Support local multihoming                        |3.1    | | |x| | |
 
−
Meet gateway specs if forward datagrams          |3.1    |x| | | | |
 
−
Configuration switch for embedded gateway        |3.1    |x| | | | |1
 
−
  Config switch default to non-gateway          |3.1    |x| | | | |1
 
−
  Auto-config based on number of interfaces    |3.1    | | | | |x|1
 
−
Able to log discarded datagrams                  |3.1    | |x| | | |
 
−
  Record in counter                            |3.1    | |x| | | |
 
−
                                                |        | | | | | |
 
−
Silently discard Version != 4                    |3.2.1.1 |x| | | | |
 
−
Verify IP checksum, silently discard bad dgram  |3.2.1.2 |x| | | | |
 
−
Addressing:                                      |        | | | | | |
 
−
  Subnet addressing (RFC-950)                    |3.2.1.3 |x| | | | |
 
−
  Src address must be host's own IP address      |3.2.1.3 |x| | | | |
 
−
  Silently discard datagram with bad dest addr  |3.2.1.3 |x| | | | |
 
−
  Silently discard datagram with bad src addr    |3.2.1.3 |x| | | | |
 
−
Support reassembly                              |3.2.1.4 |x| | | | |
 
−
Retain same Id field in identical datagram      |3.2.1.5 | | |x| | |
 
−
                                                |        | | | | | |
 
−
TOS:                                            |        | | | | | |
 
−
  Allow transport layer to set TOS              |3.2.1.6 |x| | | | |
 
−
  Pass received TOS up to transport layer        |3.2.1.6 | |x| | | |
 
−
  Use RFC-795 link-layer mappings for TOS        |3.2.1.6 | | | |x| |
 
−
TTL:                                            |        | | | | | |
 
−
  Send packet with TTL of 0                      |3.2.1.7 | | | | |x|
 
−
  Discard received packets with TTL < 2          |3.2.1.7 | | | | |x|
 
−
  Allow transport layer to set TTL              |3.2.1.7 |x| | | | |
 
−
  Fixed TTL is configurable                      |3.2.1.7 |x| | | | |
 
−
                                                |        | | | | | |
 
−
IP Options:                                      |        | | | | | |
 
−
  Allow transport layer to send IP options      |3.2.1.8 |x| | | | |
 
−
  Pass all IP options rcvd to higher layer      |3.2.1.8 |x| | | | |
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 72]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                      INTERNET LAYER                October 1989
 
−
 
 
−
 
 
−
  IP layer silently ignore unknown options      |3.2.1.8 |x| | | | |
 
−
  Security option                                |3.2.1.8a| | |x| | |
 
−
  Send Stream Identifier option                  |3.2.1.8b| | | |x| |
 
−
  Silently ignore Stream Identifer option        |3.2.1.8b|x| | | | |
 
−
  Record Route option                            |3.2.1.8d| | |x| | |
 
−
  Timestamp option                              |3.2.1.8e| | |x| | |
 
−
Source Route Option:                            |        | | | | | |
 
−
  Originate & terminate Source Route options    |3.2.1.8c|x| | | | |
 
−
  Datagram with completed SR passed up to TL    |3.2.1.8c|x| | | | |
 
−
  Build correct (non-redundant) return route    |3.2.1.8c|x| | | | |
 
−
  Send multiple SR options in one header        |3.2.1.8c| | | | |x|
 
−
                                                |        | | | | | |
 
−
ICMP:                                            |        | | | | | |
 
−
  Silently discard ICMP msg with unknown type    |3.2.2  |x| | | | |
 
−
  Include more than 8 octets of orig datagram    |3.2.2  | | |x| | |
 
−
      Included octets same as received          |3.2.2  |x| | | | |
 
−
  Demux ICMP Error to transport protocol        |3.2.2  |x| | | | |
 
−
  Send ICMP error message with TOS=0            |3.2.2  | |x| | | |
 
−
  Send ICMP error message for:                  |        | | | | | |
 
−
  - ICMP error msg                              |3.2.2  | | | | |x|
 
−
  - IP b'cast or IP m'cast                      |3.2.2  | | | | |x|
 
−
  - Link-layer b'cast                          |3.2.2  | | | | |x|
 
−
  - Non-initial fragment                        |3.2.2  | | | | |x|
 
−
  - Datagram with non-unique src address        |3.2.2  | | | | |x|
 
−
  Return ICMP error msgs (when not prohibited)  |3.3.8  |x| | | | |
 
−
                                                |        | | | | | |
 
−
  Dest Unreachable:                              |        | | | | | |
 
−
    Generate Dest Unreachable (code 2/3)        |3.2.2.1 | |x| | | |
 
−
    Pass ICMP Dest Unreachable to higher layer  |3.2.2.1 |x| | | | |
 
−
    Higher layer act on Dest Unreach            |3.2.2.1 | |x| | | |
 
−
      Interpret Dest Unreach as only hint        |3.2.2.1 |x| | | | |
 
−
  Redirect:                                      |        | | | | | |
 
−
    Host send Redirect                          |3.2.2.2 | | | |x| |
 
−
    Update route cache when recv Redirect        |3.2.2.2 |x| | | | |
 
−
    Handle both Host and Net Redirects          |3.2.2.2 |x| | | | |
 
−
    Discard illegal Redirect                    |3.2.2.2 | |x| | | |
 
−
  Source Quench:                                |        | | | | | |
 
−
    Send Source Quench if buffering exceeded    |3.2.2.3 | | |x| | |
 
−
    Pass Source Quench to higher layer          |3.2.2.3 |x| | | | |
 
−
    Higher layer act on Source Quench            |3.2.2.3 | |x| | | |
 
−
  Time Exceeded: pass to higher layer            |3.2.2.4 |x| | | | |
 
−
  Parameter Problem:                            |        | | | | | |
 
−
    Send Parameter Problem messages              |3.2.2.5 | |x| | | |
 
−
    Pass Parameter Problem to higher layer      |3.2.2.5 |x| | | | |
 
−
    Report Parameter Problem to user            |3.2.2.5 | | |x| | |
 
−
                                                |        | | | | | |
 
−
  ICMP Echo Request or Reply:                    |        | | | | | |
 
−
    Echo server and Echo client                  |3.2.2.6 |x| | | | |
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 73]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                      INTERNET LAYER                October 1989
 
−
 
 
−
 
 
−
    Echo client                                  |3.2.2.6 | |x| | | |
 
−
    Discard Echo Request to broadcast address    |3.2.2.6 | | |x| | |
 
−
    Discard Echo Request to multicast address    |3.2.2.6 | | |x| | |
 
−
    Use specific-dest addr as Echo Reply src    |3.2.2.6 |x| | | | |
 
−
    Send same data in Echo Reply                |3.2.2.6 |x| | | | |
 
−
    Pass Echo Reply to higher layer              |3.2.2.6 |x| | | | |
 
−
    Reflect Record Route, Time Stamp options    |3.2.2.6 | |x| | | |
 
−
    Reverse and reflect Source Route option      |3.2.2.6 |x| | | | |
 
−
                                                |        | | | | | |
 
−
  ICMP Information Request or Reply:            |3.2.2.7 | | | |x| |
 
−
  ICMP Timestamp and Timestamp Reply:            |3.2.2.8 | | |x| | |
 
−
    Minimize delay variability                  |3.2.2.8 | |x| | | |1
 
−
    Silently discard b'cast Timestamp            |3.2.2.8 | | |x| | |1
 
−
    Silently discard m'cast Timestamp            |3.2.2.8 | | |x| | |1
 
−
    Use specific-dest addr as TS Reply src      |3.2.2.8 |x| | | | |1
 
−
    Reflect Record Route, Time Stamp options    |3.2.2.6 | |x| | | |1
 
−
    Reverse and reflect Source Route option      |3.2.2.8 |x| | | | |1
 
−
    Pass Timestamp Reply to higher layer        |3.2.2.8 |x| | | | |1
 
−
    Obey rules for "standard value"              |3.2.2.8 |x| | | | |1
 
−
                                                |        | | | | | |
 
−
  ICMP Address Mask Request and Reply:          |        | | | | | |
 
−
    Addr Mask source configurable                |3.2.2.9 |x| | | | |
 
−
    Support static configuration of addr mask    |3.2.2.9 |x| | | | |
 
−
    Get addr mask dynamically during booting    |3.2.2.9 | | |x| | |
 
−
    Get addr via ICMP Addr Mask Request/Reply    |3.2.2.9 | | |x| | |
 
−
      Retransmit Addr Mask Req if no Reply      |3.2.2.9 |x| | | | |3
 
−
      Assume default mask if no Reply            |3.2.2.9 | |x| | | |3
 
−
      Update address mask from first Reply only  |3.2.2.9 |x| | | | |3
 
−
    Reasonableness check on Addr Mask            |3.2.2.9 | |x| | | |
 
−
    Send unauthorized Addr Mask Reply msgs      |3.2.2.9 | | | | |x|
 
−
      Explicitly configured to be agent          |3.2.2.9 |x| | | | |
 
−
    Static config=> Addr-Mask-Authoritative flag |3.2.2.9 | |x| | | |
 
−
      Broadcast Addr Mask Reply when init.      |3.2.2.9 |x| | | | |3
 
−
                                                |        | | | | | |
 
−
ROUTING OUTBOUND DATAGRAMS:                      |        | | | | | |
 
−
  Use address mask in local/remote decision      |3.3.1.1 |x| | | | |
 
−
  Operate with no gateways on conn network      |3.3.1.1 |x| | | | |
 
−
  Maintain "route cache" of next-hop gateways    |3.3.1.2 |x| | | | |
 
−
  Treat Host and Net Redirect the same          |3.3.1.2 | |x| | | |
 
−
  If no cache entry, use default gateway        |3.3.1.2 |x| | | | |
 
−
    Support multiple default gateways            |3.3.1.2 |x| | | | |
 
−
  Provide table of static routes                |3.3.1.2 | | |x| | |
 
−
    Flag: route overridable by Redirects        |3.3.1.2 | | |x| | |
 
−
  Key route cache on host, not net address      |3.3.1.3 | | |x| | |
 
−
  Include TOS in route cache                    |3.3.1.3 | |x| | | |
 
−
                                                |        | | | | | |
 
−
  Able to detect failure of next-hop gateway    |3.3.1.4 |x| | | | |
 
−
  Assume route is good forever                  |3.3.1.4 | | | |x| |
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 74]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                      INTERNET LAYER                October 1989
 
−
 
 
−
 
 
−
  Ping gateways continuously                    |3.3.1.4 | | | | |x|
 
−
  Ping only when traffic being sent              |3.3.1.4 |x| | | | |
 
−
  Ping only when no positive indication          |3.3.1.4 |x| | | | |
 
−
  Higher and lower layers give advice            |3.3.1.4 | |x| | | |
 
−
  Switch from failed default g'way to another    |3.3.1.5 |x| | | | |
 
−
  Manual method of entering config info          |3.3.1.6 |x| | | | |
 
−
                                                |        | | | | | |
 
−
REASSEMBLY and FRAGMENTATION:                    |        | | | | | |
 
−
  Able to reassemble incoming datagrams          |3.3.2  |x| | | | |
 
−
    At least 576 byte datagrams                  |3.3.2  |x| | | | |
 
−
    EMTU_R configurable or indefinite            |3.3.2  | |x| | | |
 
−
  Transport layer able to learn MMS_R            |3.3.2  |x| | | | |
 
−
  Send ICMP Time Exceeded on reassembly timeout  |3.3.2  |x| | | | |
 
−
    Fixed reassembly timeout value              |3.3.2  | |x| | | |
 
−
                                                |        | | | | | |
 
−
  Pass MMS_S to higher layers                    |3.3.3  |x| | | | |
 
−
  Local fragmentation of outgoing packets        |3.3.3  | | |x| | |
 
−
    Else don't send bigger than MMS_S          |3.3.3  |x| | | | |
 
−
  Send max 576 to off-net destination            |3.3.3  | |x| | | |
 
−
  All-Subnets-MTU configuration flag            |3.3.3  | | |x| | |
 
−
                                                |        | | | | | |
 
−
MULTIHOMING:                                    |        | | | | | |
 
−
  Reply with same addr as spec-dest addr        |3.3.4.2 | |x| | | |
 
−
  Allow application to choose local IP addr      |3.3.4.2 |x| | | | |
 
−
  Silently discard d'gram in "wrong" interface  |3.3.4.2 | | |x| | |
 
−
  Only send d'gram through "right" interface    |3.3.4.2 | | |x| | |4
 
−
                                                |        | | | | | |
 
−
SOURCE-ROUTE FORWARDING:                        |        | | | | | |
 
−
  Forward datagram with Source Route option      |3.3.5  | | |x| | |1
 
−
    Obey corresponding gateway rules            |3.3.5  |x| | | | |1
 
−
      Update TTL by gateway rules                |3.3.5  |x| | | | |1
 
−
      Able to generate ICMP err code 4, 5        |3.3.5  |x| | | | |1
 
−
      IP src addr not local host                |3.3.5  | | |x| | |1
 
−
      Update Timestamp, Record Route options    |3.3.5  |x| | | | |1
 
−
    Configurable switch for non-local SRing      |3.3.5  |x| | | | |1
 
−
      Defaults to OFF                            |3.3.5  |x| | | | |1
 
−
    Satisfy gwy access rules for non-local SRing |3.3.5  |x| | | | |1
 
−
    If not forward, send Dest Unreach (cd 5)    |3.3.5  | |x| | | |2
 
−
                                                |        | | | | | |
 
−
BROADCAST:                                      |        | | | | | |
 
−
  Broadcast addr as IP source addr              |3.2.1.3 | | | | |x|
 
−
  Receive 0 or -1 broadcast formats OK          |3.3.6  | |x| | | |
 
−
  Config'ble option to send 0 or -1 b'cast      |3.3.6  | | |x| | |
 
−
    Default to -1 broadcast                      |3.3.6  | |x| | | |
 
−
  Recognize all broadcast address formats        |3.3.6  |x| | | | |
 
−
  Use IP b'cast/m'cast addr in link-layer b'cast |3.3.6  |x| | | | |
 
−
  Silently discard link-layer-only b'cast dg's  |3.3.6  | |x| | | |
 
−
  Use Limited Broadcast addr for connected net  |3.3.6  | |x| | | |
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 75]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                      INTERNET LAYER                October 1989
 
−
 
 
−
 
 
−
                                                |        | | | | | |
 
−
MULTICAST:                                      |        | | | | | |
 
−
  Support local IP multicasting (RFC-1112)      |3.3.7  | |x| | | |
 
−
  Support IGMP (RFC-1112)                        |3.3.7  | | |x| | |
 
−
  Join all-hosts group at startup                |3.3.7  | |x| | | |
 
−
  Higher layers learn i'face m'cast capability  |3.3.7  | |x| | | |
 
−
                                                |        | | | | | |
 
−
INTERFACE:                                      |        | | | | | |
 
−
  Allow transport layer to use all IP mechanisms |3.4    |x| | | | |
 
−
  Pass interface ident up to transport layer    |3.4    |x| | | | |
 
−
  Pass all IP options up to transport layer      |3.4    |x| | | | |
 
−
  Transport layer can send certain ICMP messages |3.4    |x| | | | |
 
−
  Pass spec'd ICMP messages up to transp. layer  |3.4    |x| | | | |
 
−
    Include IP hdr+8 octets or more from orig.  |3.4    |x| | | | |
 
−
  Able to leap tall buildings at a single bound  |3.5    | |x| | | |
 
−
 
 
−
Footnotes:
 
−
 
 
−
(1)  Only if feature is implemented.
 
−
 
 
−
(2)  This requirement is overruled if datagram is an ICMP error message.
 
−
 
 
−
(3)  Only if feature is implemented and is configured "on".
 
−
 
 
−
(4)  Unless has embedded gateway functionality or is source routed.
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 76]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                  TRANSPORT LAYER -- UDP            October 1989
 
−
 
 
−
 
 
−
4. TRANSPORT PROTOCOLS
 
−
 
 
−
  4.1  USER DATAGRAM PROTOCOL -- UDP
 
−
 
 
−
      4.1.1  INTRODUCTION
 
−
 
 
−
        The User Datagram Protocol UDP [UDP:1] offers only a minimal
 
−
        transport service -- non-guaranteed datagram delivery -- and
 
−
        gives applications direct access to the datagram service of the
 
−
        IP layer.  UDP is used by applications that do not require the
 
−
        level of service of TCP or that wish to use communications
 
−
        services (e.g., multicast or broadcast delivery) not available
 
−
        from TCP.
 
−
 
 
−
        UDP is almost a null protocol; the only services it provides
 
−
        over IP are checksumming of data and multiplexing by port
 
−
        number.  Therefore, an application program running over UDP
 
−
        must deal directly with end-to-end communication problems that
 
−
        a connection-oriented protocol would have handled -- e.g.,
 
−
        retransmission for reliable delivery, packetization and
 
−
        reassembly, flow control, congestion avoidance, etc., when
 
−
        these are required.  The fairly complex coupling between IP and
 
−
        TCP will be mirrored in the coupling between UDP and many
 
−
        applications using UDP.
 
−
 
 
−
      4.1.2  PROTOCOL WALK-THROUGH
 
−
 
 
−
        There are no known errors in the specification of UDP.
 
−
 
 
−
      4.1.3  SPECIFIC ISSUES
 
−
 
 
−
        4.1.3.1  Ports
 
−
 
 
−
            UDP well-known ports follow the same rules as TCP well-known
 
−
            ports; see Section 4.2.2.1 below.
 
−
 
 
−
            If a datagram arrives addressed to a UDP port for which
 
−
            there is no pending LISTEN call, UDP SHOULD send an ICMP
 
−
            Port Unreachable message.
 
−
 
 
−
        4.1.3.2  IP Options
 
−
 
 
−
            UDP MUST pass any IP option that it receives from the IP
 
−
            layer transparently to the application layer.
 
−
 
 
−
            An application MUST be able to specify IP options to be sent
 
−
            in its UDP datagrams, and UDP MUST pass these options to the
 
−
            IP layer.
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 77]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                  TRANSPORT LAYER -- UDP            October 1989
 
−
 
 
−
 
 
−
            DISCUSSION:
 
−
                At present, the only options that need be passed
 
−
                through UDP are Source Route, Record Route, and Time
 
−
                Stamp.  However, new options may be defined in the
 
−
                future, and UDP need not and should not make any
 
−
                assumptions about the format or content of options it
 
−
                passes to or from the application; an exception to this
 
−
                might be an IP-layer security option.
 
−
 
 
−
                An application based on UDP will need to obtain a
 
−
                source route from a request datagram and supply a
 
−
                reversed route for sending the corresponding reply.
 
−
 
 
−
        4.1.3.3  ICMP Messages
 
−
 
 
−
            UDP MUST pass to the application layer all ICMP error
 
−
            messages that it receives from the IP layer.  Conceptually
 
−
            at least, this may be accomplished with an upcall to the
 
−
            ERROR_REPORT routine (see Section 4.2.4.1).
 
−
 
 
−
            DISCUSSION:
 
−
                Note that ICMP error messages resulting from sending a
 
−
                UDP datagram are received asynchronously.  A UDP-based
 
−
                application that wants to receive ICMP error messages
 
−
                is responsible for maintaining the state necessary to
 
−
                demultiplex these messages when they arrive; for
 
−
                example, the application may keep a pending receive
 
−
                operation for this purpose.  The application is also
 
−
                responsible to avoid confusion from a delayed ICMP
 
−
                error message resulting from an earlier use of the same
 
−
                port(s).
 
−
 
 
−
        4.1.3.4  UDP Checksums
 
−
 
 
−
            A host MUST implement the facility to generate and validate
 
−
            UDP checksums.  An application MAY optionally be able to
 
−
            control whether a UDP checksum will be generated, but it
 
−
            MUST default to checksumming on.
 
−
 
 
−
            If a UDP datagram is received with a checksum that is non-
 
−
            zero and invalid, UDP MUST silently discard the datagram.
 
−
            An application MAY optionally be able to control whether UDP
 
−
            datagrams without checksums should be discarded or passed to
 
−
            the application.
 
−
 
 
−
            DISCUSSION:
 
−
                Some applications that normally run only across local
 
−
                area networks have chosen to turn off UDP checksums for
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 78]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                  TRANSPORT LAYER -- UDP            October 1989
 
−
 
 
−
 
 
−
                efficiency.  As a result, numerous cases of undetected
 
−
                errors have been reported.  The advisability of ever
 
−
                turning off UDP checksumming is very controversial.
 
−
 
 
−
            IMPLEMENTATION:
 
−
                There is a common implementation error in UDP
 
−
                checksums.  Unlike the TCP checksum, the UDP checksum
 
−
                is optional; the value zero is transmitted in the
 
−
                checksum field of a UDP header to indicate the absence
 
−
                of a checksum.  If the transmitter really calculates a
 
−
                UDP checksum of zero, it must transmit the checksum as
 
−
                all 1's (65535).  No special action is required at the
 
−
                receiver, since zero and 65535 are equivalent in 1's
 
−
                complement arithmetic.
 
−
 
 
−
        4.1.3.5  UDP Multihoming
 
−
 
 
−
            When a UDP datagram is received, its specific-destination
 
−
            address MUST be passed up to the application layer.
 
−
 
 
−
            An application program MUST be able to specify the IP source
 
−
            address to be used for sending a UDP datagram or to leave it
 
−
            unspecified (in which case the networking software will
 
−
            choose an appropriate source address).  There SHOULD be a
 
−
            way to communicate the chosen source address up to the
 
−
            application layer (e.g, so that the application can later
 
−
            receive a reply datagram only from the corresponding
 
−
            interface).
 
−
 
 
−
            DISCUSSION:
 
−
                A request/response application that uses UDP should use
 
−
                a source address for the response that is the same as
 
−
                the specific destination address of the request.  See
 
−
                the "General Issues" section of [INTRO:1].
 
−
 
 
−
        4.1.3.6  Invalid Addresses
 
−
 
 
−
            A UDP datagram received with an invalid IP source address
 
−
            (e.g., a broadcast or multicast address) must be discarded
 
−
            by UDP or by the IP layer (see Section 3.2.1.3).
 
−
 
 
−
            When a host sends a UDP datagram, the source address MUST be
 
−
            (one of) the IP address(es) of the host.
 
−
 
 
−
      4.1.4  UDP/APPLICATION LAYER INTERFACE
 
−
 
 
−
        The application interface to UDP MUST provide the full services
 
−
        of the IP/transport interface described in Section 3.4 of this
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 79]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                  TRANSPORT LAYER -- UDP            October 1989
 
−
 
 
−
 
 
−
        document.  Thus, an application using UDP needs the functions
 
−
        of the GET_SRCADDR(), GET_MAXSIZES(), ADVISE_DELIVPROB(), and
 
−
        RECV_ICMP() calls described in Section 3.4.  For example,
 
−
        GET_MAXSIZES() can be used to learn the effective maximum UDP
 
−
        maximum datagram size for a particular {interface,remote
 
−
        host,TOS} triplet.
 
−
 
 
−
        An application-layer program MUST be able to set the TTL and
 
−
        TOS values as well as IP options for sending a UDP datagram,
 
−
        and these values must be passed transparently to the IP layer.
 
−
        UDP MAY pass the received TOS up to the application layer.
 
−
 
 
−
      4.1.5  UDP REQUIREMENTS SUMMARY
 
−
 
 
−
 
 
−
                                                |        | | | |S| |
 
−
                                                |        | | | |H| |F
 
−
                                                |        | | | |O|M|o
 
−
                                                |        | |S| |U|U|o
 
−
                                                |        | |H| |L|S|t
 
−
                                                |        |M|O| |D|T|n
 
−
                                                |        |U|U|M| | |o
 
−
                                                |        |S|L|A|N|N|t
 
−
                                                |        |T|D|Y|O|O|t
 
−
FEATURE                                          |SECTION | | | |T|T|e
 
−
-------------------------------------------------|--------|-|-|-|-|-|--
 
−
                                                |        | | | | | |
 
−
    UDP                                          |        | | | | | |
 
−
-------------------------------------------------|--------|-|-|-|-|-|--
 
−
                                                |        | | | | | |
 
−
UDP send Port Unreachable                        |4.1.3.1 | |x| | | |
 
−
                                                |        | | | | | |
 
−
IP Options in UDP                                |        | | | | | |
 
−
- Pass rcv'd IP options to applic layer        |4.1.3.2 |x| | | | |
 
−
- Applic layer can specify IP options in Send  |4.1.3.2 |x| | | | |
 
−
- UDP passes IP options down to IP layer        |4.1.3.2 |x| | | | |
 
−
                                                |        | | | | | |
 
−
Pass ICMP msgs up to applic layer                |4.1.3.3 |x| | | | |
 
−
                                                |        | | | | | |
 
−
UDP checksums:                                  |        | | | | | |
 
−
- Able to generate/check checksum              |4.1.3.4 |x| | | | |
 
−
- Silently discard bad checksum                |4.1.3.4 |x| | | | |
 
−
- Sender Option to not generate checksum        |4.1.3.4 | | |x| | |
 
−
  - Default is to checksum                      |4.1.3.4 |x| | | | |
 
−
- Receiver Option to require checksum          |4.1.3.4 | | |x| | |
 
−
                                                |        | | | | | |
 
−
UDP Multihoming                                  |        | | | | | |
 
−
- Pass spec-dest addr to application            |4.1.3.5 |x| | | | |
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 80]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                  TRANSPORT LAYER -- UDP            October 1989
 
−
 
 
−
 
 
−
- Applic layer can specify Local IP addr        |4.1.3.5 |x| | | | |
 
−
- Applic layer specify wild Local IP addr      |4.1.3.5 |x| | | | |
 
−
- Applic layer notified of Local IP addr used  |4.1.3.5 | |x| | | |
 
−
                                                |        | | | | | |
 
−
Bad IP src addr silently discarded by UDP/IP    |4.1.3.6 |x| | | | |
 
−
Only send valid IP source address                |4.1.3.6 |x| | | | |
 
−
UDP Application Interface Services              |        | | | | | |
 
−
Full IP interface of 3.4 for application        |4.1.4  |x| | | | |
 
−
- Able to spec TTL, TOS, IP opts when send dg  |4.1.4  |x| | | | |
 
−
- Pass received TOS up to applic layer          |4.1.4  | | |x| | |
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 81]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
 
−
 
 
−
 
 
−
  4.2  TRANSMISSION CONTROL PROTOCOL -- TCP
 
−
 
 
−
      4.2.1  INTRODUCTION
 
−
 
 
−
        The Transmission Control Protocol TCP [TCP:1] is the primary
 
−
        virtual-circuit transport protocol for the Internet suite.  TCP
 
−
        provides reliable, in-sequence delivery of a full-duplex stream
 
−
        of octets (8-bit bytes).  TCP is used by those applications
 
−
        needing reliable, connection-oriented transport service, e.g.,
 
−
        mail (SMTP), file transfer (FTP), and virtual terminal service
 
−
        (Telnet); requirements for these application-layer protocols
 
−
        are described in [INTRO:1].
 
−
 
 
−
      4.2.2  PROTOCOL WALK-THROUGH
 
−
 
 
−
        4.2.2.1  Well-Known Ports: RFC-793 Section 2.7
 
−
 
 
−
            DISCUSSION:
 
−
                TCP reserves port numbers in the range 0-255 for
 
−
                "well-known" ports, used to access services that are
 
−
                standardized across the Internet.  The remainder of the
 
−
                port space can be freely allocated to application
 
−
                processes.  Current well-known port definitions are
 
−
                listed in the RFC entitled "Assigned Numbers"
 
−
                [INTRO:6].  A prerequisite for defining a new well-
 
−
                known port is an RFC documenting the proposed service
 
−
                in enough detail to allow new implementations.
 
−
 
 
−
                Some systems extend this notion by adding a third
 
−
                subdivision of the TCP port space: reserved ports,
 
−
                which are generally used for operating-system-specific
 
−
                services.  For example, reserved ports might fall
 
−
                between 256 and some system-dependent upper limit.
 
−
                Some systems further choose to protect well-known and
 
−
                reserved ports by permitting only privileged users to
 
−
                open TCP connections with those port values.  This is
 
−
                perfectly reasonable as long as the host does not
 
−
                assume that all hosts protect their low-numbered ports
 
−
                in this manner.
 
−
 
 
−
        4.2.2.2  Use of Push: RFC-793 Section 2.8
 
−
 
 
−
            When an application issues a series of SEND calls without
 
−
            setting the PUSH flag, the TCP MAY aggregate the data
 
−
            internally without sending it.  Similarly, when a series of
 
−
            segments is received without the PSH bit, a TCP MAY queue
 
−
            the data internally without passing it to the receiving
 
−
            application.
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 82]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
 
−
 
 
−
 
 
−
            The PSH bit is not a record marker and is independent of
 
−
            segment boundaries.  The transmitter SHOULD collapse
 
−
            successive PSH bits when it packetizes data, to send the
 
−
            largest possible segment.
 
−
 
 
−
            A TCP MAY implement PUSH flags on SEND calls.  If PUSH flags
 
−
            are not implemented, then the sending TCP: (1) must not
 
−
            buffer data indefinitely, and (2) MUST set the PSH bit in
 
−
            the last buffered segment (i.e., when there is no more
 
−
            queued data to be sent).
 
−
 
 
−
            The discussion in RFC-793 on pages 48, 50, and 74
 
−
            erroneously implies that a received PSH flag must be passed
 
−
            to the application layer.  Passing a received PSH flag to
 
−
            the application layer is now OPTIONAL.
 
−
 
 
−
            An application program is logically required to set the PUSH
 
−
            flag in a SEND call whenever it needs to force delivery of
 
−
            the data to avoid a communication deadlock.  However, a TCP
 
−
            SHOULD send a maximum-sized segment whenever possible, to
 
−
            improve performance (see Section 4.2.3.4).
 
−
 
 
−
            DISCUSSION:
 
−
                When the PUSH flag is not implemented on SEND calls,
 
−
                i.e., when the application/TCP interface uses a pure
 
−
                streaming model, responsibility for aggregating any
 
−
                tiny data fragments to form reasonable sized segments
 
−
                is partially borne by the application layer.
 
−
 
 
−
                Generally, an interactive application protocol must set
 
−
                the PUSH flag at least in the last SEND call in each
 
−
                command or response sequence.  A bulk transfer protocol
 
−
                like FTP should set the PUSH flag on the last segment
 
−
                of a file or when necessary to prevent buffer deadlock.
 
−
 
 
−
                At the receiver, the PSH bit forces buffered data to be
 
−
                delivered to the application (even if less than a full
 
−
                buffer has been received). Conversely, the lack of a
 
−
                PSH bit can be used to avoid unnecessary wakeup calls
 
−
                to the application process; this can be an important
 
−
                performance optimization for large timesharing hosts.
 
−
                Passing the PSH bit to the receiving application allows
 
−
                an analogous optimization within the application.
 
−
 
 
−
        4.2.2.3  Window Size: RFC-793 Section 3.1
 
−
 
 
−
            The window size MUST be treated as an unsigned number, or
 
−
            else large window sizes will appear like negative windows
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 83]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
 
−
 
 
−
 
 
−
            and TCP will not work.  It is RECOMMENDED that
 
−
            implementations reserve 32-bit fields for the send and
 
−
            receive window sizes in the connection record and do all
 
−
            window computations with 32 bits.
 
−
 
 
−
            DISCUSSION:
 
−
                It is known that the window field in the TCP header is
 
−
                too small for high-speed, long-delay paths.
 
−
                Experimental TCP options have been defined to extend
 
−
                the window size; see for example [TCP:11].  In
 
−
                anticipation of the adoption of such an extension, TCP
 
−
                implementors should treat windows as 32 bits.
 
−
 
 
−
        4.2.2.4  Urgent Pointer: RFC-793 Section 3.1
 
−
 
 
−
            The second sentence is in error: the urgent pointer points
 
−
            to the sequence number of the LAST octet (not LAST+1) in a
 
−
            sequence of urgent data.  The description on page 56 (last
 
−
            sentence) is correct.
 
−
 
 
−
            A TCP MUST support a sequence of urgent data of any length.
 
−
 
 
−
            A TCP MUST inform the application layer asynchronously
 
−
            whenever it receives an Urgent pointer and there was
 
−
            previously no pending urgent data, or whenever the Urgent
 
−
            pointer advances in the data stream.  There MUST be a way
 
−
            for the application to learn how much urgent data remains to
 
−
            be read from the connection, or at least to determine
 
−
            whether or not more urgent data remains to be read.
 
−
 
 
−
            DISCUSSION:
 
−
                Although the Urgent mechanism may be used for any
 
−
                application, it is normally used to send "interrupt"-
 
−
                type commands to a Telnet program (see "Using Telnet
 
−
                Synch Sequence" section in [INTRO:1]).
 
−
 
 
−
                The asynchronous or "out-of-band" notification will
 
−
                allow the application to go into "urgent mode", reading
 
−
                data from the TCP connection.  This allows control
 
−
                commands to be sent to an application whose normal
 
−
                input buffers are full of unprocessed data.
 
−
 
 
−
            IMPLEMENTATION:
 
−
                The generic ERROR-REPORT() upcall described in Section
 
−
                4.2.4.1 is a possible mechanism for informing the
 
−
                application of the arrival of urgent data.
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 84]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
 
−
 
 
−
 
 
−
        4.2.2.5  TCP Options: RFC-793 Section 3.1
 
−
 
 
−
            A TCP MUST be able to receive a TCP option in any segment.
 
−
            A TCP MUST ignore without error any TCP option it does not
 
−
            implement, assuming that the option has a length field (all
 
−
            TCP options defined in the future will have length fields).
 
−
            TCP MUST be prepared to handle an illegal option length
 
−
            (e.g., zero) without crashing; a suggested procedure is to
 
−
            reset the connection and log the reason.
 
−
 
 
−
        4.2.2.6  Maximum Segment Size Option: RFC-793 Section 3.1
 
−
 
 
−
            TCP MUST implement both sending and receiving the Maximum
 
−
            Segment Size option [TCP:4].
 
−
 
 
−
            TCP SHOULD send an MSS (Maximum Segment Size) option in
 
−
            every SYN segment when its receive MSS differs from the
 
−
            default 536, and MAY send it always.
 
−
 
 
−
            If an MSS option is not received at connection setup, TCP
 
−
            MUST assume a default send MSS of 536 (576-40) [TCP:4].
 
−
 
 
−
            The maximum size of a segment that TCP really sends, the
 
−
            "effective send MSS," MUST be the smaller of the send MSS
 
−
            (which reflects the available reassembly buffer size at the
 
−
            remote host) and the largest size permitted by the IP layer:
 
−
 
 
−
              Eff.snd.MSS =
 
−
 
 
−
                  min(SendMSS+20, MMS_S) - TCPhdrsize - IPoptionsize
 
−
 
 
−
            where:
 
−
 
 
−
            *    SendMSS is the MSS value received from the remote host,
 
−
                or the default 536 if no MSS option is received.
 
−
 
 
−
            *    MMS_S is the maximum size for a transport-layer message
 
−
                that TCP may send.
 
−
 
 
−
            *    TCPhdrsize is the size of the TCP header; this is
 
−
                normally 20, but may be larger if TCP options are to be
 
−
                sent.
 
−
 
 
−
            *    IPoptionsize is the size of any IP options that TCP
 
−
                will pass to the IP layer with the current message.
 
−
 
 
−
 
 
−
            The MSS value to be sent in an MSS option must be less than
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 85]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
 
−
 
 
−
 
 
−
            or equal to:
 
−
 
 
−
              MMS_R - 20
 
−
 
 
−
            where MMS_R is the maximum size for a transport-layer
 
−
            message that can be received (and reassembled).  TCP obtains
 
−
            MMS_R and MMS_S from the IP layer; see the generic call
 
−
            GET_MAXSIZES in Section 3.4.
 
−
 
 
−
            DISCUSSION:
 
−
                The choice of TCP segment size has a strong effect on
 
−
                performance.  Larger segments increase throughput by
 
−
                amortizing header size and per-datagram processing
 
−
                overhead over more data bytes; however, if the packet
 
−
                is so large that it causes IP fragmentation, efficiency
 
−
                drops sharply if any fragments are lost [IP:9].
 
−
 
 
−
                Some TCP implementations send an MSS option only if the
 
−
                destination host is on a non-connected network.
 
−
                However, in general the TCP layer may not have the
 
−
                appropriate information to make this decision, so it is
 
−
                preferable to leave to the IP layer the task of
 
−
                determining a suitable MTU for the Internet path.  We
 
−
                therefore recommend that TCP always send the option (if
 
−
                not 536) and that the IP layer determine MMS_R as
 
−
                specified in 3.3.3 and 3.4.  A proposed IP-layer
 
−
                mechanism to measure the MTU would then modify the IP
 
−
                layer without changing TCP.
 
−
 
 
−
        4.2.2.7  TCP Checksum: RFC-793 Section 3.1
 
−
 
 
−
            Unlike the UDP checksum (see Section 4.1.3.4), the TCP
 
−
            checksum is never optional.  The sender MUST generate it and
 
−
            the receiver MUST check it.
 
−
 
 
−
        4.2.2.8  TCP Connection State Diagram: RFC-793 Section 3.2,
 
−
            page 23
 
−
 
 
−
            There are several problems with this diagram:
 
−
 
 
−
            (a)  The arrow from SYN-SENT to SYN-RCVD should be labeled
 
−
                with "snd SYN,ACK", to agree with the text on page 68
 
−
                and with Figure 8.
 
−
 
 
−
            (b)  There could be an arrow from SYN-RCVD state to LISTEN
 
−
                state, conditioned on receiving a RST after a passive
 
−
                open (see text page 70).
 
−
 
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 86]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
 
−
 
 
−
 
 
−
            (c)  It is possible to go directly from FIN-WAIT-1 to the
 
−
                TIME-WAIT state (see page 75 of the spec).
 
−
 
 
−
 
 
−
        4.2.2.9  Initial Sequence Number Selection: RFC-793 Section
 
−
            3.3, page 27
 
−
 
 
−
            A TCP MUST use the specified clock-driven selection of
 
−
            initial sequence numbers.
 
−
 
 
−
        4.2.2.10  Simultaneous Open Attempts: RFC-793 Section 3.4, page
 
−
            32
 
−
 
 
−
            There is an error in Figure 8: the packet on line 7 should
 
−
            be identical to the packet on line 5.
 
−
 
 
−
            A TCP MUST support simultaneous open attempts.
 
−
 
 
−
            DISCUSSION:
 
−
                It sometimes surprises implementors that if two
 
−
                applications attempt to simultaneously connect to each
 
−
                other, only one connection is generated instead of two.
 
−
                This was an intentional design decision; don't try to
 
−
                "fix" it.
 
−
 
 
−
        4.2.2.11  Recovery from Old Duplicate SYN: RFC-793 Section 3.4,
 
−
            page 33
 
−
 
 
−
            Note that a TCP implementation MUST keep track of whether a
 
−
            connection has reached SYN_RCVD state as the result of a
 
−
            passive OPEN or an active OPEN.
 
−
 
 
−
        4.2.2.12  RST Segment: RFC-793 Section 3.4
 
−
 
 
−
            A TCP SHOULD allow a received RST segment to include data.
 
−
 
 
−
            DISCUSSION
 
−
                It has been suggested that a RST segment could contain
 
−
                ASCII text that encoded and explained the cause of the
 
−
                RST.  No standard has yet been established for such
 
−
                data.
 
−
 
 
−
        4.2.2.13  Closing a Connection: RFC-793 Section 3.5
 
−
 
 
−
            A TCP connection may terminate in two ways: (1) the normal
 
−
            TCP close sequence using a FIN handshake, and (2) an "abort"
 
−
            in which one or more RST segments are sent and the
 
−
            connection state is immediately discarded.  If a TCP
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 87]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
 
−
 
 
−
 
 
−
            connection is closed by the remote site, the local
 
−
            application MUST be informed whether it closed normally or
 
−
            was aborted.
 
−
 
 
−
            The normal TCP close sequence delivers buffered data
 
−
            reliably in both directions.  Since the two directions of a
 
−
            TCP connection are closed independently, it is possible for
 
−
            a connection to be "half closed," i.e., closed in only one
 
−
            direction, and a host is permitted to continue sending data
 
−
            in the open direction on a half-closed connection.
 
−
 
 
−
            A host MAY implement a "half-duplex" TCP close sequence, so
 
−
            that an application that has called CLOSE cannot continue to
 
−
            read data from the connection.  If such a host issues a
 
−
            CLOSE call while received data is still pending in TCP, or
 
−
            if new data is received after CLOSE is called, its TCP
 
−
            SHOULD send a RST to show that data was lost.
 
−
 
 
−
            When a connection is closed actively, it MUST linger in
 
−
            TIME-WAIT state for a time 2xMSL (Maximum Segment Lifetime).
 
−
            However, it MAY accept a new SYN from the remote TCP to
 
−
            reopen the connection directly from TIME-WAIT state, if it:
 
−
 
 
−
            (1)  assigns its initial sequence number for the new
 
−
                connection to be larger than the largest sequence
 
−
                number it used on the previous connection incarnation,
 
−
                and
 
−
 
 
−
            (2)  returns to TIME-WAIT state if the SYN turns out to be
 
−
                an old duplicate.
 
−
 
 
−
 
 
−
            DISCUSSION:
 
−
                TCP's full-duplex data-preserving close is a feature
 
−
                that is not included in the analogous ISO transport
 
−
                protocol TP4.
 
−
 
 
−
                Some systems have not implemented half-closed
 
−
                connections, presumably because they do not fit into
 
−
                the I/O model of their particular operating system.  On
 
−
                these systems, once an application has called CLOSE, it
 
−
                can no longer read input data from the connection; this
 
−
                is referred to as a "half-duplex" TCP close sequence.
 
−
 
 
−
                The graceful close algorithm of TCP requires that the
 
−
                connection state remain defined on (at least)  one end
 
−
                of the connection, for a timeout period of 2xMSL, i.e.,
 
−
                4 minutes.  During this period, the (remote socket,
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 88]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
 
−
 
 
−
 
 
−
                local socket) pair that defines the connection is busy
 
−
                and cannot be reused.  To shorten the time that a given
 
−
                port pair is tied up, some TCPs allow a new SYN to be
 
−
                accepted in TIME-WAIT state.
 
−
 
 
−
        4.2.2.14  Data Communication: RFC-793 Section 3.7, page 40
 
−
 
 
−
            Since RFC-793 was written, there has been extensive work on
 
−
            TCP algorithms to achieve efficient data communication.
 
−
            Later sections of the present document describe required and
 
−
            recommended TCP algorithms to determine when to send data
 
−
            (Section 4.2.3.4), when to send an acknowledgment (Section
 
−
            4.2.3.2), and when to update the window (Section 4.2.3.3).
 
−
 
 
−
            DISCUSSION:
 
−
                One important performance issue is "Silly Window
 
−
                Syndrome" or "SWS" [TCP:5], a stable pattern of small
 
−
                incremental window movements resulting in extremely
 
−
                poor TCP performance.  Algorithms to avoid SWS are
 
−
                described below for both the sending side (Section
 
−
                4.2.3.4) and the receiving side (Section 4.2.3.3).
 
−
 
 
−
                In brief, SWS is caused by the receiver advancing the
 
−
                right window edge whenever it has any new buffer space
 
−
                available to receive data and by the sender using any
 
−
                incremental window, no matter how small, to send more
 
−
                data [TCP:5].  The result can be a stable pattern of
 
−
                sending tiny data segments, even though both sender and
 
−
                receiver have a large total buffer space for the
 
−
                connection.  SWS can only occur during the transmission
 
−
                of a large amount of data; if the connection goes
 
−
                quiescent, the problem will disappear.  It is caused by
 
−
                typical straightforward implementation of window
 
−
                management, but the sender and receiver algorithms
 
−
                given below will avoid it.
 
−
 
 
−
                Another important TCP performance issue is that some
 
−
                applications, especially remote login to character-at-
 
−
                a-time hosts, tend to send streams of one-octet data
 
−
                segments.  To avoid deadlocks, every TCP SEND call from
 
−
                such applications must be "pushed", either explicitly
 
−
                by the application or else implicitly by TCP.  The
 
−
                result may be a stream of TCP segments that contain one
 
−
                data octet each, which makes very inefficient use of
 
−
                the Internet and contributes to Internet congestion.
 
−
                The Nagle Algorithm described in Section 4.2.3.4
 
−
                provides a simple and effective solution to this
 
−
                problem.  It does have the effect of clumping
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 89]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
 
−
 
 
−
 
 
−
                characters over Telnet connections; this may initially
 
−
                surprise users accustomed to single-character echo, but
 
−
                user acceptance has not been a problem.
 
−
 
 
−
                Note that the Nagle algorithm and the send SWS
 
−
                avoidance algorithm play complementary roles in
 
−
                improving performance.  The Nagle algorithm discourages
 
−
                sending tiny segments when the data to be sent
 
−
                increases in small increments, while the SWS avoidance
 
−
                algorithm discourages small segments resulting from the
 
−
                right window edge advancing in small increments.
 
−
 
 
−
                A careless implementation can send two or more
 
−
                acknowledgment segments per data segment received.  For
 
−
                example, suppose the receiver acknowledges every data
 
−
                segment immediately.  When the application program
 
−
                subsequently consumes the data and increases the
 
−
                available receive buffer space again, the receiver may
 
−
                send a second acknowledgment segment to update the
 
−
                window at the sender.  The extreme case occurs with
 
−
                single-character segments on TCP connections using the
 
−
                Telnet protocol for remote login service.  Some
 
−
                implementations have been observed in which each
 
−
                incoming 1-character segment generates three return
 
−
                segments: (1) the acknowledgment, (2) a one byte
 
−
                increase in the window, and (3) the echoed character,
 
−
                respectively.
 
−
 
 
−
        4.2.2.15  Retransmission Timeout: RFC-793 Section 3.7, page 41
 
−
 
 
−
            The algorithm suggested in RFC-793 for calculating the
 
−
            retransmission timeout is now known to be inadequate; see
 
−
            Section 4.2.3.1 below.
 
−
 
 
−
            Recent work by Jacobson [TCP:7] on Internet congestion and
 
−
            TCP retransmission stability has produced a transmission
 
−
            algorithm combining "slow start" with "congestion
 
−
            avoidance".  A TCP MUST implement this algorithm.
 
−
 
 
−
            If a retransmitted packet is identical to the original
 
−
            packet (which implies not only that the data boundaries have
 
−
            not changed, but also that the window and acknowledgment
 
−
            fields of the header have not changed), then the same IP
 
−
            Identification field MAY be used (see Section 3.2.1.5).
 
−
 
 
−
            IMPLEMENTATION:
 
−
                Some TCP implementors have chosen to "packetize" the
 
−
                data stream, i.e., to pick segment boundaries when
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 90]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
 
−
 
 
−
 
 
−
                segments are originally sent and to queue these
 
−
                segments in a "retransmission queue" until they are
 
−
                acknowledged.  Another design (which may be simpler) is
 
−
                to defer packetizing until each time data is
 
−
                transmitted or retransmitted, so there will be no
 
−
                segment retransmission queue.
 
−
 
 
−
                In an implementation with a segment retransmission
 
−
                queue, TCP performance may be enhanced by repacketizing
 
−
                the segments awaiting acknowledgment when the first
 
−
                retransmission timeout occurs.  That is, the
 
−
                outstanding segments that fitted would be combined into
 
−
                one maximum-sized segment, with a new IP Identification
 
−
                value.  The TCP would then retain this combined segment
 
−
                in the retransmit queue until it was acknowledged.
 
−
                However, if the first two segments in the
 
−
                retransmission queue totalled more than one maximum-
 
−
                sized segment, the TCP would retransmit only the first
 
−
                segment using the original IP Identification field.
 
−
 
 
−
        4.2.2.16  Managing the Window: RFC-793 Section 3.7, page 41
 
−
 
 
−
            A TCP receiver SHOULD NOT shrink the window, i.e., move the
 
−
            right window edge to the left.  However, a sending TCP MUST
 
−
            be robust against window shrinking, which may cause the
 
−
            "useable window" (see Section 4.2.3.4) to become negative.
 
−
 
 
−
            If this happens, the sender SHOULD NOT send new data, but
 
−
            SHOULD retransmit normally the old unacknowledged data
 
−
            between SND.UNA and SND.UNA+SND.WND.  The sender MAY also
 
−
            retransmit old data beyond SND.UNA+SND.WND, but SHOULD NOT
 
−
            time out the connection if data beyond the right window edge
 
−
            is not acknowledged.  If the window shrinks to zero, the TCP
 
−
            MUST probe it in the standard way (see next Section).
 
−
 
 
−
            DISCUSSION:
 
−
                Many TCP implementations become confused if the window
 
−
                shrinks from the right after data has been sent into a
 
−
                larger window.  Note that TCP has a heuristic to select
 
−
                the latest window update despite possible datagram
 
−
                reordering; as a result, it may ignore a window update
 
−
                with a smaller window than previously offered if
 
−
                neither the sequence number nor the acknowledgment
 
−
                number is increased.
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 91]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
 
−
 
 
−
 
 
−
        4.2.2.17  Probing Zero Windows: RFC-793 Section 3.7, page 42
 
−
 
 
−
            Probing of zero (offered) windows MUST be supported.
 
−
 
 
−
            A TCP MAY keep its offered receive window closed
 
−
            indefinitely.  As long as the receiving TCP continues to
 
−
            send acknowledgments in response to the probe segments, the
 
−
            sending TCP MUST allow the connection to stay open.
 
−
 
 
−
            DISCUSSION:
 
−
                It is extremely important to remember that ACK
 
−
                (acknowledgment) segments that contain no data are not
 
−
                reliably transmitted by TCP.  If zero window probing is
 
−
                not supported, a connection may hang forever when an
 
−
                ACK segment that re-opens the window is lost.
 
−
 
 
−
                The delay in opening a zero window generally occurs
 
−
                when the receiving application stops taking data from
 
−
                its TCP.  For example, consider a printer daemon
 
−
                application, stopped because the printer ran out of
 
−
                paper.
 
−
 
 
−
            The transmitting host SHOULD send the first zero-window
 
−
            probe when a zero window has existed for the retransmission
 
−
            timeout period (see Section 4.2.2.15), and SHOULD increase
 
−
            exponentially the interval between successive probes.
 
−
 
 
−
            DISCUSSION:
 
−
                This procedure minimizes delay if the zero-window
 
−
                condition is due to a lost ACK segment containing a
 
−
                window-opening update.  Exponential backoff is
 
−
                recommended, possibly with some maximum interval not
 
−
                specified here.  This procedure is similar to that of
 
−
                the retransmission algorithm, and it may be possible to
 
−
                combine the two procedures in the implementation.
 
−
 
 
−
        4.2.2.18  Passive OPEN Calls:  RFC-793 Section 3.8
 
−
 
 
−
            Every passive OPEN call either creates a new connection
 
−
            record in LISTEN state, or it returns an error; it MUST NOT
 
−
            affect any previously created connection record.
 
−
 
 
−
            A TCP that supports multiple concurrent users MUST provide
 
−
            an OPEN call that will functionally allow an application to
 
−
            LISTEN on a port while a connection block with the same
 
−
            local port is in SYN-SENT or SYN-RECEIVED state.
 
−
 
 
−
            DISCUSSION:
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 92]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
 
−
 
 
−
 
 
−
                Some applications (e.g., SMTP servers) may need to
 
−
                handle multiple connection attempts at about the same
 
−
                time.  The probability of a connection attempt failing
 
−
                is reduced by giving the application some means of
 
−
                listening for a new connection at the same time that an
 
−
                earlier connection attempt is going through the three-
 
−
                way handshake.
 
−
 
 
−
            IMPLEMENTATION:
 
−
                Acceptable implementations of concurrent opens may
 
−
                permit multiple passive OPEN calls, or they may allow
 
−
                "cloning" of LISTEN-state connections from a single
 
−
                passive OPEN call.
 
−
 
 
−
        4.2.2.19  Time to Live: RFC-793 Section 3.9, page 52
 
−
 
 
−
            RFC-793 specified that TCP was to request the IP layer to
 
−
            send TCP segments with TTL = 60.  This is obsolete; the TTL
 
−
            value used to send TCP segments MUST be configurable.  See
 
−
            Section 3.2.1.7 for discussion.
 
−
 
 
−
        4.2.2.20  Event Processing: RFC-793 Section 3.9
 
−
 
 
−
            While it is not strictly required, a TCP SHOULD be capable
 
−
            of queueing out-of-order TCP segments.  Change the "may" in
 
−
            the last sentence of the first paragraph on page 70 to
 
−
            "should".
 
−
 
 
−
            DISCUSSION:
 
−
                Some small-host implementations have omitted segment
 
−
                queueing because of limited buffer space.  This
 
−
                omission may be expected to adversely affect TCP
 
−
                throughput, since loss of a single segment causes all
 
−
                later segments to appear to be "out of sequence".
 
−
 
 
−
            In general, the processing of received segments MUST be
 
−
            implemented to aggregate ACK segments whenever possible.
 
−
            For example, if the TCP is processing a series of queued
 
−
            segments, it MUST process them all before sending any ACK
 
−
            segments.
 
−
 
 
−
            Here are some detailed error corrections and notes on the
 
−
            Event Processing section of RFC-793.
 
−
 
 
−
            (a)  CLOSE Call, CLOSE-WAIT state, p. 61: enter LAST-ACK
 
−
                state, not CLOSING.
 
−
 
 
−
            (b)  LISTEN state, check for SYN (pp. 65, 66): With a SYN
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 93]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
 
−
 
 
−
 
 
−
                bit, if the security/compartment or the precedence is
 
−
                wrong for the segment, a reset is sent.  The wrong form
 
−
                of reset is shown in the text; it should be:
 
−
 
 
−
                  <SEQ=0><ACK=SEG.SEQ+SEG.LEN><CTL=RST,ACK>
 
−
 
 
−
 
 
−
            (c)  SYN-SENT state, Check for SYN, p. 68: When the
 
−
                connection enters ESTABLISHED state, the following
 
−
                variables must be set:
 
−
                    SND.WND <- SEG.WND
 
−
                    SND.WL1 <- SEG.SEQ
 
−
                    SND.WL2 <- SEG.ACK
 
−
 
 
−
 
 
−
            (d)  Check security and precedence, p. 71: The first heading
 
−
                "ESTABLISHED STATE" should really be a list of all
 
−
                states other than SYN-RECEIVED: ESTABLISHED, FIN-WAIT-
 
−
                1, FIN-WAIT-2, CLOSE-WAIT, CLOSING, LAST-ACK, and
 
−
                TIME-WAIT.
 
−
 
 
−
            (e)  Check SYN bit, p. 71:  "In SYN-RECEIVED state and if
 
−
                the connection was initiated with a passive OPEN, then
 
−
                return this connection to the LISTEN state and return.
 
−
                Otherwise...".
 
−
 
 
−
            (f)  Check ACK field, SYN-RECEIVED state, p. 72: When the
 
−
                connection enters ESTABLISHED state, the variables
 
−
                listed in (c) must be set.
 
−
 
 
−
            (g)  Check ACK field, ESTABLISHED state, p. 72: The ACK is a
 
−
                duplicate if SEG.ACK =< SND.UNA (the = was omitted).
 
−
                Similarly, the window should be updated if: SND.UNA =<
 
−
                SEG.ACK =< SND.NXT.
 
−
 
 
−
            (h)  USER TIMEOUT, p. 77:
 
−
 
 
−
                It would be better to notify the application of the
 
−
                timeout rather than letting TCP force the connection
 
−
                closed.  However, see also Section 4.2.3.5.
 
−
 
 
−
 
 
−
        4.2.2.21  Acknowledging Queued Segments: RFC-793 Section 3.9
 
−
 
 
−
            A TCP MAY send an ACK segment acknowledging RCV.NXT when a
 
−
            valid segment arrives that is in the window but not at the
 
−
            left window edge.
 
−
 
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 94]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
 
−
 
 
−
 
 
−
            DISCUSSION:
 
−
                RFC-793 (see page 74) was ambiguous about whether or
 
−
                not an ACK segment should be sent when an out-of-order
 
−
                segment was received, i.e., when SEG.SEQ was unequal to
 
−
                RCV.NXT.
 
−
 
 
−
                One reason for ACKing out-of-order segments might be to
 
−
                support an experimental algorithm known as "fast
 
−
                retransmit".  With this algorithm, the sender uses the
 
−
                "redundant" ACK's to deduce that a segment has been
 
−
                lost before the retransmission timer has expired.  It
 
−
                counts the number of times an ACK has been received
 
−
                with the same value of SEG.ACK and with the same right
 
−
                window edge.  If more than a threshold number of such
 
−
                ACK's is received, then the segment containing the
 
−
                octets starting at SEG.ACK is assumed to have been lost
 
−
                and is retransmitted, without awaiting a timeout.  The
 
−
                threshold is chosen to compensate for the maximum
 
−
                likely segment reordering in the Internet.  There is
 
−
                not yet enough experience with the fast retransmit
 
−
                algorithm to determine how useful it is.
 
−
 
 
−
      4.2.3  SPECIFIC ISSUES
 
−
 
 
−
        4.2.3.1  Retransmission Timeout Calculation
 
−
 
 
−
            A host TCP MUST implement Karn's algorithm and Jacobson's
 
−
            algorithm for computing the retransmission timeout ("RTO").
 
−
 
 
−
            o    Jacobson's algorithm for computing the smoothed round-
 
−
                trip ("RTT") time incorporates a simple measure of the
 
−
                variance [TCP:7].
 
−
 
 
−
            o    Karn's algorithm for selecting RTT measurements ensures
 
−
                that ambiguous round-trip times will not corrupt the
 
−
                calculation of the smoothed round-trip time [TCP:6].
 
−
 
 
−
            This implementation also MUST include "exponential backoff"
 
−
            for successive RTO values for the same segment.
 
−
            Retransmission of SYN segments SHOULD use the same algorithm
 
−
            as data segments.
 
−
 
 
−
            DISCUSSION:
 
−
                There were two known problems with the RTO calculations
 
−
                specified in RFC-793.  First, the accurate measurement
 
−
                of RTTs is difficult when there are retransmissions.
 
−
                Second, the algorithm to compute the smoothed round-
 
−
                trip time is inadequate [TCP:7], because it incorrectly
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 95]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
 
−
 
 
−
 
 
−
                assumed that the variance in RTT values would be small
 
−
                and constant.  These problems were solved by Karn's and
 
−
                Jacobson's algorithm, respectively.
 
−
 
 
−
                The performance increase resulting from the use of
 
−
                these improvements varies from noticeable to dramatic.
 
−
                Jacobson's algorithm for incorporating the measured RTT
 
−
                variance is especially important on a low-speed link,
 
−
                where the natural variation of packet sizes causes a
 
−
                large variation in RTT.  One vendor found link
 
−
                utilization on a 9.6kb line went from 10% to 90% as a
 
−
                result of implementing Jacobson's variance algorithm in
 
−
                TCP.
 
−
 
 
−
            The following values SHOULD be used to initialize the
 
−
            estimation parameters for a new connection:
 
−
 
 
−
            (a)  RTT = 0 seconds.
 
−
 
 
−
            (b)  RTO = 3 seconds.  (The smoothed variance is to be
 
−
                initialized to the value that will result in this RTO).
 
−
 
 
−
            The recommended upper and lower bounds on the RTO are known
 
−
            to be inadequate on large internets.  The lower bound SHOULD
 
−
            be measured in fractions of a second (to accommodate high
 
−
            speed LANs) and the upper bound should be 2*MSL, i.e., 240
 
−
            seconds.
 
−
 
 
−
            DISCUSSION:
 
−
                Experience has shown that these initialization values
 
−
                are reasonable, and that in any case the Karn and
 
−
                Jacobson algorithms make TCP behavior reasonably
 
−
                insensitive to the initial parameter choices.
 
−
 
 
−
        4.2.3.2  When to Send an ACK Segment
 
−
 
 
−
            A host that is receiving a stream of TCP data segments can
 
−
            increase efficiency in both the Internet and the hosts by
 
−
            sending fewer than one ACK (acknowledgment) segment per data
 
−
            segment received; this is known as a "delayed ACK" [TCP:5].
 
−
 
 
−
            A TCP SHOULD implement a delayed ACK, but an ACK should not
 
−
            be excessively delayed; in particular, the delay MUST be
 
−
            less than 0.5 seconds, and in a stream of full-sized
 
−
            segments there SHOULD be an ACK for at least every second
 
−
            segment.
 
−
 
 
−
            DISCUSSION:
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 96]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
 
−
 
 
−
 
 
−
                A delayed ACK gives the application an opportunity to
 
−
                update the window and perhaps to send an immediate
 
−
                response.  In particular, in the case of character-mode
 
−
                remote login, a delayed ACK can reduce the number of
 
−
                segments sent by the server by a factor of 3 (ACK,
 
−
                window update, and echo character all combined in one
 
−
                segment).
 
−
 
 
−
                In addition, on some large multi-user hosts, a delayed
 
−
                ACK can substantially reduce protocol processing
 
−
                overhead by reducing the total number of packets to be
 
−
                processed [TCP:5].  However, excessive delays on ACK's
 
−
                can disturb the round-trip timing and packet "clocking"
 
−
                algorithms [TCP:7].
 
−
 
 
−
        4.2.3.3  When to Send a Window Update
 
−
 
 
−
            A TCP MUST include a SWS avoidance algorithm in the receiver
 
−
            [TCP:5].
 
−
 
 
−
            IMPLEMENTATION:
 
−
                The receiver's SWS avoidance algorithm determines when
 
−
                the right window edge may be advanced; this is
 
−
                customarily known as "updating the window".  This
 
−
                algorithm combines with the delayed ACK algorithm (see
 
−
                Section 4.2.3.2) to determine when an ACK segment
 
−
                containing the current window will really be sent to
 
−
                the receiver.  We use the notation of RFC-793; see
 
−
                Figures 4 and 5 in that document.
 
−
 
 
−
                The solution to receiver SWS is to avoid advancing the
 
−
                right window edge RCV.NXT+RCV.WND in small increments,
 
−
                even if data is received from the network in small
 
−
                segments.
 
−
 
 
−
                Suppose the total receive buffer space is RCV.BUFF.  At
 
−
                any given moment, RCV.USER octets of this total may be
 
−
                tied up with data that has been received and
 
−
                acknowledged but which the user process has not yet
 
−
                consumed.  When the connection is quiescent, RCV.WND =
 
−
                RCV.BUFF and RCV.USER = 0.
 
−
 
 
−
                Keeping the right window edge fixed as data arrives and
 
−
                is acknowledged requires that the receiver offer less
 
−
                than its full buffer space, i.e., the receiver must
 
−
                specify a RCV.WND that keeps RCV.NXT+RCV.WND constant
 
−
                as RCV.NXT increases.  Thus, the total buffer space
 
−
                RCV.BUFF is generally divided into three parts:
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 97]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
 
−
 
 
−
 
 
−
 
 
−
                |<------- RCV.BUFF ---------------->|
 
−
                      1            2            3
 
−
            ----|---------|------------------|------|----
 
−
                        RCV.NXT              ^
 
−
                                          (Fixed)
 
−
 
 
−
            1 - RCV.USER =  data received but not yet consumed;
 
−
            2 - RCV.WND =  space advertised to sender;
 
−
            3 - Reduction = space available but not yet
 
−
                            advertised.
 
−
 
 
−
 
 
−
                The suggested SWS avoidance algorithm for the receiver
 
−
                is to keep RCV.NXT+RCV.WND fixed until the reduction
 
−
                satisfies:
 
−
 
 
−
                      RCV.BUFF - RCV.USER - RCV.WND  >=
 
−
 
 
−
                            min( Fr * RCV.BUFF, Eff.snd.MSS )
 
−
 
 
−
                where Fr is a fraction whose recommended value is 1/2,
 
−
                and Eff.snd.MSS is the effective send MSS for the
 
−
                connection (see Section 4.2.2.6).  When the inequality
 
−
                is satisfied, RCV.WND is set to RCV.BUFF-RCV.USER.
 
−
 
 
−
                Note that the general effect of this algorithm is to
 
−
                advance RCV.WND in increments of Eff.snd.MSS (for
 
−
                realistic receive buffers:  Eff.snd.MSS < RCV.BUFF/2).
 
−
                Note also that the receiver must use its own
 
−
                Eff.snd.MSS, assuming it is the same as the sender's.
 
−
 
 
−
        4.2.3.4  When to Send Data
 
−
 
 
−
            A TCP MUST include a SWS avoidance algorithm in the sender.
 
−
 
 
−
            A TCP SHOULD implement the Nagle Algorithm [TCP:9] to
 
−
            coalesce short segments.  However, there MUST be a way for
 
−
            an application to disable the Nagle algorithm on an
 
−
            individual connection.  In all cases, sending data is also
 
−
            subject to the limitation imposed by the Slow Start
 
−
            algorithm (Section 4.2.2.15).
 
−
 
 
−
            DISCUSSION:
 
−
                The Nagle algorithm is generally as follows:
 
−
 
 
−
                      If there is unacknowledged data (i.e., SND.NXT >
 
−
                      SND.UNA), then the sending TCP buffers all user
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 98]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
 
−
 
 
−
 
 
−
                      data (regardless of the PSH bit), until the
 
−
                      outstanding data has been acknowledged or until
 
−
                      the TCP can send a full-sized segment (Eff.snd.MSS
 
−
                      bytes; see Section 4.2.2.6).
 
−
 
 
−
                Some applications (e.g., real-time display window
 
−
                updates) require that the Nagle algorithm be turned
 
−
                off, so small data segments can be streamed out at the
 
−
                maximum rate.
 
−
 
 
−
            IMPLEMENTATION:
 
−
                The sender's SWS avoidance algorithm is more difficult
 
−
                than the receivers's, because the sender does not know
 
−
                (directly) the receiver's total buffer space RCV.BUFF.
 
−
                An approach which has been found to work well is for
 
−
                the sender to calculate Max(SND.WND), the maximum send
 
−
                window it has seen so far on the connection, and to use
 
−
                this value as an estimate of RCV.BUFF.  Unfortunately,
 
−
                this can only be an estimate; the receiver may at any
 
−
                time reduce the size of RCV.BUFF.  To avoid a resulting
 
−
                deadlock, it is necessary to have a timeout to force
 
−
                transmission of data, overriding the SWS avoidance
 
−
                algorithm.  In practice, this timeout should seldom
 
−
                occur.
 
−
 
 
−
                The "useable window" [TCP:5] is:
 
−
 
 
−
                      U = SND.UNA + SND.WND - SND.NXT
 
−
 
 
−
                i.e., the offered window less the amount of data sent
 
−
                but not acknowledged.  If D is the amount of data
 
−
                queued in the sending TCP but not yet sent, then the
 
−
                following set of rules is recommended.
 
−
 
 
−
                Send data:
 
−
 
 
−
                (1)  if a maximum-sized segment can be sent, i.e, if:
 
−
 
 
−
                          min(D,U) >= Eff.snd.MSS;
 
−
 
 
−
 
 
−
                (2)  or if the data is pushed and all queued data can
 
−
                      be sent now, i.e., if:
 
−
 
 
−
                          [SND.NXT = SND.UNA and] PUSHED and D <= U
 
−
 
 
−
                      (the bracketed condition is imposed by the Nagle
 
−
                      algorithm);
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                                [Page 99]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
 
−
 
 
−
 
 
−
                (3)  or if at least a fraction Fs of the maximum window
 
−
                      can be sent, i.e., if:
 
−
 
 
−
                          [SND.NXT = SND.UNA and]
 
−
 
 
−
                                  min(D.U) >= Fs * Max(SND.WND);
 
−
 
 
−
 
 
−
                (4)  or if data is PUSHed and the override timeout
 
−
                      occurs.
 
−
 
 
−
                Here Fs is a fraction whose recommended value is 1/2.
 
−
                The override timeout should be in the range 0.1 - 1.0
 
−
                seconds.  It may be convenient to combine this timer
 
−
                with the timer used to probe zero windows (Section
 
−
                4.2.2.17).
 
−
 
 
−
                Finally, note that the SWS avoidance algorithm just
 
−
                specified is to be used instead of the sender-side
 
−
                algorithm contained in [TCP:5].
 
−
 
 
−
        4.2.3.5  TCP Connection Failures
 
−
 
 
−
            Excessive retransmission of the same segment by TCP
 
−
            indicates some failure of the remote host or the Internet
 
−
            path.  This failure may be of short or long duration.  The
 
−
            following procedure MUST be used to handle excessive
 
−
            retransmissions of data segments [IP:11]:
 
−
 
 
−
            (a)  There are two thresholds R1 and R2 measuring the amount
 
−
                of retransmission that has occurred for the same
 
−
                segment.  R1 and R2 might be measured in time units or
 
−
                as a count of retransmissions.
 
−
 
 
−
            (b)  When the number of transmissions of the same segment
 
−
                reaches or exceeds threshold R1, pass negative advice
 
−
                (see Section 3.3.1.4) to the IP layer, to trigger
 
−
                dead-gateway diagnosis.
 
−
 
 
−
            (c)  When the number of transmissions of the same segment
 
−
                reaches a threshold R2 greater than R1, close the
 
−
                connection.
 
−
 
 
−
            (d)  An application MUST be able to set the value for R2 for
 
−
                a particular connection.  For example, an interactive
 
−
                application might set R2 to "infinity," giving the user
 
−
                control over when to disconnect.
 
−
 
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                              [Page 100]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
 
−
 
 
−
 
 
−
            (d)  TCP SHOULD inform the application of the delivery
 
−
                problem (unless such information has been disabled by
 
−
                the application; see Section 4.2.4.1), when R1 is
 
−
                reached and before R2.  This will allow a remote login
 
−
                (User Telnet) application program to inform the user,
 
−
                for example.
 
−
 
 
−
            The value of R1 SHOULD correspond to at least 3
 
−
            retransmissions, at the current RTO.  The value of R2 SHOULD
 
−
            correspond to at least 100 seconds.
 
−
 
 
−
            An attempt to open a TCP connection could fail with
 
−
            excessive retransmissions of the SYN segment or by receipt
 
−
            of a RST segment or an ICMP Port Unreachable.  SYN
 
−
            retransmissions MUST be handled in the general way just
 
−
            described for data retransmissions, including notification
 
−
            of the application layer.
 
−
 
 
−
            However, the values of R1 and R2 may be different for SYN
 
−
            and data segments.  In particular, R2 for a SYN segment MUST
 
−
            be set large enough to provide retransmission of the segment
 
−
            for at least 3 minutes.  The application can close the
 
−
            connection (i.e., give up on the open attempt) sooner, of
 
−
            course.
 
−
 
 
−
            DISCUSSION:
 
−
                Some Internet paths have significant setup times, and
 
−
                the number of such paths is likely to increase in the
 
−
                future.
 
−
 
 
−
        4.2.3.6  TCP Keep-Alives
 
−
 
 
−
            Implementors MAY include "keep-alives" in their TCP
 
−
            implementations, although this practice is not universally
 
−
            accepted.  If keep-alives are included, the application MUST
 
−
            be able to turn them on or off for each TCP connection, and
 
−
            they MUST default to off.
 
−
 
 
−
            Keep-alive packets MUST only be sent when no data or
 
−
            acknowledgement packets have been received for the
 
−
            connection within an interval.  This interval MUST be
 
−
            configurable and MUST default to no less than two hours.
 
−
 
 
−
            It is extremely important to remember that ACK segments that
 
−
            contain no data are not reliably transmitted by TCP.
 
−
            Consequently, if a keep-alive mechanism is implemented it
 
−
            MUST NOT interpret failure to respond to any specific probe
 
−
            as a dead connection.
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                              [Page 101]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
 
−
 
 
−
 
 
−
            An implementation SHOULD send a keep-alive segment with no
 
−
            data; however, it MAY be configurable to send a keep-alive
 
−
            segment containing one garbage octet, for compatibility with
 
−
            erroneous TCP implementations.
 
−
 
 
−
            DISCUSSION:
 
−
                A "keep-alive" mechanism periodically probes the other
 
−
                end of a connection when the connection is otherwise
 
−
                idle, even when there is no data to be sent.  The TCP
 
−
                specification does not include a keep-alive mechanism
 
−
                because it could:  (1) cause perfectly good connections
 
−
                to break during transient Internet failures; (2)
 
−
                consume unnecessary bandwidth ("if no one is using the
 
−
                connection, who cares if it is still good?"); and (3)
 
−
                cost money for an Internet path that charges for
 
−
                packets.
 
−
 
 
−
                Some TCP implementations, however, have included a
 
−
                keep-alive mechanism.  To confirm that an idle
 
−
                connection is still active, these implementations send
 
−
                a probe segment designed to elicit a response from the
 
−
                peer TCP.  Such a segment generally contains SEG.SEQ =
 
−
                SND.NXT-1 and may or may not contain one garbage octet
 
−
                of data.  Note that on a quiet connection SND.NXT =
 
−
                RCV.NXT, so that this SEG.SEQ will be outside the
 
−
                window.  Therefore, the probe causes the receiver to
 
−
                return an acknowledgment segment, confirming that the
 
−
                connection is still live.  If the peer has dropped the
 
−
                connection due to a network partition or a crash, it
 
−
                will respond with a RST instead of an acknowledgment
 
−
                segment.
 
−
 
 
−
                Unfortunately, some misbehaved TCP implementations fail
 
−
                to respond to a segment with SEG.SEQ = SND.NXT-1 unless
 
−
                the segment contains data.  Alternatively, an
 
−
                implementation could determine whether a peer responded
 
−
                correctly to keep-alive packets with no garbage data
 
−
                octet.
 
−
 
 
−
                A TCP keep-alive mechanism should only be invoked in
 
−
                server applications that might otherwise hang
 
−
                indefinitely and consume resources unnecessarily if a
 
−
                client crashes or aborts a connection during a network
 
−
                failure.
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                              [Page 102]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
 
−
 
 
−
 
 
−
        4.2.3.7  TCP Multihoming
 
−
 
 
−
            If an application on a multihomed host does not specify the
 
−
            local IP address when actively opening a TCP connection,
 
−
            then the TCP MUST ask the IP layer to select a local IP
 
−
            address before sending the (first) SYN.  See the function
 
−
            GET_SRCADDR() in Section 3.4.
 
−
 
 
−
            At all other times, a previous segment has either been sent
 
−
            or received on this connection, and TCP MUST use the same
 
−
            local address is used that was used in those previous
 
−
            segments.
 
−
 
 
−
        4.2.3.8  IP Options
 
−
 
 
−
            When received options are passed up to TCP from the IP
 
−
            layer, TCP MUST ignore options that it does not understand.
 
−
 
 
−
            A TCP MAY support the Time Stamp and Record Route options.
 
−
 
 
−
            An application MUST be able to specify a source route when
 
−
            it actively opens a TCP connection, and this MUST take
 
−
            precedence over a source route received in a datagram.
 
−
 
 
−
            When a TCP connection is OPENed passively and a packet
 
−
            arrives with a completed IP Source Route option (containing
 
−
            a return route), TCP MUST save the return route and use it
 
−
            for all segments sent on this connection.  If a different
 
−
            source route arrives in a later segment, the later
 
−
            definition SHOULD override the earlier one.
 
−
 
 
−
        4.2.3.9  ICMP Messages
 
−
 
 
−
            TCP MUST act on an ICMP error message passed up from the IP
 
−
            layer, directing it to the connection that created the
 
−
            error.  The necessary demultiplexing information can be
 
−
            found in the IP header contained within the ICMP message.
 
−
 
 
−
            o    Source Quench
 
−
 
 
−
                TCP MUST react to a Source Quench by slowing
 
−
                transmission on the connection.  The RECOMMENDED
 
−
                procedure is for a Source Quench to trigger a "slow
 
−
                start," as if a retransmission timeout had occurred.
 
−
 
 
−
            o    Destination Unreachable -- codes 0, 1, 5
 
−
 
 
−
                Since these Unreachable messages indicate soft error
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                              [Page 103]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
 
−
 
 
−
 
 
−
                conditions, TCP MUST NOT abort the connection, and it
 
−
                SHOULD make the information available to the
 
−
                application.
 
−
 
 
−
                DISCUSSION:
 
−
                      TCP could report the soft error condition directly
 
−
                      to the application layer with an upcall to the
 
−
                      ERROR_REPORT routine, or it could merely note the
 
−
                      message and report it to the application only when
 
−
                      and if the TCP connection times out.
 
−
 
 
−
            o    Destination Unreachable -- codes 2-4
 
−
 
 
−
                These are hard error conditions, so TCP SHOULD abort
 
−
                the connection.
 
−
 
 
−
            o    Time Exceeded -- codes 0, 1
 
−
 
 
−
                This should be handled the same way as Destination
 
−
                Unreachable codes 0, 1, 5 (see above).
 
−
 
 
−
            o    Parameter Problem
 
−
 
 
−
                This should be handled the same way as Destination
 
−
                Unreachable codes 0, 1, 5 (see above).
 
−
 
 
−
 
 
−
        4.2.3.10  Remote Address Validation
 
−
 
 
−
            A TCP implementation MUST reject as an error a local OPEN
 
−
            call for an invalid remote IP address (e.g., a broadcast or
 
−
            multicast address).
 
−
 
 
−
            An incoming SYN with an invalid source address must be
 
−
            ignored either by TCP or by the IP layer (see Section
 
−
            3.2.1.3).
 
−
 
 
−
            A TCP implementation MUST silently discard an incoming SYN
 
−
            segment that is addressed to a broadcast or multicast
 
−
            address.
 
−
 
 
−
        4.2.3.11  TCP Traffic Patterns
 
−
 
 
−
            IMPLEMENTATION:
 
−
                The TCP protocol specification [TCP:1] gives the
 
−
                implementor much freedom in designing the algorithms
 
−
                that control the message flow over the connection --
 
−
                packetizing, managing the window, sending
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                              [Page 104]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
 
−
 
 
−
 
 
−
                acknowledgments, etc.  These design decisions are
 
−
                difficult because a TCP must adapt to a wide range of
 
−
                traffic patterns.  Experience has shown that a TCP
 
−
                implementor needs to verify the design on two extreme
 
−
                traffic patterns:
 
−
 
 
−
                o    Single-character Segments
 
−
 
 
−
                      Even if the sender is using the Nagle Algorithm,
 
−
                      when a TCP connection carries remote login traffic
 
−
                      across a low-delay LAN the receiver will generally
 
−
                      get a stream of single-character segments.  If
 
−
                      remote terminal echo mode is in effect, the
 
−
                      receiver's system will generally echo each
 
−
                      character as it is received.
 
−
 
 
−
                o    Bulk Transfer
 
−
 
 
−
                      When TCP is used for bulk transfer, the data
 
−
                      stream should be made up (almost) entirely of
 
−
                      segments of the size of the effective MSS.
 
−
                      Although TCP uses a sequence number space with
 
−
                      byte (octet) granularity, in bulk-transfer mode
 
−
                      its operation should be as if TCP used a sequence
 
−
                      space that counted only segments.
 
−
 
 
−
                Experience has furthermore shown that a single TCP can
 
−
                effectively and efficiently handle these two extremes.
 
−
 
 
−
                The most important tool for verifying a new TCP
 
−
                implementation is a packet trace program.  There is a
 
−
                large volume of experience showing the importance of
 
−
                tracing a variety of traffic patterns with other TCP
 
−
                implementations and studying the results carefully.
 
−
 
 
−
 
 
−
        4.2.3.12  Efficiency
 
−
 
 
−
            IMPLEMENTATION:
 
−
                Extensive experience has led to the following
 
−
                suggestions for efficient implementation of TCP:
 
−
 
 
−
                (a)  Don't Copy Data
 
−
 
 
−
                      In bulk data transfer, the primary CPU-intensive
 
−
                      tasks are copying data from one place to another
 
−
                      and checksumming the data.  It is vital to
 
−
                      minimize the number of copies of TCP data.  Since
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                              [Page 105]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
 
−
 
 
−
 
 
−
                      the ultimate speed limitation may be fetching data
 
−
                      across the memory bus, it may be useful to combine
 
−
                      the copy with checksumming, doing both with a
 
−
                      single memory fetch.
 
−
 
 
−
                (b)  Hand-Craft the Checksum Routine
 
−
 
 
−
                      A good TCP checksumming routine is typically two
 
−
                      to five times faster than a simple and direct
 
−
                      implementation of the definition.  Great care and
 
−
                      clever coding are often required and advisable to
 
−
                      make the checksumming code "blazing fast".  See
 
−
                      [TCP:10].
 
−
 
 
−
                (c)  Code for the Common Case
 
−
 
 
−
                      TCP protocol processing can be complicated, but
 
−
                      for most segments there are only a few simple
 
−
                      decisions to be made.  Per-segment processing will
 
−
                      be greatly speeded up by coding the main line to
 
−
                      minimize the number of decisions in the most
 
−
                      common case.
 
−
 
 
−
 
 
−
      4.2.4  TCP/APPLICATION LAYER INTERFACE
 
−
 
 
−
        4.2.4.1  Asynchronous Reports
 
−
 
 
−
            There MUST be a mechanism for reporting soft TCP error
 
−
            conditions to the application.  Generically, we assume this
 
−
            takes the form of an application-supplied ERROR_REPORT
 
−
            routine that may be upcalled [INTRO:7] asynchronously from
 
−
            the transport layer:
 
−
 
 
−
              ERROR_REPORT(local connection name, reason, subreason)
 
−
 
 
−
            The precise encoding of the reason and subreason parameters
 
−
            is not specified here.  However, the conditions that are
 
−
            reported asynchronously to the application MUST include:
 
−
 
 
−
            *    ICMP error message arrived (see 4.2.3.9)
 
−
 
 
−
            *    Excessive retransmissions (see 4.2.3.5)
 
−
 
 
−
            *    Urgent pointer advance (see 4.2.2.4).
 
  
−
            However, an application program that does not want to
+
        The application layer MUST be able to specify the Type-of-
−
            receive such ERROR_REPORT calls SHOULD be able to
+
        Service (TOS) for segments that are sent on a connection.
 +
        It not required, but the application SHOULD be able to
 +
        change the TOS during the connection lifetime.  TCP SHOULD
 +
        pass the current TOS value without change to the IP layer,
 +
        when it sends segments on the connection.
  
 +
        The TOS will be specified independently in each direction on
 +
        the connection, so that the receiver application will
 +
        specify the TOS used for ACK segments.
  
 +
        TCP MAY pass the most recently received TOS up to the
 +
        application.
  
−
Internet Engineering Task Force                              [Page 106]
+
        DISCUSSION
 +
              Some applications (e.g., SMTP) change the nature of
 +
              their communication during the lifetime of a
 +
              connection, and therefore would like to change the TOS
 +
              specification.
  
 +
              Note also that the OPEN call specified in RFC-793
 +
              includes a parameter ("options") in which the caller
 +
              can specify IP options such as source route, record
 +
              route, or timestamp.
  
 +
      4.2.4.3  Flush Call
  
 +
        Some TCP implementations have included a FLUSH call, which
 +
        will empty the TCP send queue of any data for which the user
 +
        has issued SEND calls but which is still to the right of the
 +
        current send window.  That is, it flushes as much queued
 +
        send data as possible without losing sequence number
 +
        synchronization.  This is useful for implementing the "abort
 +
        output" function of Telnet.
  
 
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
 
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
  
 +
      4.2.4.4  Multihoming
  
−
            effectively disable these calls.
+
        The user interface outlined in sections 2.7 and 3.8 of RFC-
 +
        793 needs to be extended for multihoming.  The OPEN call
 +
        MUST have an optional parameter:
  
−
            DISCUSSION:
+
            OPEN( ... [local IP address,] ... )
−
                These error reports generally reflect soft errors that
 
−
                can be ignored without harm by many applications. It
 
−
                has been suggested that these error report calls should
 
−
                default to "disabled," but this is not required.
 
  
−
         4.2.4.2  Type-of-Service
+
         to allow the specification of the local IP address.
  
−
            The application layer MUST be able to specify the Type-of-
+
        DISCUSSION:
−
            Service (TOS) for segments that are sent on a connection.
+
              Some TCP-based applications need to specify the local
−
            It not required, but the application SHOULD be able to
+
              IP address to be used to open a particular connection;
−
            change the TOS during the connection lifetime.  TCP SHOULD
+
              FTP is an example.
−
            pass the current TOS value without change to the IP layer,
 
−
            when it sends segments on the connection.
 
  
−
            The TOS will be specified independently in each direction on
+
         IMPLEMENTATION:
−
            the connection, so that the receiver application will
+
              A passive OPEN call with a specified "local IP address"
−
            specify the TOS used for ACK segments.
+
              parameter will await an incoming connection request to
−
 
+
              that address.  If the parameter is unspecified, a
−
            TCP MAY pass the most recently received TOS up to the
+
              passive OPEN will await an incoming connection request
−
            application.
+
              to any local IP address, and then bind the local IP
−
 
+
              address of the connection to the particular address
−
            DISCUSSION
+
              that is used.
−
                Some applications (e.g., SMTP) change the nature of
 
−
                their communication during the lifetime of a
 
−
                connection, and therefore would like to change the TOS
 
−
                specification.
 
−
 
 
−
                Note also that the OPEN call specified in RFC-793
 
−
                includes a parameter ("options") in which the caller
 
−
                can specify IP options such as source route, record
 
−
                route, or timestamp.
 
−
 
 
−
        4.2.4.3  Flush Call
 
−
 
 
−
            Some TCP implementations have included a FLUSH call, which
 
−
            will empty the TCP send queue of any data for which the user
 
−
            has issued SEND calls but which is still to the right of the
 
−
            current send window.  That is, it flushes as much queued
 
−
            send data as possible without losing sequence number
 
−
            synchronization.  This is useful for implementing the "abort
 
−
            output" function of Telnet.
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                              [Page 107]
 
−
 
 
−
 
 
−
 
 
−
 
 
−
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
 
−
 
 
−
 
 
−
         4.2.4.4  Multihoming
 
−
 
 
−
            The user interface outlined in sections 2.7 and 3.8 of RFC-
 
−
            793 needs to be extended for multihoming.  The OPEN call
 
−
            MUST have an optional parameter:
 
−
 
 
−
                OPEN( ... [local IP address,] ... )
 
−
 
 
−
            to allow the specification of the local IP address.
 
−
 
 
−
            DISCUSSION:
 
−
                Some TCP-based applications need to specify the local
 
−
                IP address to be used to open a particular connection;
 
−
                FTP is an example.
 
−
 
 
−
            IMPLEMENTATION:
 
−
                A passive OPEN call with a specified "local IP address"
 
−
                parameter will await an incoming connection request to
 
−
                that address.  If the parameter is unspecified, a
 
−
                passive OPEN will await an incoming connection request
 
−
                to any local IP address, and then bind the local IP
 
−
                address of the connection to the particular address
 
−
                that is used.
 
  
−
                For an active OPEN call, a specified "local IP address"
+
              For an active OPEN call, a specified "local IP address"
−
                parameter will be used for opening the connection.  If
+
              parameter will be used for opening the connection.  If
−
                the parameter is unspecified, the networking software
+
              the parameter is unspecified, the networking software
−
                will choose an appropriate local IP address (see
+
              will choose an appropriate local IP address (see
−
                Section 3.3.4.2) for the connection
+
              Section 3.3.4.2) for the connection
  
−
      4.2.5  TCP REQUIREMENT SUMMARY
+
  4.2.5  TCP REQUIREMENT SUMMARY
  
−
                                                |        | | | |S| |
+
                                              |        | | | |S| |
−
                                                |        | | | |H| |F
+
                                              |        | | | |H| |F
−
                                                |        | | | |O|M|o
+
                                              |        | | | |O|M|o
−
                                                |        | |S| |U|U|o
+
                                              |        | |S| |U|U|o
−
                                                |        | |H| |L|S|t
+
                                              |        | |H| |L|S|t
−
                                                |        |M|O| |D|T|n
+
                                              |        |M|O| |D|T|n
−
                                                |        |U|U|M| | |o
+
                                              |        |U|U|M| | |o
−
                                                |        |S|L|A|N|N|t
+
                                              |        |S|L|A|N|N|t
−
                                                |        |T|D|Y|O|O|t
+
                                              |        |T|D|Y|O|O|t
 
FEATURE                                          |SECTION | | | |T|T|e
 
FEATURE                                          |SECTION | | | |T|T|e
 
-------------------------------------------------|--------|-|-|-|-|-|--
 
-------------------------------------------------|--------|-|-|-|-|-|--
−
                                                |        | | | | | |
+
                                              |        | | | | | |
 
Push flag                                        |        | | | | | |
 
Push flag                                        |        | | | | | |
 
   Aggregate or queue un-pushed data              |4.2.2.2 | | |x| | |
 
   Aggregate or queue un-pushed data              |4.2.2.2 | | |x| | |
 
   Sender collapse successive PSH flags          |4.2.2.2 | |x| | | |
 
   Sender collapse successive PSH flags          |4.2.2.2 | |x| | | |
 
   SEND call can specify PUSH                    |4.2.2.2 | | |x| | |
 
   SEND call can specify PUSH                    |4.2.2.2 | | |x| | |
−
 
−
 
−
 
−
Internet Engineering Task Force                              [Page 108]
 
−
 
−
 
−
 
  
 
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
 
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
  
−
 
+
If cannot: sender buffer indefinitely        |4.2.2.2 | | | | |x|
−
    If cannot: sender buffer indefinitely        |4.2.2.2 | | | | |x|
+
If cannot: PSH last segment                  |4.2.2.2 |x| | | | |
−
    If cannot: PSH last segment                  |4.2.2.2 |x| | | | |
 
 
   Notify receiving ALP of PSH                    |4.2.2.2 | | |x| | |1
 
   Notify receiving ALP of PSH                    |4.2.2.2 | | |x| | |1
 
   Send max size segment when possible            |4.2.2.2 | |x| | | |
 
   Send max size segment when possible            |4.2.2.2 | |x| | | |
−
                                                |        | | | | | |
+
                                              |        | | | | | |
 
Window                                          |        | | | | | |
 
Window                                          |        | | | | | |
 
   Treat as unsigned number                      |4.2.2.3 |x| | | | |
 
   Treat as unsigned number                      |4.2.2.3 |x| | | | |
Line 6,389: Line 5,163:
 
   Receiver's window closed indefinitely          |4.2.2.17| | |x| | |
 
   Receiver's window closed indefinitely          |4.2.2.17| | |x| | |
 
   Sender probe zero window                      |4.2.2.17|x| | | | |
 
   Sender probe zero window                      |4.2.2.17|x| | | | |
−
    First probe after RTO                        |4.2.2.17| |x| | | |
+
First probe after RTO                        |4.2.2.17| |x| | | |
−
    Exponential backoff                          |4.2.2.17| |x| | | |
+
Exponential backoff                          |4.2.2.17| |x| | | |
 
   Allow window stay zero indefinitely            |4.2.2.17|x| | | | |
 
   Allow window stay zero indefinitely            |4.2.2.17|x| | | | |
 
   Sender timeout OK conn with zero wind          |4.2.2.17| | | | |x|
 
   Sender timeout OK conn with zero wind          |4.2.2.17| | | | |x|
−
                                                |        | | | | | |
+
                                              |        | | | | | |
 
Urgent Data                                      |        | | | | | |
 
Urgent Data                                      |        | | | | | |
 
   Pointer points to last octet                  |4.2.2.4 |x| | | | |
 
   Pointer points to last octet                  |4.2.2.4 |x| | | | |
Line 6,399: Line 5,173:
 
   Inform ALP asynchronously of urgent data      |4.2.2.4 |x| | | | |1
 
   Inform ALP asynchronously of urgent data      |4.2.2.4 |x| | | | |1
 
   ALP can learn if/how much urgent data Q'd      |4.2.2.4 |x| | | | |1
 
   ALP can learn if/how much urgent data Q'd      |4.2.2.4 |x| | | | |1
−
                                                |        | | | | | |
+
                                              |        | | | | | |
 
TCP Options                                      |        | | | | | |
 
TCP Options                                      |        | | | | | |
 
   Receive TCP option in any segment              |4.2.2.5 |x| | | | |
 
   Receive TCP option in any segment              |4.2.2.5 |x| | | | |
Line 6,409: Line 5,183:
 
   Send-MSS default is 536                        |4.2.2.6 |x| | | | |
 
   Send-MSS default is 536                        |4.2.2.6 |x| | | | |
 
   Calculate effective send seg size              |4.2.2.6 |x| | | | |
 
   Calculate effective send seg size              |4.2.2.6 |x| | | | |
−
                                                |        | | | | | |
+
                                              |        | | | | | |
 
TCP Checksums                                    |        | | | | | |
 
TCP Checksums                                    |        | | | | | |
 
   Sender compute checksum                        |4.2.2.7 |x| | | | |
 
   Sender compute checksum                        |4.2.2.7 |x| | | | |
 
   Receiver check checksum                        |4.2.2.7 |x| | | | |
 
   Receiver check checksum                        |4.2.2.7 |x| | | | |
−
                                                |        | | | | | |
+
                                              |        | | | | | |
 
Use clock-driven ISN selection                  |4.2.2.9 |x| | | | |
 
Use clock-driven ISN selection                  |4.2.2.9 |x| | | | |
−
                                                |        | | | | | |
+
                                              |        | | | | | |
 
Opening Connections                              |        | | | | | |
 
Opening Connections                              |        | | | | | |
 
   Support simultaneous open attempts            |4.2.2.10|x| | | | |
 
   Support simultaneous open attempts            |4.2.2.10|x| | | | |
Line 6,422: Line 5,196:
 
   Function: simultan. LISTENs for same port      |4.2.2.18|x| | | | |
 
   Function: simultan. LISTENs for same port      |4.2.2.18|x| | | | |
 
   Ask IP for src address for SYN if necc.        |4.2.3.7 |x| | | | |
 
   Ask IP for src address for SYN if necc.        |4.2.3.7 |x| | | | |
−
    Otherwise, use local addr of conn.          |4.2.3.7 |x| | | | |
+
Otherwise, use local addr of conn.          |4.2.3.7 |x| | | | |
 
   OPEN to broadcast/multicast IP Address        |4.2.3.14| | | | |x|
 
   OPEN to broadcast/multicast IP Address        |4.2.3.14| | | | |x|
 
   Silently discard seg to bcast/mcast addr      |4.2.3.14|x| | | | |
 
   Silently discard seg to bcast/mcast addr      |4.2.3.14|x| | | | |
−
 
−
 
−
 
−
Internet Engineering Task Force                              [Page 109]
 
−
 
−
 
−
 
  
 
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
 
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
  
−
 
+
                                              |        | | | | | |
−
                                                |        | | | | | |
 
 
Closing Connections                              |        | | | | | |
 
Closing Connections                              |        | | | | | |
 
   RST can contain data                          |4.2.2.12| |x| | | |
 
   RST can contain data                          |4.2.2.12| |x| | | |
 
   Inform application of aborted conn            |4.2.2.13|x| | | | |
 
   Inform application of aborted conn            |4.2.2.13|x| | | | |
 
   Half-duplex close connections                  |4.2.2.13| | |x| | |
 
   Half-duplex close connections                  |4.2.2.13| | |x| | |
−
    Send RST to indicate data lost              |4.2.2.13| |x| | | |
+
Send RST to indicate data lost              |4.2.2.13| |x| | | |
 
   In TIME-WAIT state for 2xMSL seconds          |4.2.2.13|x| | | | |
 
   In TIME-WAIT state for 2xMSL seconds          |4.2.2.13|x| | | | |
−
    Accept SYN from TIME-WAIT state              |4.2.2.13| | |x| | |
+
Accept SYN from TIME-WAIT state              |4.2.2.13| | |x| | |
−
                                                |        | | | | | |
+
                                              |        | | | | | |
 
Retransmissions                                  |        | | | | | |
 
Retransmissions                                  |        | | | | | |
 
   Jacobson Slow Start algorithm                  |4.2.2.15|x| | | | |
 
   Jacobson Slow Start algorithm                  |4.2.2.15|x| | | | |
Line 6,454: Line 5,220:
 
   SYN RTO calc same as data                      |4.2.3.1 | |x| | | |
 
   SYN RTO calc same as data                      |4.2.3.1 | |x| | | |
 
   Recommended initial values and bounds          |4.2.3.1 | |x| | | |
 
   Recommended initial values and bounds          |4.2.3.1 | |x| | | |
−
                                                |        | | | | | |
+
                                              |        | | | | | |
 
Generating ACK's:                                |        | | | | | |
 
Generating ACK's:                                |        | | | | | |
 
   Queue out-of-order segments                    |4.2.2.20| |x| | | |
 
   Queue out-of-order segments                    |4.2.2.20| |x| | | |
Line 6,460: Line 5,226:
 
   Send ACK for out-of-order segment              |4.2.2.21| | |x| | |
 
   Send ACK for out-of-order segment              |4.2.2.21| | |x| | |
 
   Delayed ACK's                                  |4.2.3.2 | |x| | | |
 
   Delayed ACK's                                  |4.2.3.2 | |x| | | |
−
    Delay < 0.5 seconds                          |4.2.3.2 |x| | | | |
+
Delay < 0.5 seconds                          |4.2.3.2 |x| | | | |
−
    Every 2nd full-sized segment ACK'd          |4.2.3.2 |x| | | | |
+
Every 2nd full-sized segment ACK'd          |4.2.3.2 |x| | | | |
 
   Receiver SWS-Avoidance Algorithm              |4.2.3.3 |x| | | | |
 
   Receiver SWS-Avoidance Algorithm              |4.2.3.3 |x| | | | |
−
                                                |        | | | | | |
+
                                              |        | | | | | |
 
Sending data                                    |        | | | | | |
 
Sending data                                    |        | | | | | |
 
   Configurable TTL                              |4.2.2.19|x| | | | |
 
   Configurable TTL                              |4.2.2.19|x| | | | |
 
   Sender SWS-Avoidance Algorithm                |4.2.3.4 |x| | | | |
 
   Sender SWS-Avoidance Algorithm                |4.2.3.4 |x| | | | |
 
   Nagle algorithm                                |4.2.3.4 | |x| | | |
 
   Nagle algorithm                                |4.2.3.4 | |x| | | |
−
    Application can disable Nagle algorithm      |4.2.3.4 |x| | | | |
+
Application can disable Nagle algorithm      |4.2.3.4 |x| | | | |
−
                                                |        | | | | | |
+
                                              |        | | | | | |
 
Connection Failures:                            |        | | | | | |
 
Connection Failures:                            |        | | | | | |
 
   Negative advice to IP on R1 retxs              |4.2.3.5 |x| | | | |
 
   Negative advice to IP on R1 retxs              |4.2.3.5 |x| | | | |
Line 6,477: Line 5,243:
 
   Recommended values for R1, R2                  |4.2.3.5 | |x| | | |
 
   Recommended values for R1, R2                  |4.2.3.5 | |x| | | |
 
   Same mechanism for SYNs                        |4.2.3.5 |x| | | | |
 
   Same mechanism for SYNs                        |4.2.3.5 |x| | | | |
−
    R2 at least 3 minutes for SYN                |4.2.3.5 |x| | | | |
+
R2 at least 3 minutes for SYN                |4.2.3.5 |x| | | | |
−
                                                |        | | | | | |
+
                                              |        | | | | | |
 
Send Keep-alive Packets:                        |4.2.3.6 | | |x| | |
 
Send Keep-alive Packets:                        |4.2.3.6 | | |x| | |
 
   - Application can request                      |4.2.3.6 |x| | | | |
 
   - Application can request                      |4.2.3.6 |x| | | | |
Line 6,484: Line 5,250:
 
   - Only send if idle for interval              |4.2.3.6 |x| | | | |
 
   - Only send if idle for interval              |4.2.3.6 |x| | | | |
 
   - Interval configurable                        |4.2.3.6 |x| | | | |
 
   - Interval configurable                        |4.2.3.6 |x| | | | |
−
 
−
 
−
 
−
Internet Engineering Task Force                              [Page 110]
 
−
 
−
 
−
 
  
 
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
 
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
−
 
  
 
   - Default at least 2 hrs.                      |4.2.3.6 |x| | | | |
 
   - Default at least 2 hrs.                      |4.2.3.6 |x| | | | |
 
   - Tolerant of lost ACK's                      |4.2.3.6 |x| | | | |
 
   - Tolerant of lost ACK's                      |4.2.3.6 |x| | | | |
−
                                                |        | | | | | |
+
                                              |        | | | | | |
 
IP Options                                      |        | | | | | |
 
IP Options                                      |        | | | | | |
 
   Ignore options TCP doesn't understand          |4.2.3.8 |x| | | | |
 
   Ignore options TCP doesn't understand          |4.2.3.8 |x| | | | |
Line 6,503: Line 5,261:
 
   Record Route support                          |4.2.3.8 | | |x| | |
 
   Record Route support                          |4.2.3.8 | | |x| | |
 
   Source Route:                                  |        | | | | | |
 
   Source Route:                                  |        | | | | | |
−
    ALP can specify                              |4.2.3.8 |x| | | | |1
+
ALP can specify                              |4.2.3.8 |x| | | | |1
−
      Overrides src rt in datagram              |4.2.3.8 |x| | | | |
+
  Overrides src rt in datagram              |4.2.3.8 |x| | | | |
−
    Build return route from src rt              |4.2.3.8 |x| | | | |
+
Build return route from src rt              |4.2.3.8 |x| | | | |
−
    Later src route overrides                    |4.2.3.8 | |x| | | |
+
Later src route overrides                    |4.2.3.8 | |x| | | |
−
                                                |        | | | | | |
+
                                              |        | | | | | |
 
Receiving ICMP Messages from IP                  |4.2.3.9 |x| | | | |
 
Receiving ICMP Messages from IP                  |4.2.3.9 |x| | | | |
 
   Dest. Unreach (0,1,5) => inform ALP            |4.2.3.9 | |x| | | |
 
   Dest. Unreach (0,1,5) => inform ALP            |4.2.3.9 | |x| | | |
Line 6,515: Line 5,273:
 
   Time Exceeded => tell ALP, don't abort        |4.2.3.9 | |x| | | |
 
   Time Exceeded => tell ALP, don't abort        |4.2.3.9 | |x| | | |
 
   Param Problem => tell ALP, don't abort        |4.2.3.9 | |x| | | |
 
   Param Problem => tell ALP, don't abort        |4.2.3.9 | |x| | | |
−
                                                |        | | | | | |
+
                                              |        | | | | | |
 
Address Validation                              |        | | | | | |
 
Address Validation                              |        | | | | | |
 
   Reject OPEN call to invalid IP address        |4.2.3.10|x| | | | |
 
   Reject OPEN call to invalid IP address        |4.2.3.10|x| | | | |
 
   Reject SYN from invalid IP address            |4.2.3.10|x| | | | |
 
   Reject SYN from invalid IP address            |4.2.3.10|x| | | | |
 
   Silently discard SYN to bcast/mcast addr      |4.2.3.10|x| | | | |
 
   Silently discard SYN to bcast/mcast addr      |4.2.3.10|x| | | | |
−
                                                |        | | | | | |
+
                                              |        | | | | | |
 
TCP/ALP Interface Services                      |        | | | | | |
 
TCP/ALP Interface Services                      |        | | | | | |
 
   Error Report mechanism                        |4.2.4.1 |x| | | | |
 
   Error Report mechanism                        |4.2.4.1 |x| | | | |
 
   ALP can disable Error Report Routine          |4.2.4.1 | |x| | | |
 
   ALP can disable Error Report Routine          |4.2.4.1 | |x| | | |
 
   ALP can specify TOS for sending                |4.2.4.2 |x| | | | |
 
   ALP can specify TOS for sending                |4.2.4.2 |x| | | | |
−
    Passed unchanged to IP                      |4.2.4.2 | |x| | | |
+
Passed unchanged to IP                      |4.2.4.2 | |x| | | |
 
   ALP can change TOS during connection          |4.2.4.2 | |x| | | |
 
   ALP can change TOS during connection          |4.2.4.2 | |x| | | |
 
   Pass received TOS up to ALP                    |4.2.4.2 | | |x| | |
 
   Pass received TOS up to ALP                    |4.2.4.2 | | |x| | |
Line 6,536: Line 5,294:
  
 
(1)  "ALP" means Application-Layer program.
 
(1)  "ALP" means Application-Layer program.
−
 
−
 
−
 
−
 
−
 
−
 
−
 
−
 
−
 
−
 
−
Internet Engineering Task Force                              [Page 111]
 
−
 
−
 
−
 
  
 
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
 
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
  
−
 
+
== REFERENCES ==
−
5.  REFERENCES
 
  
 
INTRODUCTORY REFERENCES
 
INTRODUCTORY REFERENCES
−
 
  
 
[INTRO:1] "Requirements for Internet Hosts -- Application and Support,"
 
[INTRO:1] "Requirements for Internet Hosts -- Application and Support,"
−
    IETF Host Requirements Working Group, R. Braden, Ed., RFC-1123,
+
  IETF Host Requirements Working Group, R. Braden, Ed., RFC-1123,
−
    October 1989.
+
  October 1989.
  
 
[INTRO:2]  "Requirements for Internet Gateways,"  R. Braden and J.
 
[INTRO:2]  "Requirements for Internet Gateways,"  R. Braden and J.
−
    Postel, RFC-1009, June 1987.
+
  Postel, RFC-1009, June 1987.
  
 
[INTRO:3]  "DDN Protocol Handbook," NIC-50004, NIC-50005, NIC-50006,
 
[INTRO:3]  "DDN Protocol Handbook," NIC-50004, NIC-50005, NIC-50006,
−
    (three volumes), SRI International, December 1985.
+
  (three volumes), SRI International, December 1985.
  
 
[INTRO:4]  "Official Internet Protocols," J. Reynolds and J. Postel,
 
[INTRO:4]  "Official Internet Protocols," J. Reynolds and J. Postel,
−
    RFC-1011, May 1987.
+
  RFC-1011, May 1987.
  
−
    This document is republished periodically with new RFC numbers; the
+
  This document is republished periodically with new RFC numbers; the
−
    latest version must be used.
+
  latest version must be used.
  
 
[INTRO:5]  "Protocol Document Order Information," O. Jacobsen and J.
 
[INTRO:5]  "Protocol Document Order Information," O. Jacobsen and J.
−
    Postel, RFC-980, March 1986.
+
  Postel, RFC-980, March 1986.
  
 
[INTRO:6]  "Assigned Numbers," J. Reynolds and J. Postel, RFC-1010, May
 
[INTRO:6]  "Assigned Numbers," J. Reynolds and J. Postel, RFC-1010, May
−
    1987.
+
  1987.
  
−
    This document is republished periodically with new RFC numbers; the
+
  This document is republished periodically with new RFC numbers; the
−
    latest version must be used.
+
  latest version must be used.
  
 
[INTRO:7] "Modularity and Efficiency in Protocol Implementations," D.
 
[INTRO:7] "Modularity and Efficiency in Protocol Implementations," D.
−
    Clark, RFC-817, July 1982.
+
  Clark, RFC-817, July 1982.
  
 
[INTRO:8] "The Structuring of Systems Using Upcalls," D. Clark, 10th ACM
 
[INTRO:8] "The Structuring of Systems Using Upcalls," D. Clark, 10th ACM
−
    SOSP, Orcas Island, Washington, December 1985.
+
  SOSP, Orcas Island, Washington, December 1985.
−
 
 
  
 
Secondary References:
 
Secondary References:
−
 
  
 
[INTRO:9]  "A Protocol for Packet Network Intercommunication," V. Cerf
 
[INTRO:9]  "A Protocol for Packet Network Intercommunication," V. Cerf
−
    and R. Kahn, IEEE Transactions on Communication, May 1974.
+
  and R. Kahn, IEEE Transactions on Communication, May 1974.
  
 
[INTRO:10]  "The ARPA Internet Protocol," J. Postel, C. Sunshine, and D.
 
[INTRO:10]  "The ARPA Internet Protocol," J. Postel, C. Sunshine, and D.
−
    Cohen, Computer Networks, Vol. 5, No. 4, July 1981.
+
  Cohen, Computer Networks, Vol. 5, No. 4, July 1981.
  
 
[INTRO:11]  "The DARPA Internet Protocol Suite," B. Leiner, J. Postel,
 
[INTRO:11]  "The DARPA Internet Protocol Suite," B. Leiner, J. Postel,
−
    R. Cole and D. Mills, Proceedings INFOCOM 85, IEEE, Washington DC,
+
  R. Cole and D. Mills, Proceedings INFOCOM 85, IEEE, Washington DC,
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                              [Page 112]
 
−
 
 
−
 
 
−
 
 
  
 
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
 
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
  
−
 
+
  March 1985.  Also in: IEEE Communications Magazine, March 1985.
−
    March 1985.  Also in: IEEE Communications Magazine, March 1985.
+
  Also available as ISI-RS-85-153.
−
    Also available as ISI-RS-85-153.
 
  
 
[INTRO:12] "Final Text of DIS8473, Protocol for Providing the
 
[INTRO:12] "Final Text of DIS8473, Protocol for Providing the
−
    Connectionless Mode Network Service," ANSI, published as RFC-994,
+
  Connectionless Mode Network Service," ANSI, published as RFC-994,
−
    March 1986.
+
  March 1986.
  
 
[INTRO:13] "End System to Intermediate System Routing Exchange
 
[INTRO:13] "End System to Intermediate System Routing Exchange
−
    Protocol," ANSI X3S3.3, published as RFC-995, April 1986.
+
  Protocol," ANSI X3S3.3, published as RFC-995, April 1986.
−
 
 
  
 
LINK LAYER REFERENCES
 
LINK LAYER REFERENCES
−
 
  
 
[LINK:1] "Trailer Encapsulations," S. Leffler and M. Karels, RFC-893,
 
[LINK:1] "Trailer Encapsulations," S. Leffler and M. Karels, RFC-893,
−
    April 1984.
+
  April 1984.
  
 
[LINK:2] "An Ethernet Address Resolution Protocol," D. Plummer, RFC-826,
 
[LINK:2] "An Ethernet Address Resolution Protocol," D. Plummer, RFC-826,
−
    November 1982.
+
  November 1982.
  
 
[LINK:3] "A Standard for the Transmission of IP Datagrams over Ethernet
 
[LINK:3] "A Standard for the Transmission of IP Datagrams over Ethernet
−
    Networks," C. Hornig, RFC-894, April 1984.
+
  Networks," C. Hornig, RFC-894, April 1984.
  
 
[LINK:4] "A Standard for the Transmission of IP Datagrams over IEEE 802
 
[LINK:4] "A Standard for the Transmission of IP Datagrams over IEEE 802
−
    "Networks," J. Postel and J. Reynolds, RFC-1042, February 1988.
+
  "Networks," J. Postel and J. Reynolds, RFC-1042, February 1988.
−
 
 
−
    This RFC contains a great deal of information of importance to
 
−
    Internet implementers planning to use IEEE 802 networks.
 
  
 +
  This RFC contains a great deal of information of importance to
 +
  Internet implementers planning to use IEEE 802 networks.
  
 
IP LAYER REFERENCES
 
IP LAYER REFERENCES
−
 
  
 
[IP:1] "Internet Protocol (IP)," J. Postel, RFC-791, September 1981.
 
[IP:1] "Internet Protocol (IP)," J. Postel, RFC-791, September 1981.
  
 
[IP:2] "Internet Control Message Protocol (ICMP)," J. Postel, RFC-792,
 
[IP:2] "Internet Control Message Protocol (ICMP)," J. Postel, RFC-792,
−
    September 1981.
+
  September 1981.
  
 
[IP:3] "Internet Standard Subnetting Procedure," J. Mogul and J. Postel,
 
[IP:3] "Internet Standard Subnetting Procedure," J. Mogul and J. Postel,
−
    RFC-950, August 1985.
+
  RFC-950, August 1985.
  
 
[IP:4]  "Host Extensions for IP Multicasting," S. Deering, RFC-1112,
 
[IP:4]  "Host Extensions for IP Multicasting," S. Deering, RFC-1112,
−
    August 1989.
+
  August 1989.
  
 
[IP:5] "Military Standard Internet Protocol," MIL-STD-1777, Department
 
[IP:5] "Military Standard Internet Protocol," MIL-STD-1777, Department
−
    of Defense, August 1983.
+
  of Defense, August 1983.
−
 
 
−
    This specification, as amended by RFC-963, is intended to describe
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                              [Page 113]
 
−
 
 
−
 
 
  
 +
  This specification, as amended by RFC-963, is intended to describe
  
 
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
 
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
  
−
 
+
  the Internet Protocol but has some serious omissions (e.g., the
−
    the Internet Protocol but has some serious omissions (e.g., the
+
  mandatory subnet extension [IP:3] and the optional multicasting
−
    mandatory subnet extension [IP:3] and the optional multicasting
+
  extension [IP:4]).  It is also out of date.  If there is a
−
    extension [IP:4]).  It is also out of date.  If there is a
+
  conflict, RFC-791, RFC-792, and RFC-950 must be taken as
−
    conflict, RFC-791, RFC-792, and RFC-950 must be taken as
+
  authoritative, while the present document is authoritative over
−
    authoritative, while the present document is authoritative over
+
  all.
−
    all.
 
  
 
[IP:6] "Some Problems with the Specification of the Military Standard
 
[IP:6] "Some Problems with the Specification of the Military Standard
−
    Internet Protocol," D. Sidhu, RFC-963, November 1985.
+
  Internet Protocol," D. Sidhu, RFC-963, November 1985.
  
 
[IP:7] "The TCP Maximum Segment Size and Related Topics," J. Postel,
 
[IP:7] "The TCP Maximum Segment Size and Related Topics," J. Postel,
−
    RFC-879, November 1983.
+
  RFC-879, November 1983.
  
−
    Discusses and clarifies the relationship between the TCP Maximum
+
  Discusses and clarifies the relationship between the TCP Maximum
−
    Segment Size option and the IP datagram size.
+
  Segment Size option and the IP datagram size.
  
 
[IP:8] "Internet Protocol Security Options,"  B. Schofield, RFC-1108,
 
[IP:8] "Internet Protocol Security Options,"  B. Schofield, RFC-1108,
−
    October 1989.
+
  October 1989.
  
 
[IP:9] "Fragmentation Considered Harmful," C. Kent and J. Mogul, ACM
 
[IP:9] "Fragmentation Considered Harmful," C. Kent and J. Mogul, ACM
−
    SIGCOMM-87, August 1987.  Published as ACM Comp Comm Review, Vol.
+
  SIGCOMM-87, August 1987.  Published as ACM Comp Comm Review, Vol.
−
    17, no. 5.
+
  17, no. 5.
  
−
    This useful paper discusses the problems created by Internet
+
  This useful paper discusses the problems created by Internet
−
    fragmentation and presents alternative solutions.
+
  fragmentation and presents alternative solutions.
  
 
[IP:10] "IP Datagram Reassembly Algorithms," D. Clark, RFC-815, July
 
[IP:10] "IP Datagram Reassembly Algorithms," D. Clark, RFC-815, July
−
    1982.
+
  1982.
  
−
    This and the following paper should be read by every implementor.
+
  This and the following paper should be read by every implementor.
  
 
[IP:11] "Fault Isolation and Recovery," D. Clark, RFC-816, July 1982.
 
[IP:11] "Fault Isolation and Recovery," D. Clark, RFC-816, July 1982.
  
 
SECONDARY IP REFERENCES:
 
SECONDARY IP REFERENCES:
−
 
  
 
[IP:12] "Broadcasting Internet Datagrams in the Presence of Subnets," J.
 
[IP:12] "Broadcasting Internet Datagrams in the Presence of Subnets," J.
−
    Mogul, RFC-922, October 1984.
+
  Mogul, RFC-922, October 1984.
  
 
[IP:13] "Name, Addresses, Ports, and Routes," D. Clark, RFC-814, July
 
[IP:13] "Name, Addresses, Ports, and Routes," D. Clark, RFC-814, July
−
    1982.
+
  1982.
  
 
[IP:14] "Something a Host Could Do with Source Quench: The Source Quench
 
[IP:14] "Something a Host Could Do with Source Quench: The Source Quench
−
    Introduced Delay (SQUID)," W. Prue and J. Postel, RFC-1016, July
+
  Introduced Delay (SQUID)," W. Prue and J. Postel, RFC-1016, July
−
    1987.
+
  1987.
−
 
 
−
    This RFC first described directed broadcast addresses.  However,
 
−
    the bulk of the RFC is concerned with gateways, not hosts.
 
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                              [Page 114]
 
−
 
 
−
 
 
  
 +
  This RFC first described directed broadcast addresses.  However,
 +
  the bulk of the RFC is concerned with gateways, not hosts.
  
 
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
 
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
−
 
  
 
UDP REFERENCES:
 
UDP REFERENCES:
−
 
  
 
[UDP:1] "User Datagram Protocol," J. Postel, RFC-768, August 1980.
 
[UDP:1] "User Datagram Protocol," J. Postel, RFC-768, August 1980.
−
 
  
 
TCP REFERENCES:
 
TCP REFERENCES:
−
 
  
 
[TCP:1] "Transmission Control Protocol," J. Postel, RFC-793, September
 
[TCP:1] "Transmission Control Protocol," J. Postel, RFC-793, September
−
    1981.
+
  1981.
−
 
 
  
 
[TCP:2] "Transmission Control Protocol," MIL-STD-1778, US Department of
 
[TCP:2] "Transmission Control Protocol," MIL-STD-1778, US Department of
−
    Defense, August 1984.
+
  Defense, August 1984.
−
 
 
−
    This specification as amended by RFC-964 is intended to describe
 
−
    the same protocol as RFC-793 [TCP:1].  If there is a conflict,
 
−
    RFC-793 takes precedence, and the present document is authoritative
 
−
    over both.
 
  
 +
  This specification as amended by RFC-964 is intended to describe
 +
  the same protocol as RFC-793 [TCP:1].  If there is a conflict,
 +
  RFC-793 takes precedence, and the present document is authoritative
 +
  over both.
  
 
[TCP:3] "Some Problems with the Specification of the Military Standard
 
[TCP:3] "Some Problems with the Specification of the Military Standard
−
    Transmission Control Protocol," D. Sidhu and T. Blumer, RFC-964,
+
  Transmission Control Protocol," D. Sidhu and T. Blumer, RFC-964,
−
    November 1985.
+
  November 1985.
−
 
 
  
 
[TCP:4] "The TCP Maximum Segment Size and Related Topics," J. Postel,
 
[TCP:4] "The TCP Maximum Segment Size and Related Topics," J. Postel,
−
    RFC-879, November 1983.
+
  RFC-879, November 1983.
−
 
 
  
 
[TCP:5] "Window and Acknowledgment Strategy in TCP," D. Clark, RFC-813,
 
[TCP:5] "Window and Acknowledgment Strategy in TCP," D. Clark, RFC-813,
−
    July 1982.
+
  July 1982.
−
 
 
  
 
[TCP:6] "Round Trip Time Estimation," P. Karn & C. Partridge, ACM
 
[TCP:6] "Round Trip Time Estimation," P. Karn & C. Partridge, ACM
−
    SIGCOMM-87, August 1987.
+
  SIGCOMM-87, August 1987.
−
 
 
  
 
[TCP:7] "Congestion Avoidance and Control," V. Jacobson, ACM SIGCOMM-88,
 
[TCP:7] "Congestion Avoidance and Control," V. Jacobson, ACM SIGCOMM-88,
−
    August 1988.
+
  August 1988.
−
 
 
  
 
SECONDARY TCP REFERENCES:
 
SECONDARY TCP REFERENCES:
−
 
  
 
[TCP:8] "Modularity and Efficiency in Protocol Implementation," D.
 
[TCP:8] "Modularity and Efficiency in Protocol Implementation," D.
−
    Clark, RFC-817, July 1982.
+
  Clark, RFC-817, July 1982.
−
 
 
−
 
 
−
 
 
−
Internet Engineering Task Force                              [Page 115]
 
−
 
 
−
 
 
−
 
 
  
 
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
 
RFC1122                  TRANSPORT LAYER -- TCP            October 1989
−
 
  
 
[TCP:9] "Congestion Control in IP/TCP," J. Nagle, RFC-896, January 1984.
 
[TCP:9] "Congestion Control in IP/TCP," J. Nagle, RFC-896, January 1984.
−
 
  
 
[TCP:10] "Computing the Internet Checksum," R. Braden, D. Borman, and C.
 
[TCP:10] "Computing the Internet Checksum," R. Braden, D. Borman, and C.
−
    Partridge, RFC-1071, September 1988.
+
  Partridge, RFC-1071, September 1988.
−
 
 
  
 
[TCP:11] "TCP Extensions for Long-Delay Paths," V. Jacobson & R. Braden,
 
[TCP:11] "TCP Extensions for Long-Delay Paths," V. Jacobson & R. Braden,
−
    RFC-1072, October 1988.
+
  RFC-1072, October 1988.
−
 
 
  
 
Security Considerations
 
Security Considerations
  
−
  There are many security issues in the communication layers of host
+
There are many security issues in the communication layers of host
−
  software, but a full discussion is beyond the scope of this RFC.
+
software, but a full discussion is beyond the scope of this RFC.
  
−
  The Internet architecture generally provides little protection
+
The Internet architecture generally provides little protection
−
  against spoofing of IP source addresses, so any security mechanism
+
against spoofing of IP source addresses, so any security mechanism
−
  that is based upon verifying the IP source address of a datagram
+
that is based upon verifying the IP source address of a datagram
−
  should be treated with suspicion.  However, in restricted
+
should be treated with suspicion.  However, in restricted
−
  environments some source-address checking may be possible.  For
+
environments some source-address checking may be possible.  For
−
  example, there might be a secure LAN whose gateway to the rest of the
+
example, there might be a secure LAN whose gateway to the rest of the
−
  Internet discarded any incoming datagram with a source address that
+
Internet discarded any incoming datagram with a source address that
−
  spoofed the LAN address.  In this case, a host on the LAN could use
+
spoofed the LAN address.  In this case, a host on the LAN could use
−
  the source address to test for local vs. remote source.  This problem
+
the source address to test for local vs. remote source.  This problem
−
  is complicated by source routing, and some have suggested that
+
is complicated by source routing, and some have suggested that
−
  source-routed datagram forwarding by hosts (see Section 3.3.5) should
+
source-routed datagram forwarding by hosts (see Section 3.3.5) should
−
  be outlawed for security reasons.
+
be outlawed for security reasons.
  
−
  Security-related issues are mentioned in sections concerning the IP
+
Security-related issues are mentioned in sections concerning the IP
−
  Security option (Section 3.2.1.8), the ICMP Parameter Problem message
+
Security option (Section 3.2.1.8), the ICMP Parameter Problem message
−
  (Section 3.2.2.5), IP options in UDP datagrams (Section 4.1.3.2), and
+
(Section 3.2.2.5), IP options in UDP datagrams (Section 4.1.3.2), and
−
  reserved TCP ports (Section 4.2.2.1).
+
reserved TCP ports (Section 4.2.2.1).
  
 
Author's Address
 
Author's Address
  
−
  Robert Braden
+
Robert Braden
−
  USC/Information Sciences Institute
+
USC/Information Sciences Institute
−
  4676 Admiralty Way
+
4676 Admiralty Way
−
  Marina del Rey, CA 90292-6695
+
Marina del Rey, CA 90292-6695
−
 
 
−
  Phone: (213) 822 1511
 
−
 
 
− 
−
 
 
−
 
 
−
 
 
−
 
 
−
 
 
  
 +
Phone: (213) 822 1511
  
−
Internet Engineering Task Force                              [Page 116]
+

Latest revision as of 15:45, 15 October 2020

Network Working Group Internet Engineering Task Force Request for Comments: 1122 R. Braden, Editor 

                                                        October 1989

    Requirements for Internet Hosts -- Communication Layers

Status of This Memo

This RFC is an official specification for the Internet community. It incorporates by reference, amends, corrects, and supplements the primary protocol standards documents relating to hosts. Distribution of this document is unlimited.

Summary

This is one RFC of a pair that defines and discusses the requirements for Internet host software. This RFC covers the communications protocol layers: link layer, IP layer, and transport layer; its companion RFC-1123 covers the application and support protocols. 

                       Table of Contents

RFC1122 INTRODUCTION October 1989

RFC1122 INTRODUCTION October 1989

RFC1122 INTRODUCTION October 1989

RFC1122 INTRODUCTION October 1989

INTRODUCTION

This document is one of a pair that defines and discusses the requirements for host system implementations of the Internet protocol suite. This RFC covers the communication protocol layers: link layer, IP layer, and transport layer. Its companion RFC, "Requirements for Internet Hosts -- Application and Support" [INTRO:1], covers the application layer protocols. This document should also be read in conjunction with "Requirements for Internet Gateways" [INTRO:2].

These documents are intended to provide guidance for vendors, implementors, and users of Internet communication software. They represent the consensus of a large body of technical experience and wisdom, contributed by the members of the Internet research and vendor communities.

This RFC enumerates standard protocols that a host connected to the Internet must use, and it incorporates by reference the RFCs and other documents describing the current specifications for these protocols. It corrects errors in the referenced documents and adds additional discussion and guidance for an implementor.

For each protocol, this document also contains an explicit set of requirements, recommendations, and options. The reader must understand that the list of requirements in this document is incomplete by itself; the complete set of requirements for an Internet host is primarily defined in the standard protocol specification documents, with the corrections, amendments, and supplements contained in this RFC.

A good-faith implementation of the protocols that was produced after careful reading of the RFC's and with some interaction with the Internet technical community, and that followed good communications software engineering practices, should differ from the requirements of this document in only minor ways. Thus, in many cases, the "requirements" in this RFC are already stated or implied in the standard protocol documents, so that their inclusion here is, in a sense, redundant. However, they were included because some past implementation has made the wrong choice, causing problems of interoperability, performance, and/or robustness.

This document includes discussion and explanation of many of the requirements and recommendations. A simple list of requirements would be dangerous, because:

o Some required features are more important than others, and some 

    features are optional.

RFC1122 INTRODUCTION October 1989

o There may be valid reasons why particular vendor products that 

    are designed for restricted contexts might choose to use
    different specifications.

However, the specifications of this document must be followed to meet the general goal of arbitrary host interoperation across the diversity and complexity of the Internet system. Although most current implementations fail to meet these requirements in various ways, some minor and some major, this specification is the ideal towards which we need to move.

These requirements are based on the current level of Internet architecture. This document will be updated as required to provide additional clarifications or to include additional information in those areas in which specifications are still evolving.

This introductory section begins with a brief overview of the Internet architecture as it relates to hosts, and then gives some general advice to host software vendors. Finally, there is some guidance on reading the rest of the document and some terminology.

1.1 The Internet Architecture 

  General background and discussion on the Internet architecture and
  supporting protocol suite can be found in the DDN Protocol
  Handbook [INTRO:3]; for background see for example [INTRO:9],
  [INTRO:10], and [INTRO:11].  Reference [INTRO:5] describes the
  procedure for obtaining Internet protocol documents, while
  [INTRO:6] contains a list of the numbers assigned within Internet
  protocols.

  1.1.1  Internet Hosts

     A host computer, or simply "host," is the ultimate consumer of
     communication services.  A host generally executes application
     programs on behalf of user(s), employing network and/or
     Internet communication services in support of this function.
     An Internet host corresponds to the concept of an "End-System"
     used in the OSI protocol suite [INTRO:13].

     An Internet communication system consists of interconnected
     packet networks supporting communication among host computers
     using the Internet protocols.  The networks are interconnected
     using packet-switching computers called "gateways" or "IP
     routers" by the Internet community, and "Intermediate Systems"
     by the OSI world [INTRO:13].  The RFC "Requirements for
     Internet Gateways" [INTRO:2] contains the official
     specifications for Internet gateways.  That RFC together with

RFC1122 INTRODUCTION October 1989 

     the present document and its companion [INTRO:1] define the
     rules for the current realization of the Internet architecture.

     Internet hosts span a wide range of size, speed, and function.
     They range in size from small microprocessors through
     workstations to mainframes and supercomputers.  In function,
     they range from single-purpose hosts (such as terminal servers)
     to full-service hosts that support a variety of online network
     services, typically including remote login, file transfer, and
     electronic mail.

     A host is generally said to be multihomed if it has more than
     one interface to the same or to different networks.  See
     Section 1.1.3 on "Terminology".

  1.1.2  Architectural Assumptions

     The current Internet architecture is based on a set of
     assumptions about the communication system.  The assumptions
     most relevant to hosts are as follows:

     (a)  The Internet is a network of networks.

          Each host is directly connected to some particular
          network(s); its connection to the Internet is only
          conceptual.  Two hosts on the same network communicate
          with each other using the same set of protocols that they
          would use to communicate with hosts on distant networks.

     (b)  Gateways don't keep connection state information.

          To improve robustness of the communication system,
          gateways are designed to be stateless, forwarding each IP
          datagram independently of other datagrams.  As a result,
          redundant paths can be exploited to provide robust service
          in spite of failures of intervening gateways and networks.

          All state information required for end-to-end flow control
          and reliability is implemented in the hosts, in the
          transport layer or in application programs.  All
          connection control information is thus co-located with the
          end points of the communication, so it will be lost only
          if an end point fails.

     (c)  Routing complexity should be in the gateways.

          Routing is a complex and difficult problem, and ought to
          be performed by the gateways, not the hosts.  An important

RFC1122 INTRODUCTION October 1989 

          objective is to insulate host software from changes caused
          by the inevitable evolution of the Internet routing
          architecture.

     (d)  The System must tolerate wide network variation.

          A basic objective of the Internet design is to tolerate a
          wide range of network characteristics -- e.g., bandwidth,
          delay, packet loss, packet reordering, and maximum packet
          size.  Another objective is robustness against failure of
          individual networks, gateways, and hosts, using whatever
          bandwidth is still available.  Finally, the goal is full
          "open system interconnection": an Internet host must be
          able to interoperate robustly and effectively with any
          other Internet host, across diverse Internet paths.

          Sometimes host implementors have designed for less
          ambitious goals.  For example, the LAN environment is
          typically much more benign than the Internet as a whole;
          LANs have low packet loss and delay and do not reorder
          packets.  Some vendors have fielded host implementations
          that are adequate for a simple LAN environment, but work
          badly for general interoperation.  The vendor justifies
          such a product as being economical within the restricted
          LAN market.  However, isolated LANs seldom stay isolated
          for long; they are soon gatewayed to each other, to
          organization-wide internets, and eventually to the global
          Internet system.  In the end, neither the customer nor the
          vendor is served by incomplete or substandard Internet
          host software.

          The requirements spelled out in this document are designed
          for a full-function Internet host, capable of full
          interoperation over an arbitrary Internet path.

  1.1.3  Internet Protocol Suite

     To communicate using the Internet system, a host must implement
     the layered set of protocols comprising the Internet protocol
     suite.  A host typically must implement at least one protocol
     from each layer.

     The protocol layers used in the Internet architecture are as
     follows [INTRO:4]:

     o  Application Layer

RFC1122 INTRODUCTION October 1989 

          The application layer is the top layer of the Internet
          protocol suite.  The Internet suite does not further
          subdivide the application layer, although some of the
          Internet application layer protocols do contain some
          internal sub-layering.  The application layer of the
          Internet suite essentially combines the functions of the
          top two layers -- Presentation and Application -- of the
          OSI reference model.

          We distinguish two categories of application layer
          protocols:  user protocols that provide service directly
          to users, and support protocols that provide common system
          functions.  Requirements for user and support protocols
          will be found in the companion RFC [INTRO:1].

          The most common Internet user protocols are:

            o  Telnet (remote login)
            o  FTP    (file transfer)
            o  SMTP   (electronic mail delivery)

          There are a number of other standardized user protocols
          [INTRO:4] and many private user protocols.

          Support protocols, used for host name mapping, booting,
          and management, include SNMP, BOOTP, RARP, and the Domain
          Name System (DNS) protocols.

     o  Transport Layer

          The transport layer provides end-to-end communication
          services for applications.  There are two primary
          transport layer protocols at present:

            o Transmission Control Protocol (TCP)
            o User Datagram Protocol (UDP)

          TCP is a reliable connection-oriented transport service
          that provides end-to-end reliability, resequencing, and
          flow control.  UDP is a connectionless ("datagram")
          transport service.

          Other transport protocols have been developed by the
          research community, and the set of official Internet
          transport protocols may be expanded in the future.

          Transport layer protocols are discussed in Chapter 4.

RFC1122 INTRODUCTION October 1989 

     o  Internet Layer

          All Internet transport protocols use the Internet Protocol
          (IP) to carry data from source host to destination host.
          IP is a connectionless or datagram internetwork service,
          providing no end-to-end delivery guarantees. Thus, IP
          datagrams may arrive at the destination host damaged,
          duplicated, out of order, or not at all.  The layers above
          IP are responsible for reliable delivery service when it
          is required.  The IP protocol includes provision for
          addressing, type-of-service specification, fragmentation
          and reassembly, and security information.

          The datagram or connectionless nature of the IP protocol
          is a fundamental and characteristic feature of the
          Internet architecture.  Internet IP was the model for the
          OSI Connectionless Network Protocol [INTRO:12].

          ICMP is a control protocol that is considered to be an
          integral part of IP, although it is architecturally
          layered upon IP, i.e., it uses IP to carry its data end-
          to-end just as a transport protocol like TCP or UDP does.
          ICMP provides error reporting, congestion reporting, and
          first-hop gateway redirection.

          IGMP is an Internet layer protocol used for establishing
          dynamic host groups for IP multicasting.

          The Internet layer protocols IP, ICMP, and IGMP are
          discussed in Chapter 3.

     o  Link Layer

          To communicate on its directly-connected network, a host
          must implement the communication protocol used to
          interface to that network.  We call this a link layer or
          media-access layer protocol.

          There is a wide variety of link layer protocols,
          corresponding to the many different types of networks.
          See Chapter 2.

  1.1.4  Embedded Gateway Code

     Some Internet host software includes embedded gateway
     functionality, so that these hosts can forward packets as a

RFC1122 INTRODUCTION October 1989 

     gateway would, while still performing the application layer
     functions of a host.

     Such dual-purpose systems must follow the Gateway Requirements
     RFC [INTRO:2]  with respect to their gateway functions, and
     must follow the present document with respect to their host
     functions.  In all overlapping cases, the two specifications
     should be in agreement.

     There are varying opinions in the Internet community about
     embedded gateway functionality.  The main arguments are as
     follows:

     o    Pro: in a local network environment where networking is
          informal, or in isolated internets, it may be convenient
          and economical to use existing host systems as gateways.

          There is also an architectural argument for embedded
          gateway functionality: multihoming is much more common
          than originally foreseen, and multihoming forces a host to
          make routing decisions as if it were a gateway.  If the
          multihomed  host contains an embedded gateway, it will
          have full routing knowledge and as a result will be able
          to make more optimal routing decisions.

     o    Con: Gateway algorithms and protocols are still changing,
          and they will continue to change as the Internet system
          grows larger.  Attempting to include a general gateway
          function within the host IP layer will force host system
          maintainers to track these (more frequent) changes.  Also,
          a larger pool of gateway implementations will make
          coordinating the changes more difficult.  Finally, the
          complexity of a gateway IP layer is somewhat greater than
          that of a host, making the implementation and operation
          tasks more complex.

          In addition, the style of operation of some hosts is not
          appropriate for providing stable and robust gateway
          service.

     There is considerable merit in both of these viewpoints.  One
     conclusion can be drawn: an host administrator must have
     conscious control over whether or not a given host acts as a
     gateway.  See Section 3.1 for the detailed requirements.

RFC1122 INTRODUCTION October 1989

1.2 General Considerations 

  There are two important lessons that vendors of Internet host
  software have learned and which a new vendor should consider
  seriously.

  1.2.1  Continuing Internet Evolution

     The enormous growth of the Internet has revealed problems of
     management and scaling in a large datagram-based packet
     communication system.  These problems are being addressed, and
     as a result there will be continuing evolution of the
     specifications described in this document.  These changes will
     be carefully planned and controlled, since there is extensive
     participation in this planning by the vendors and by the
     organizations responsible for operations of the networks.

     Development, evolution, and revision are characteristic of
     computer network protocols today, and this situation will
     persist for some years.  A vendor who develops computer
     communication software for the Internet protocol suite (or any
     other protocol suite!) and then fails to maintain and update
     that software for changing specifications is going to leave a
     trail of unhappy customers.  The Internet is a large
     communication network, and the users are in constant contact
     through it.  Experience has shown that knowledge of
     deficiencies in vendor software propagates quickly through the
     Internet technical community.

  1.2.2  Robustness Principle

     At every layer of the protocols, there is a general rule whose
     application can lead to enormous benefits in robustness and
     interoperability [IP:1]:

            "Be liberal in what you accept, and
             conservative in what you send"

     Software should be written to deal with every conceivable
     error, no matter how unlikely; sooner or later a packet will
     come in with that particular combination of errors and
     attributes, and unless the software is prepared, chaos can
     ensue.  In general, it is best to assume that the network is
     filled with malevolent entities that will send in packets
     designed to have the worst possible effect.  This assumption
     will lead to suitable protective design, although the most
     serious problems in the Internet have been caused by
     unenvisaged mechanisms triggered by low-probability events;

RFC1122 INTRODUCTION October 1989 

     mere human malice would never have taken so devious a course!

     Adaptability to change must be designed into all levels of
     Internet host software.  As a simple example, consider a
     protocol specification that contains an enumeration of values
     for a particular header field -- e.g., a type field, a port
     number, or an error code; this enumeration must be assumed to
     be incomplete.  Thus, if a protocol specification defines four
     possible error codes, the software must not break when a fifth
     code shows up.  An undefined code might be logged (see below),
     but it must not cause a failure.

     The second part of the principle is almost as important:
     software on other hosts may contain deficiencies that make it
     unwise to exploit legal but obscure protocol features.  It is
     unwise to stray far from the obvious and simple, lest untoward
     effects result elsewhere.  A corollary of this is "watch out
     for misbehaving hosts"; host software should be prepared, not
     just to survive other misbehaving hosts, but also to cooperate
     to limit the amount of disruption such hosts can cause to the
     shared communication facility.

  1.2.3  Error Logging

     The Internet includes a great variety of host and gateway
     systems, each implementing many protocols and protocol layers,
     and some of these contain bugs and mis-features in their
     Internet protocol software.  As a result of complexity,
     diversity, and distribution of function, the diagnosis of
     Internet problems is often very difficult.

     Problem diagnosis will be aided if host implementations include
     a carefully designed facility for logging erroneous or
     "strange" protocol events.  It is important to include as much
     diagnostic information as possible when an error is logged.  In
     particular, it is often useful to record the header(s) of a
     packet that caused an error.  However, care must be taken to
     ensure that error logging does not consume prohibitive amounts
     of resources or otherwise interfere with the operation of the
     host.

     There is a tendency for abnormal but harmless protocol events
     to overflow error logging files; this can be avoided by using a
     "circular" log, or by enabling logging only while diagnosing a
     known failure.  It may be useful to filter and count duplicate
     successive messages.  One strategy that seems to work well is:
     (1) always count abnormalities and make such counts accessible
     through the management protocol (see [INTRO:1]); and (2) allow

RFC1122 INTRODUCTION October 1989 

     the logging of a great variety of events to be selectively
     enabled.  For example, it might useful to be able to "log
     everything" or to "log everything for host X".

     Note that different managements may have differing policies
     about the amount of error logging that they want normally
     enabled in a host.  Some will say, "if it doesn't hurt me, I
     don't want to know about it", while others will want to take a
     more watchful and aggressive attitude about detecting and
     removing protocol abnormalities.

  1.2.4  Configuration

     It would be ideal if a host implementation of the Internet
     protocol suite could be entirely self-configuring.  This would
     allow the whole suite to be implemented in ROM or cast into
     silicon, it would simplify diskless workstations, and it would
     be an immense boon to harried LAN administrators as well as
     system vendors.  We have not reached this ideal; in fact, we
     are not even close.

     At many points in this document, you will find a requirement
     that a parameter be a configurable option.  There are several
     different reasons behind such requirements.  In a few cases,
     there is current uncertainty or disagreement about the best
     value, and it may be necessary to update the recommended value
     in the future.  In other cases, the value really depends on
     external factors -- e.g., the size of the host and the
     distribution of its communication load, or the speeds and
     topology of nearby networks -- and self-tuning algorithms are
     unavailable and may be insufficient.  In some cases,
     configurability is needed because of administrative
     requirements.

     Finally, some configuration options are required to communicate
     with obsolete or incorrect implementations of the protocols,
     distributed without sources, that unfortunately persist in many
     parts of the Internet.  To make correct systems coexist with
     these faulty systems, administrators often have to "mis-
     configure" the correct systems.  This problem will correct
     itself gradually as the faulty systems are retired, but it
     cannot be ignored by vendors.

     When we say that a parameter must be configurable, we do not
     intend to require that its value be explicitly read from a
     configuration file at every boot time.  We recommend that
     implementors set up a default for each parameter, so a
     configuration file is only necessary to override those defaults

RFC1122 INTRODUCTION October 1989 

     that are inappropriate in a particular installation.  Thus, the
     configurability requirement is an assurance that it will be
     POSSIBLE to override the default when necessary, even in a
     binary-only or ROM-based product.

     This document requires a particular value for such defaults in
     some cases.  The choice of default is a sensitive issue when
     the configuration item controls the accommodation to existing
     faulty systems.  If the Internet is to converge successfully to
     complete interoperability, the default values built into
     implementations must implement the official protocol, not
     "mis-configurations" to accommodate faulty implementations.
     Although marketing considerations have led some vendors to
     choose mis-configuration defaults, we urge vendors to choose
     defaults that will conform to the standard.

     Finally, we note that a vendor needs to provide adequate
     documentation on all configuration parameters, their limits and
     effects.

1.3 Reading this Document 

  1.3.1  Organization

     Protocol layering, which is generally used as an organizing
     principle in implementing network software, has also been used
     to organize this document.  In describing the rules, we assume
     that an implementation does strictly mirror the layering of the
     protocols.  Thus, the following three major sections specify
     the requirements for the link layer, the internet layer, and
     the transport layer, respectively.  A companion RFC [INTRO:1]
     covers application level software.  This layerist organization
     was chosen for simplicity and clarity.

     However, strict layering is an imperfect model, both for the
     protocol suite and for recommended implementation approaches.
     Protocols in different layers interact in complex and sometimes
     subtle ways, and particular functions often involve multiple
     layers.  There are many design choices in an implementation,
     many of which involve creative "breaking" of strict layering.
     Every implementor is urged to read references [INTRO:7] and
     [INTRO:8].

     This document describes the conceptual service interface
     between layers using a functional ("procedure call") notation,
     like that used in the TCP specification [TCP:1].  A host
     implementation must support the logical information flow

RFC1122 INTRODUCTION October 1989 

     implied by these calls, but need not literally implement the
     calls themselves.  For example, many implementations reflect
     the coupling between the transport layer and the IP layer by
     giving them shared access to common data structures.  These
     data structures, rather than explicit procedure calls, are then
     the agency for passing much of the information that is
     required.

     In general, each major section of this document is organized
     into the following subsections:

     (1)  Introduction

     (2)  Protocol Walk-Through -- considers the protocol
          specification documents section-by-section, correcting
          errors, stating requirements that may be ambiguous or
          ill-defined, and providing further clarification or
          explanation.

     (3)  Specific Issues -- discusses protocol design and
          implementation issues that were not included in the walk-
          through.

     (4)  Interfaces -- discusses the service interface to the next
          higher layer.

     (5)  Summary -- contains a summary of the requirements of the
          section.

     Under many of the individual topics in this document, there is
     parenthetical material labeled "DISCUSSION" or
     "IMPLEMENTATION". This material is intended to give
     clarification and explanation of the preceding requirements
     text.  It also includes some suggestions on possible future
     directions or developments.  The implementation material
     contains suggested approaches that an implementor may want to
     consider.

     The summary sections are intended to be guides and indexes to
     the text, but are necessarily cryptic and incomplete.  The
     summaries should never be used or referenced separately from
     the complete RFC.

  1.3.2  Requirements

     In this document, the words that are used to define the
     significance of each particular requirement are capitalized.

RFC1122 INTRODUCTION October 1989 

     These words are:

     *    "MUST"

          This word or the adjective "REQUIRED" means that the item
          is an absolute requirement of the specification.

     *    "SHOULD"

          This word or the adjective "RECOMMENDED" means that there
          may exist valid reasons in particular circumstances to
          ignore this item, but the full implications should be
          understood and the case carefully weighed before choosing
          a different course.

     *    "MAY"

          This word or the adjective "OPTIONAL" means that this item
          is truly optional.  One vendor may choose to include the
          item because a particular marketplace requires it or
          because it enhances the product, for example; another
          vendor may omit the same item.

     An implementation is not compliant if it fails to satisfy one
     or more of the MUST requirements for the protocols it
     implements.  An implementation that satisfies all the MUST and
     all the SHOULD requirements for its protocols is said to be
     "unconditionally compliant"; one that satisfies all the MUST
     requirements but not all the SHOULD requirements for its
     protocols is said to be "conditionally compliant".

  1.3.3  Terminology

     This document uses the following technical terms:

     Segment
          A segment is the unit of end-to-end transmission in the
          TCP protocol.  A segment consists of a TCP header followed
          by application data.  A segment is transmitted by
          encapsulation inside an IP datagram.

     Message
          In this description of the lower-layer protocols, a
          message is the unit of transmission in a transport layer
          protocol.  In particular, a TCP segment is a message.  A
          message consists of a transport protocol header followed
          by application protocol data.  To be transmitted end-to-

RFC1122 INTRODUCTION October 1989 

          end through the Internet, a message must be encapsulated
          inside a datagram.

     IP Datagram
          An IP datagram is the unit of end-to-end transmission in
          the IP protocol.  An IP datagram consists of an IP header
          followed by transport layer data, i.e., of an IP header
          followed by a message.

          In the description of the internet layer (Section 3), the
          unqualified term "datagram" should be understood to refer
          to an IP datagram.

     Packet
          A packet is the unit of data passed across the interface
          between the internet layer and the link layer.  It
          includes an IP header and data.  A packet may be a
          complete IP datagram or a fragment of an IP datagram.

     Frame
          A frame is the unit of transmission in a link layer
          protocol, and consists of a link-layer header followed by
          a packet.

     Connected Network
          A network to which a host is interfaced is often known as
          the "local network" or the "subnetwork" relative to that
          host.  However, these terms can cause confusion, and
          therefore we use the term "connected network" in this
          document.

     Multihomed
          A host is said to be multihomed if it has multiple IP
          addresses.  For a discussion of multihoming, see Section
          3.3.4 below.

     Physical network interface
          This is a physical interface to a connected network and
          has a (possibly unique) link-layer address.  Multiple
          physical network interfaces on a single host may share the
          same link-layer address, but the address must be unique
          for different hosts on the same physical network.

     Logical [network] interface
          We define a logical [network] interface to be a logical
          path, distinguished by a unique IP address, to a connected
          network.  See Section 3.3.4.

RFC1122 INTRODUCTION October 1989 

     Specific-destination address
          This is the effective destination address of a datagram,
          even if it is broadcast or multicast; see Section 3.2.1.3.

     Path
          At a given moment, all the IP datagrams from a particular
          source host to a particular destination host will
          typically traverse the same sequence of gateways.  We use
          the term "path" for this sequence.  Note that a path is
          uni-directional; it is not unusual to have different paths
          in the two directions between a given host pair.

     MTU
          The maximum transmission unit, i.e., the size of the
          largest packet that can be transmitted.

     The terms frame, packet, datagram, message, and segment are
     illustrated by the following schematic diagrams:

     A. Transmission on connected network:
       _______________________________________________
      | LL hdr | IP hdr |         (data)              |
      |________|________|_____________________________|

       <---------- Frame ----------------------------->
                <----------Packet -------------------->

     B. Before IP fragmentation or after IP reassembly:
                ______________________________________
               | IP hdr | transport| Application Data |
               |________|____hdr___|__________________|

                <--------  Datagram ------------------>
                         <-------- Message ----------->
       or, for TCP:
                ______________________________________
               | IP hdr |  TCP hdr | Application Data |
               |________|__________|__________________|

                <--------  Datagram ------------------>
                         <-------- Segment ----------->

RFC1122 INTRODUCTION October 1989

1.4 Acknowledgments 

  This document incorporates contributions and comments from a large
  group of Internet protocol experts, including representatives of
  university and research labs, vendors, and government agencies.
  It was assembled primarily by the Host Requirements Working Group
  of the Internet Engineering Task Force (IETF).

  The Editor would especially like to acknowledge the tireless
  dedication of the following people, who attended many long
  meetings and generated 3 million bytes of electronic mail over the
  past 18 months in pursuit of this document: Philip Almquist, Dave
  Borman (Cray Research), Noel Chiappa, Dave Crocker (DEC), Steve
  Deering (Stanford), Mike Karels (Berkeley), Phil Karn (Bellcore),
  John Lekashman (NASA), Charles Lynn (BBN), Keith McCloghrie (TWG),
  Paul Mockapetris (ISI), Thomas Narten (Purdue), Craig Partridge
  (BBN), Drew Perkins (CMU), and James Van Bokkelen (FTP Software).

  In addition, the following people made major contributions to the
  effort: Bill Barns (Mitre), Steve Bellovin (AT&T), Mike Brescia
  (BBN), Ed Cain (DCA), Annette DeSchon (ISI), Martin Gross (DCA),
  Phill Gross (NRI), Charles Hedrick (Rutgers), Van Jacobson (LBL),
  John Klensin (MIT), Mark Lottor (SRI), Milo Medin (NASA), Bill
  Melohn (Sun Microsystems), Greg Minshall (Kinetics), Jeff Mogul
  (DEC), John Mullen (CMC), Jon Postel (ISI), John Romkey (Epilogue
  Technology), and Mike StJohns (DCA).  The following also made
  significant contributions to particular areas: Eric Allman
  (Berkeley), Rob Austein (MIT), Art Berggreen (ACC), Keith Bostic
  (Berkeley), Vint Cerf (NRI), Wayne Hathaway (NASA), Matt Korn
  (IBM), Erik Naggum (Naggum Software, Norway), Robert Ullmann
  (Prime Computer), David Waitzman (BBN), Frank Wancho (USA), Arun
  Welch (Ohio State), Bill Westfield (Cisco), and Rayan Zachariassen
  (Toronto).

  We are grateful to all, including any contributors who may have
  been inadvertently omitted from this list.

RFC1122 LINK LAYER October 1989

LINK LAYER

2.1 INTRODUCTION 

  All Internet systems, both hosts and gateways, have the same
  requirements for link layer protocols.  These requirements are
  given in Chapter 3 of "Requirements for Internet Gateways"
  [INTRO:2], augmented with the material in this section.

2.2 PROTOCOL WALK-THROUGH 

  None.

2.3 SPECIFIC ISSUES 

  2.3.1  Trailer Protocol Negotiation

     The trailer protocol [LINK:1] for link-layer encapsulation MAY
     be used, but only when it has been verified that both systems
     (host or gateway) involved in the link-layer communication
     implement trailers.  If the system does not dynamically
     negotiate use of the trailer protocol on a per-destination
     basis, the default configuration MUST disable the protocol.

     DISCUSSION:
          The trailer protocol is a link-layer encapsulation
          technique that rearranges the data contents of packets
          sent on the physical network.  In some cases, trailers
          improve the throughput of higher layer protocols by
          reducing the amount of data copying within the operating
          system.  Higher layer protocols are unaware of trailer
          use, but both the sending and receiving host MUST
          understand the protocol if it is used.

          Improper use of trailers can result in very confusing
          symptoms.  Only packets with specific size attributes are
          encapsulated using trailers, and typically only a small
          fraction of the packets being exchanged have these
          attributes.  Thus, if a system using trailers exchanges
          packets with a system that does not, some packets
          disappear into a black hole while others are delivered
          successfully.

     IMPLEMENTATION:
          On an Ethernet, packets encapsulated with trailers use a
          distinct Ethernet type [LINK:1], and trailer negotiation
          is performed at the time that ARP is used to discover the
          link-layer address of a destination system.

RFC1122 LINK LAYER October 1989 

          Specifically, the ARP exchange is completed in the usual
          manner using the normal IP protocol type, but a host that
          wants to speak trailers will send an additional "trailer
          ARP reply" packet, i.e., an ARP reply that specifies the
          trailer encapsulation protocol type but otherwise has the
          format of a normal ARP reply.  If a host configured to use
          trailers receives a trailer ARP reply message from a
          remote machine, it can add that machine to the list of
          machines that understand trailers, e.g., by marking the
          corresponding entry in the ARP cache.

          Hosts wishing to receive trailer encapsulations send
          trailer ARP replies whenever they complete exchanges of
          normal ARP messages for IP.  Thus, a host that received an
          ARP request for its IP protocol address would send a
          trailer ARP reply in addition to the normal IP ARP reply;
          a host that sent the IP ARP request would send a trailer
          ARP reply when it received the corresponding IP ARP reply.
          In this way, either the requesting or responding host in
          an IP ARP exchange may request that it receive trailer
          encapsulations.

          This scheme, using extra trailer ARP reply packets rather
          than sending an ARP request for the trailer protocol type,
          was designed to avoid a continuous exchange of ARP packets
          with a misbehaving host that, contrary to any
          specification or common sense, responded to an ARP reply
          for trailers with another ARP reply for IP.  This problem
          is avoided by sending a trailer ARP reply in response to
          an IP ARP reply only when the IP ARP reply answers an
          outstanding request; this is true when the hardware
          address for the host is still unknown when the IP ARP
          reply is received.  A trailer ARP reply may always be sent
          along with an IP ARP reply responding to an IP ARP
          request.

  2.3.2  Address Resolution Protocol -- ARP

     2.3.2.1  ARP Cache Validation

        An implementation of the Address Resolution Protocol (ARP)
        [LINK:2] MUST provide a mechanism to flush out-of-date cache
        entries.  If this mechanism involves a timeout, it SHOULD be
        possible to configure the timeout value.

        A mechanism to prevent ARP flooding (repeatedly sending an
        ARP Request for the same IP address, at a high rate) MUST be
        included.  The recommended maximum rate is 1 per second per

RFC1122 LINK LAYER October 1989 

        destination.

        DISCUSSION:
             The ARP specification [LINK:2] suggests but does not
             require a timeout mechanism to invalidate cache entries
             when hosts change their Ethernet addresses.  The
             prevalence of proxy ARP (see Section 2.4 of [INTRO:2])
             has significantly increased the likelihood that cache
             entries in hosts will become invalid, and therefore
             some ARP-cache invalidation mechanism is now required
             for hosts.  Even in the absence of proxy ARP, a long-
             period cache timeout is useful in order to
             automatically correct any bad ARP data that might have
             been cached.

        IMPLEMENTATION:
             Four mechanisms have been used, sometimes in
             combination, to flush out-of-date cache entries.

             (1)  Timeout -- Periodically time out cache entries,
                  even if they are in use.  Note that this timeout
                  should be restarted when the cache entry is
                  "refreshed" (by observing the source fields,
                  regardless of target address, of an ARP broadcast
                  from the system in question).  For proxy ARP
                  situations, the timeout needs to be on the order
                  of a minute.

             (2)  Unicast Poll -- Actively poll the remote host by
                  periodically sending a point-to-point ARP Request
                  to it, and delete the entry if no ARP Reply is
                  received from N successive polls.  Again, the
                  timeout should be on the order of a minute, and
                  typically N is 2.

             (3)  Link-Layer Advice -- If the link-layer driver
                  detects a delivery problem, flush the
                  corresponding ARP cache entry.

             (4)  Higher-layer Advice -- Provide a call from the
                  Internet layer to the link layer to indicate a
                  delivery problem.  The effect of this call would
                  be to invalidate the corresponding cache entry.
                  This call would be analogous to the
                  "ADVISE_DELIVPROB()" call from the transport layer
                  to the Internet layer (see Section 3.4), and in
                  fact the ADVISE_DELIVPROB routine might in turn
                  call the link-layer advice routine to invalidate

RFC1122 LINK LAYER October 1989 

                  the ARP cache entry.

             Approaches (1) and (2) involve ARP cache timeouts on
             the order of a minute or less.  In the absence of proxy
             ARP, a timeout this short could create noticeable
             overhead traffic on a very large Ethernet.  Therefore,
             it may be necessary to configure a host to lengthen the
             ARP cache timeout.

     2.3.2.2  ARP Packet Queue

        The link layer SHOULD save (rather than discard) at least
        one (the latest) packet of each set of packets destined to
        the same unresolved IP address, and transmit the saved
        packet when the address has been resolved.

        DISCUSSION:
             Failure to follow this recommendation causes the first
             packet of every exchange to be lost.  Although higher-
             layer protocols can generally cope with packet loss by
             retransmission, packet loss does impact performance.
             For example, loss of a TCP open request causes the
             initial round-trip time estimate to be inflated.  UDP-
             based applications such as the Domain Name System are
             more seriously affected.

  2.3.3  Ethernet and IEEE 802 Encapsulation

     The IP encapsulation for Ethernets is described in RFC-894
     [LINK:3], while RFC-1042 [LINK:4] describes the IP
     encapsulation for IEEE 802 networks.  RFC-1042 elaborates and
     replaces the discussion in Section 3.4 of [INTRO:2].

     Every Internet host connected to a 10Mbps Ethernet cable:

     o    MUST be able to send and receive packets using RFC-894
          encapsulation;

     o    SHOULD be able to receive RFC-1042 packets, intermixed
          with RFC-894 packets; and

     o    MAY be able to send packets using RFC-1042 encapsulation.

     An Internet host that implements sending both the RFC-894 and
     the RFC-1042 encapsulations MUST provide a configuration switch
     to select which is sent, and this switch MUST default to RFC-
     894.

RFC1122 LINK LAYER October 1989 

     Note that the standard IP encapsulation in RFC-1042 does not
     use the protocol id value (K1=6) that IEEE reserved for IP;
     instead, it uses a value (K1=170) that implies an extension
     (the "SNAP") which can be used to hold the Ether-Type field.
     An Internet system MUST NOT send 802 packets using K1=6.

     Address translation from Internet addresses to link-layer
     addresses on Ethernet and IEEE 802 networks MUST be managed by
     the Address Resolution Protocol (ARP).

     The MTU for an Ethernet is 1500 and for 802.3 is 1492.

     DISCUSSION:
          The IEEE 802.3 specification provides for operation over a
          10Mbps Ethernet cable, in which case Ethernet and IEEE
          802.3 frames can be physically intermixed.  A receiver can
          distinguish Ethernet and 802.3 frames by the value of the
          802.3 Length field; this two-octet field coincides in the
          header with the Ether-Type field of an Ethernet frame.  In
          particular, the 802.3 Length field must be less than or
          equal to 1500, while all valid Ether-Type values are
          greater than 1500.

          Another compatibility problem arises with link-layer
          broadcasts.  A broadcast sent with one framing will not be
          seen by hosts that can receive only the other framing.

          The provisions of this section were designed to provide
          direct interoperation between 894-capable and 1042-capable
          systems on the same cable, to the maximum extent possible.
          It is intended to support the present situation where
          894-only systems predominate, while providing an easy
          transition to a possible future in which 1042-capable
          systems become common.

          Note that 894-only systems cannot interoperate directly
          with 1042-only systems.  If the two system types are set
          up as two different logical networks on the same cable,
          they can communicate only through an IP gateway.
          Furthermore, it is not useful or even possible for a
          dual-format host to discover automatically which format to
          send, because of the problem of link-layer broadcasts.

2.4 LINK/INTERNET LAYER INTERFACE 

  The packet receive interface between the IP layer and the link
  layer MUST include a flag to indicate whether the incoming packet
  was addressed to a link-layer broadcast address.

RFC1122 LINK LAYER October 1989 

  DISCUSSION
       Although the IP layer does not generally know link layer
       addresses (since every different network medium typically has
       a different address format), the broadcast address on a
       broadcast-capable medium is an important special case.  See
       Section 3.2.2, especially the DISCUSSION concerning broadcast
       storms.

  The packet send interface between the IP and link layers MUST
  include the 5-bit TOS field (see Section 3.2.1.6).

  The link layer MUST NOT report a Destination Unreachable error to
  IP solely because there is no ARP cache entry for a destination.

2.5 LINK LAYER REQUIREMENTS SUMMARY 

                                              |       | | | |S| |
                                              |       | | | |H| |F
                                              |       | | | |O|M|o
                                              |       | |S| |U|U|o
                                              |       | |H| |L|S|t
                                              |       |M|O| |D|T|n
                                              |       |U|U|M| | |o
                                              |       |S|L|A|N|N|t
                                              |       |T|D|Y|O|O|t

FEATURE |SECTION| | | |T|T|e

|-------|-|-|-|-|-|-- 

                                              |       | | | | | |

Trailer encapsulation |2.3.1 | | |x| | | Send Trailers by default without negotiation |2.3.1 | | | | |x| ARP |2.3.2 | | | | | | 

 Flush out-of-date ARP cache entries             |2.3.2.1|x| | | | |
 Prevent ARP floods                              |2.3.2.1|x| | | | |
 Cache timeout configurable                      |2.3.2.1| |x| | | |
 Save at least one (latest) unresolved pkt       |2.3.2.2| |x| | | |

Ethernet and IEEE 802 Encapsulation |2.3.3 | | | | | | 

 Host able to:                                   |2.3.3  | | | | | |
Send & receive RFC-894 encapsulation          |2.3.3  |x| | | | |
Receive RFC-1042 encapsulation                |2.3.3  | |x| | | |
Send RFC-1042 encapsulation                   |2.3.3  | | |x| | |
  Then config. sw. to select, RFC-894 dflt    |2.3.3  |x| | | | |
 Send K1=6 encapsulation                         |2.3.3  | | | | |x|
 Use ARP on Ethernet and IEEE 802 nets           |2.3.3  |x| | | | |

Link layer report b'casts to IP layer |2.4 |x| | | | | IP layer pass TOS to link layer |2.4 |x| | | | | No ARP cache entry treated as Dest. Unreach. |2.4 | | | | |x|

RFC1122 INTERNET LAYER October 1989

INTERNET LAYER PROTOCOLS

3.1 INTRODUCTION 

  The Robustness Principle: "Be liberal in what you accept, and
  conservative in what you send" is particularly important in the
  Internet layer, where one misbehaving host can deny Internet
  service to many other hosts.

  The protocol standards used in the Internet layer are:

  o    RFC-791 [IP:1] defines the IP protocol and gives an
       introduction to the architecture of the Internet.

  o    RFC-792 [IP:2] defines ICMP, which provides routing,
       diagnostic and error functionality for IP.  Although ICMP
       messages are encapsulated within IP datagrams, ICMP
       processing is considered to be (and is typically implemented
       as) part of the IP layer.  See Section 3.2.2.

  o    RFC-950 [IP:3] defines the mandatory subnet extension to the
       addressing architecture.

  o    RFC-1112 [IP:4] defines the Internet Group Management
       Protocol IGMP, as part of a recommended extension to hosts
       and to the host-gateway interface to support Internet-wide
       multicasting at the IP level.  See Section 3.2.3.

       The target of an IP multicast may be an arbitrary group of
       Internet hosts.  IP multicasting is designed as a natural
       extension of the link-layer multicasting facilities of some
       networks, and it provides a standard means for local access
       to such link-layer multicasting facilities.

  Other important references are listed in Section 5 of this
  document.

  The Internet layer of host software MUST implement both IP and
  ICMP.  See Section 3.3.7 for the requirements on support of IGMP.

  The host IP layer has two basic functions:  (1) choose the "next
  hop" gateway or host for outgoing IP datagrams and (2) reassemble
  incoming IP datagrams.  The IP layer may also (3) implement
  intentional fragmentation of outgoing datagrams.  Finally, the IP
  layer must (4) provide diagnostic and error functionality.  We
  expect that IP layer functions may increase somewhat in the
  future, as further Internet control and management facilities are
  developed.

RFC1122 INTERNET LAYER October 1989 

  For normal datagrams, the processing is straightforward.  For
  incoming datagrams, the IP layer:

  (1)  verifies that the datagram is correctly formatted;

  (2)  verifies that it is destined to the local host;

  (3)  processes options;

  (4)  reassembles the datagram if necessary; and

  (5)  passes the encapsulated message to the appropriate
       transport-layer protocol module.

  For outgoing datagrams, the IP layer:

  (1)  sets any fields not set by the transport layer;

  (2)  selects the correct first hop on the connected network (a
       process called "routing");

  (3)  fragments the datagram if necessary and if intentional
       fragmentation is implemented (see Section 3.3.3); and

  (4)  passes the packet(s) to the appropriate link-layer driver.

  A host is said to be multihomed if it has multiple IP addresses.
  Multihoming introduces considerable confusion and complexity into
  the protocol suite, and it is an area in which the Internet
  architecture falls seriously short of solving all problems.  There
  are two distinct problem areas in multihoming:

  (1)  Local multihoming --  the host itself is multihomed; or

  (2)  Remote multihoming -- the local host needs to communicate
       with a remote multihomed host.

  At present, remote multihoming MUST be handled at the application
  layer, as discussed in the companion RFC [INTRO:1].  A host MAY
  support local multihoming, which is discussed in this document,
  and in particular in Section 3.3.4.

  Any host that forwards datagrams generated by another host is
  acting as a gateway and MUST also meet the specifications laid out
  in the gateway requirements RFC [INTRO:2].  An Internet host that
  includes embedded gateway code MUST have a configuration switch to
  disable the gateway function, and this switch MUST default to the

RFC1122 INTERNET LAYER October 1989 

  non-gateway mode.  In this mode, a datagram arriving through one
  interface will not be forwarded to another host or gateway (unless
  it is source-routed), regardless of whether the host is single-
  homed or multihomed.  The host software MUST NOT automatically
  move into gateway mode if the host has more than one interface, as
  the operator of the machine may neither want to provide that
  service nor be competent to do so.

  In the following, the action specified in certain cases is to
  "silently discard" a received datagram.  This means that the
  datagram will be discarded without further processing and that the
  host will not send any ICMP error message (see Section 3.2.2) as a
  result.  However, for diagnosis of problems a host SHOULD provide
  the capability of logging the error (see Section 1.2.3), including
  the contents of the silently-discarded datagram, and SHOULD record
  the event in a statistics counter.

  DISCUSSION:
       Silent discard of erroneous datagrams is generally intended
       to prevent "broadcast storms".

3.2 PROTOCOL WALK-THROUGH 

  3.2.1 Internet Protocol -- IP

     3.2.1.1  Version Number: RFC-791 Section 3.1

        A datagram whose version number is not 4 MUST be silently
        discarded.

     3.2.1.2  Checksum: RFC-791 Section 3.1

        A host MUST verify the IP header checksum on every received
        datagram and silently discard every datagram that has a bad
        checksum.

     3.2.1.3  Addressing: RFC-791 Section 3.2

        There are now five classes of IP addresses: Class A through
        Class E.  Class D addresses are used for IP multicasting
        [IP:4], while Class E addresses are reserved for
        experimental use.

        A multicast (Class D) address is a 28-bit logical address
        that stands for a group of hosts, and may be either
        permanent or transient.  Permanent multicast addresses are
        allocated by the Internet Assigned Number Authority
        [INTRO:6], while transient addresses may be allocated

RFC1122 INTERNET LAYER October 1989 

        dynamically to transient groups.  Group membership is
        determined dynamically using IGMP [IP:4].

        We now summarize the important special cases for Class A, B,
        and C IP addresses, using the following notation for an IP
        address:

            { <Network-number>, <Host-number> }

        or
            { <Network-number>, <Subnet-number>, <Host-number> }

        and the notation "-1" for a field that contains all 1 bits.
        This notation is not intended to imply that the 1-bits in an
        address mask need be contiguous.

        (a)  { 0, 0 }

             This host on this network.  MUST NOT be sent, except as
             a source address as part of an initialization procedure
             by which the host learns its own IP address.

             See also Section 3.3.6 for a non-standard use of {0,0}.

        (b)  { 0, <Host-number> }

             Specified host on this network.  It MUST NOT be sent,
             except as a source address as part of an initialization
             procedure by which the host learns its full IP address.

        (c)  { -1, -1 }

             Limited broadcast.  It MUST NOT be used as a source
             address.

             A datagram with this destination address will be
             received by every host on the connected physical
             network but will not be forwarded outside that network.

        (d)  { <Network-number>, -1 }

             Directed broadcast to the specified network.  It MUST
             NOT be used as a source address.

        (e)  { <Network-number>, <Subnet-number>, -1 }

             Directed broadcast to the specified subnet.  It MUST
             NOT be used as a source address.

RFC1122 INTERNET LAYER October 1989 

        (f)  { <Network-number>, -1, -1 }

             Directed broadcast to all subnets of the specified
             subnetted network.  It MUST NOT be used as a source
             address.

        (g)  { 127, <any> }

             Internal host loopback address.  Addresses of this form
             MUST NOT appear outside a host.

        The <Network-number> is administratively assigned so that
        its value will be unique in the entire world.

        IP addresses are not permitted to have the value 0 or -1 for
        any of the <Host-number>, <Network-number>, or <Subnet-
        number> fields (except in the special cases listed above).
        This implies that each of these fields will be at least two
        bits long.

        For further discussion of broadcast addresses, see Section
        3.3.6.

        A host MUST support the subnet extensions to IP [IP:3].  As
        a result, there will be an address mask of the form:
        {-1, -1, 0} associated with each of the host's local IP
        addresses; see Sections 3.2.2.9 and 3.3.1.1.

        When a host sends any datagram, the IP source address MUST
        be one of its own IP addresses (but not a broadcast or
        multicast address).

        A host MUST silently discard an incoming datagram that is
        not destined for the host.  An incoming datagram is destined
        for the host if the datagram's destination address field is:

        (1)  (one of) the host's IP address(es); or

        (2)  an IP broadcast address valid for the connected
             network; or

        (3)  the address for a multicast group of which the host is
             a member on the incoming physical interface.

        For most purposes, a datagram addressed to a broadcast or
        multicast destination is processed as if it had been
        addressed to one of the host's IP addresses; we use the term
        "specific-destination address" for the equivalent local IP

RFC1122 INTERNET LAYER October 1989 

        address of the host.  The specific-destination address is
        defined to be the destination address in the IP header
        unless the header contains a broadcast or multicast address,
        in which case the specific-destination is an IP address
        assigned to the physical interface on which the datagram
        arrived.

        A host MUST silently discard an incoming datagram containing
        an IP source address that is invalid by the rules of this
        section.  This validation could be done in either the IP
        layer or by each protocol in the transport layer.

        DISCUSSION:
             A mis-addressed datagram might be caused by a link-
             layer broadcast of a unicast datagram or by a gateway
             or host that is confused or mis-configured.

             An architectural goal for Internet hosts was to allow
             IP addresses to be featureless 32-bit numbers, avoiding
             algorithms that required a knowledge of the IP address
             format.  Otherwise, any future change in the format or
             interpretation of IP addresses will require host
             software changes.  However, validation of broadcast and
             multicast addresses violates this goal; a few other
             violations are described elsewhere in this document.

             Implementers should be aware that applications
             depending upon the all-subnets directed broadcast
             address (f) may be unusable on some networks.  All-
             subnets broadcast is not widely implemented in vendor
             gateways at present, and even when it is implemented, a
             particular network administration may disable it in the
             gateway configuration.

     3.2.1.4  Fragmentation and Reassembly: RFC-791 Section 3.2

        The Internet model requires that every host support
        reassembly.  See Sections 3.3.2 and 3.3.3 for the
        requirements on fragmentation and reassembly.

     3.2.1.5  Identification: RFC-791 Section 3.2

        When sending an identical copy of an earlier datagram, a
        host MAY optionally retain the same Identification field in
        the copy.

RFC1122 INTERNET LAYER October 1989 

        DISCUSSION:
             Some Internet protocol experts have maintained that
             when a host sends an identical copy of an earlier
             datagram, the new copy should contain the same
             Identification value as the original.  There are two
             suggested advantages:  (1) if the datagrams are
             fragmented and some of the fragments are lost, the
             receiver may be able to reconstruct a complete datagram
             from fragments of the original and the copies; (2) a
             congested gateway might use the IP Identification field
             (and Fragment Offset) to discard duplicate datagrams
             from the queue.

             However, the observed patterns of datagram loss in the
             Internet do not favor the probability of retransmitted
             fragments filling reassembly gaps, while other
             mechanisms (e.g., TCP repacketizing upon
             retransmission) tend to prevent retransmission of an
             identical datagram [IP:9].  Therefore, we believe that
             retransmitting the same Identification field is not
             useful.  Also, a connectionless transport protocol like
             UDP would require the cooperation of the application
             programs to retain the same Identification value in
             identical datagrams.

     3.2.1.6  Type-of-Service: RFC-791 Section 3.2

        The "Type-of-Service" byte in the IP header is divided into
        two sections:  the Precedence field (high-order 3 bits), and
        a field that is customarily called "Type-of-Service" or
        "TOS" (low-order 5 bits).  In this document, all references
        to "TOS" or the "TOS field" refer to the low-order 5 bits
        only.

        The Precedence field is intended for Department of Defense
        applications of the Internet protocols.  The use of non-zero
        values in this field is outside the scope of this document
        and the IP standard specification.  Vendors should consult
        the Defense Communication Agency (DCA) for guidance on the
        IP Precedence field and its implications for other protocol
        layers.  However, vendors should note that the use of
        precedence will most likely require that its value be passed
        between protocol layers in just the same way as the TOS
        field is passed.

        The IP layer MUST provide a means for the transport layer to
        set the TOS field of every datagram that is sent; the
        default is all zero bits.  The IP layer SHOULD pass received

RFC1122 INTERNET LAYER October 1989 

        TOS values up to the transport layer.

        The particular link-layer mappings of TOS contained in RFC-
        795 SHOULD NOT be implemented.

        DISCUSSION:
             While the TOS field has been little used in the past,
             it is expected to play an increasing role in the near
             future.  The TOS field is expected to be used to
             control two aspects of gateway operations: routing and
             queueing algorithms.  See Section 2 of [INTRO:1] for
             the requirements on application programs to specify TOS
             values.

             The TOS field may also be mapped into link-layer
             service selectors.  This has been applied to provide
             effective sharing of serial lines by different classes
             of TCP traffic, for example.  However, the mappings
             suggested in RFC-795 for networks that were included in
             the Internet as of 1981 are now obsolete.

     3.2.1.7  Time-to-Live: RFC-791 Section 3.2

        A host MUST NOT send a datagram with a Time-to-Live (TTL)
        value of zero.

        A host MUST NOT discard a datagram just because it was
        received with TTL less than 2.

        The IP layer MUST provide a means for the transport layer to
        set the TTL field of every datagram that is sent.  When a
        fixed TTL value is used, it MUST be configurable.  The
        current suggested value will be published in the "Assigned
        Numbers" RFC.

        DISCUSSION:
             The TTL field has two functions: limit the lifetime of
             TCP segments (see RFC-793 [TCP:1], p. 28), and
             terminate Internet routing loops.  Although TTL is a
             time in seconds, it also has some attributes of a hop-
             count, since each gateway is required to reduce the TTL
             field by at least one.

             The intent is that TTL expiration will cause a datagram
             to be discarded by a gateway but not by the destination
             host; however, hosts that act as gateways by forwarding
             datagrams must follow the gateway rules for TTL.

RFC1122 INTERNET LAYER October 1989 

             A higher-layer protocol may want to set the TTL in
             order to implement an "expanding scope" search for some
             Internet resource.  This is used by some diagnostic
             tools, and is expected to be useful for locating the
             "nearest" server of a given class using IP
             multicasting, for example.  A particular transport
             protocol may also want to specify its own TTL bound on
             maximum datagram lifetime.

             A fixed value must be at least big enough for the
             Internet "diameter," i.e., the longest possible path.
             A reasonable value is about twice the diameter, to
             allow for continued Internet growth.

     3.2.1.8  Options: RFC-791 Section 3.2

        There MUST be a means for the transport layer to specify IP
        options to be included in transmitted IP datagrams (see
        Section 3.4).

        All IP options (except NOP or END-OF-LIST) received in
        datagrams MUST be passed to the transport layer (or to ICMP
        processing when the datagram is an ICMP message).  The IP
        and transport layer MUST each interpret those IP options
        that they understand and silently ignore the others.

        Later sections of this document discuss specific IP option
        support required by each of ICMP, TCP, and UDP.

        DISCUSSION:
             Passing all received IP options to the transport layer
             is a deliberate "violation of strict layering" that is
             designed to ease the introduction of new transport-
             relevant IP options in the future.  Each layer must
             pick out any options that are relevant to its own
             processing and ignore the rest.  For this purpose,
             every IP option except NOP and END-OF-LIST will include
             a specification of its own length.

             This document does not define the order in which a
             receiver must process multiple options in the same IP
             header.  Hosts sending multiple options must be aware
             that this introduces an ambiguity in the meaning of
             certain options when combined with a source-route
             option.

        IMPLEMENTATION:
             The IP layer must not crash as the result of an option

RFC1122 INTERNET LAYER October 1989 

             length that is outside the possible range.  For
             example, erroneous option lengths have been observed to
             put some IP implementations into infinite loops.

        Here are the requirements for specific IP options:

        (a)  Security Option

             Some environments require the Security option in every
             datagram; such a requirement is outside the scope of
             this document and the IP standard specification.  Note,
             however, that the security options described in RFC-791
             and RFC-1038 are obsolete.  For DoD applications,
             vendors should consult [IP:8] for guidance.

        (b)  Stream Identifier Option

             This option is obsolete; it SHOULD NOT be sent, and it
             MUST be silently ignored if received.

        (c)  Source Route Options

             A host MUST support originating a source route and MUST
             be able to act as the final destination of a source
             route.

             If host receives a datagram containing a completed
             source route (i.e., the pointer points beyond the last
             field), the datagram has reached its final destination;
             the option as received (the recorded route) MUST be
             passed up to the transport layer (or to ICMP message
             processing).  This recorded route will be reversed and
             used to form a return source route for reply datagrams
             (see discussion of IP Options in Section 4).  When a
             return source route is built, it MUST be correctly
             formed even if the recorded route included the source
             host (see case (B) in the discussion below).

             An IP header containing more than one Source Route
             option MUST NOT be sent; the effect on routing of
             multiple Source Route options is implementation-
             specific.

             Section 3.3.5 presents the rules for a host acting as
             an intermediate hop in a source route, i.e., forwarding

RFC1122 INTERNET LAYER October 1989 

             a source-routed datagram.

             DISCUSSION:
                  If a source-routed datagram is fragmented, each
                  fragment will contain a copy of the source route.
                  Since the processing of IP options (including a
                  source route) must precede reassembly, the
                  original datagram will not be reassembled until
                  the final destination is reached.

                  Suppose a source routed datagram is to be routed
                  from host S to host D via gateways G1, G2, ... Gn.
                  There was an ambiguity in the specification over
                  whether the source route option in a datagram sent
                  out by S should be (A) or (B):

                      (A):  {>>G2, G3, ... Gn, D}     <--- CORRECT

                      (B):  {S, >>G2, G3, ... Gn, D}  <---- WRONG

                  (where >> represents the pointer).  If (A) is
                  sent, the datagram received at D will contain the
                  option: {G1, G2, ... Gn >>}, with S and D as the
                  IP source and destination addresses.  If (B) were
                  sent, the datagram received at D would again
                  contain S and D as the same IP source and
                  destination addresses, but the option would be:
                  {S, G1, ...Gn >>}; i.e., the originating host
                  would be the first hop in the route.

        (d)  Record Route Option

             Implementation of originating and processing the Record
             Route option is OPTIONAL.

        (e)  Timestamp Option

             Implementation of originating and processing the
             Timestamp option is OPTIONAL.  If it is implemented,
             the following rules apply:

             o    The originating host MUST record a timestamp in a
                  Timestamp option whose Internet address fields are
                  not pre-specified or whose first pre-specified
                  address is the host's interface address.

RFC1122 INTERNET LAYER October 1989 

             o    The destination host MUST (if possible) add the
                  current timestamp to a Timestamp option before
                  passing the option to the transport layer or to
                  ICMP for processing.

             o    A timestamp value MUST follow the rules given in
                  Section 3.2.2.8 for the ICMP Timestamp message.

  3.2.2 Internet Control Message Protocol -- ICMP

     ICMP messages are grouped into two classes.

     *
          ICMP error messages:

           Destination Unreachable   (see Section 3.2.2.1)
           Redirect                  (see Section 3.2.2.2)
           Source Quench             (see Section 3.2.2.3)
           Time Exceeded             (see Section 3.2.2.4)
           Parameter Problem         (see Section 3.2.2.5)

     *
          ICMP query messages:

            Echo                     (see Section 3.2.2.6)
            Information              (see Section 3.2.2.7)
            Timestamp                (see Section 3.2.2.8)
            Address Mask             (see Section 3.2.2.9)

     If an ICMP message of unknown type is received, it MUST be
     silently discarded.

     Every ICMP error message includes the Internet header and at
     least the first 8 data octets of the datagram that triggered
     the error; more than 8 octets MAY be sent; this header and data
     MUST be unchanged from the received datagram.

     In those cases where the Internet layer is required to pass an
     ICMP error message to the transport layer, the IP protocol
     number MUST be extracted from the original header and used to
     select the appropriate transport protocol entity to handle the
     error.

     An ICMP error message SHOULD be sent with normal (i.e., zero)
     TOS bits.

RFC1122 INTERNET LAYER October 1989 

     An ICMP error message MUST NOT be sent as the result of
     receiving:

     *    an ICMP error message, or

     *    a datagram destined to an IP broadcast or IP multicast
          address, or

     *    a datagram sent as a link-layer broadcast, or

     *    a non-initial fragment, or

     *    a datagram whose source address does not define a single
          host -- e.g., a zero address, a loopback address, a
          broadcast address, a multicast address, or a Class E
          address.

     NOTE: THESE RESTRICTIONS TAKE PRECEDENCE OVER ANY REQUIREMENT
     ELSEWHERE IN THIS DOCUMENT FOR SENDING ICMP ERROR MESSAGES.

     DISCUSSION:
          These rules will prevent the "broadcast storms" that have
          resulted from hosts returning ICMP error messages in
          response to broadcast datagrams.  For example, a broadcast
          UDP segment to a non-existent port could trigger a flood
          of ICMP Destination Unreachable datagrams from all
          machines that do not have a client for that destination
          port.  On a large Ethernet, the resulting collisions can
          render the network useless for a second or more.

          Every datagram that is broadcast on the connected network
          should have a valid IP broadcast address as its IP
          destination (see Section 3.3.6).  However, some hosts
          violate this rule.  To be certain to detect broadcast
          datagrams, therefore, hosts are required to check for a
          link-layer broadcast as well as an IP-layer broadcast
          address.

     IMPLEMENTATION:
          This requires that the link layer inform the IP layer when
          a link-layer broadcast datagram has been received; see
          Section 2.4.

     3.2.2.1  Destination Unreachable: RFC-792

        The following additional codes are hereby defined:

                6 = destination network unknown

RFC1122 INTERNET LAYER October 1989 

                7 = destination host unknown

                8 = source host isolated

                9 = communication with destination network
                        administratively prohibited

               10 = communication with destination host
                        administratively prohibited

               11 = network unreachable for type of service

               12 = host unreachable for type of service

        A host SHOULD generate Destination Unreachable messages with
        code:

        2    (Protocol Unreachable), when the designated transport
             protocol is not supported; or

        3    (Port Unreachable), when the designated transport
             protocol (e.g., UDP) is unable to demultiplex the
             datagram but has no protocol mechanism to inform the
             sender.

        A Destination Unreachable message that is received MUST be
        reported to the transport layer.  The transport layer SHOULD
        use the information appropriately; for example, see Sections
        4.1.3.3, 4.2.3.9, and 4.2.4 below.  A transport protocol
        that has its own mechanism for notifying the sender that a
        port is unreachable (e.g., TCP, which sends RST segments)
        MUST nevertheless accept an ICMP Port Unreachable for the
        same purpose.

        A Destination Unreachable message that is received with code
        0 (Net), 1 (Host), or 5 (Bad Source Route) may result from a
        routing transient and MUST therefore be interpreted as only
        a hint, not proof, that the specified destination is
        unreachable [IP:11].  For example, it MUST NOT be used as
        proof of a dead gateway (see Section 3.3.1).

     3.2.2.2  Redirect: RFC-792

        A host SHOULD NOT send an ICMP Redirect message; Redirects
        are to be sent only by gateways.

        A host receiving a Redirect message MUST update its routing
        information accordingly.  Every host MUST be prepared to

RFC1122 INTERNET LAYER October 1989 

        accept both Host and Network Redirects and to process them
        as described in Section 3.3.1.2 below.

        A Redirect message SHOULD be silently discarded if the new
        gateway address it specifies is not on the same connected
        (sub-) net through which the Redirect arrived [INTRO:2,
        Appendix A], or if the source of the Redirect is not the
        current first-hop gateway for the specified destination (see
        Section 3.3.1).

     3.2.2.3  Source Quench: RFC-792

        A host MAY send a Source Quench message if it is
        approaching, or has reached, the point at which it is forced
        to discard incoming datagrams due to a shortage of
        reassembly buffers or other resources.  See Section 2.2.3 of
        [INTRO:2] for suggestions on when to send Source Quench.

        If a Source Quench message is received, the IP layer MUST
        report it to the transport layer (or ICMP processing). In
        general, the transport or application layer SHOULD implement
        a mechanism to respond to Source Quench for any protocol
        that can send a sequence of datagrams to the same
        destination and which can reasonably be expected to maintain
        enough state information to make this feasible.  See Section
        4 for the handling of Source Quench by TCP and UDP.

        DISCUSSION:
             A Source Quench may be generated by the target host or
             by some gateway in the path of a datagram.  The host
             receiving a Source Quench should throttle itself back
             for a period of time, then gradually increase the
             transmission rate again.  The mechanism to respond to
             Source Quench may be in the transport layer (for
             connection-oriented protocols like TCP) or in the
             application layer (for protocols that are built on top
             of UDP).

             A mechanism has been proposed [IP:14] to make the IP
             layer respond directly to Source Quench by controlling
             the rate at which datagrams are sent, however, this
             proposal is currently experimental and not currently
             recommended.

     3.2.2.4  Time Exceeded: RFC-792

        An incoming Time Exceeded message MUST be passed to the
        transport layer.

RFC1122 INTERNET LAYER October 1989 

        DISCUSSION:
             A gateway will send a Time Exceeded Code 0 (In Transit)
             message when it discards a datagram due to an expired
             TTL field.  This indicates either a gateway routing
             loop or too small an initial TTL value.

             A host may receive a Time Exceeded Code 1 (Reassembly
             Timeout) message from a destination host that has timed
             out and discarded an incomplete datagram; see Section
             3.3.2 below.  In the future, receipt of this message
             might be part of some "MTU discovery" procedure, to
             discover the maximum datagram size that can be sent on
             the path without fragmentation.

     3.2.2.5  Parameter Problem: RFC-792

        A host SHOULD generate Parameter Problem messages.  An
        incoming Parameter Problem message MUST be passed to the
        transport layer, and it MAY be reported to the user.

        DISCUSSION:
             The ICMP Parameter Problem message is sent to the
             source host for any problem not specifically covered by
             another ICMP message.  Receipt of a Parameter Problem
             message generally indicates some local or remote
             implementation error.

        A new variant on the Parameter Problem message is hereby
        defined:
          Code 1 = required option is missing.

        DISCUSSION:
             This variant is currently in use in the military
             community for a missing security option.

     3.2.2.6  Echo Request/Reply: RFC-792

        Every host MUST implement an ICMP Echo server function that
        receives Echo Requests and sends corresponding Echo Replies.
        A host SHOULD also implement an application-layer interface
        for sending an Echo Request and receiving an Echo Reply, for
        diagnostic purposes.

        An ICMP Echo Request destined to an IP broadcast or IP
        multicast address MAY be silently discarded.

RFC1122 INTERNET LAYER October 1989 

        DISCUSSION:
             This neutral provision results from a passionate debate
             between those who feel that ICMP Echo to a broadcast
             address provides a valuable diagnostic capability and
             those who feel that misuse of this feature can too
             easily create packet storms.

        The IP source address in an ICMP Echo Reply MUST be the same
        as the specific-destination address (defined in Section
        3.2.1.3) of the corresponding ICMP Echo Request message.

        Data received in an ICMP Echo Request MUST be entirely
        included in the resulting Echo Reply.  However, if sending
        the Echo Reply requires intentional fragmentation that is
        not implemented, the datagram MUST be truncated to maximum
        transmission size (see Section 3.3.3) and sent.

        Echo Reply messages MUST be passed to the ICMP user
        interface, unless the corresponding Echo Request originated
        in the IP layer.

        If a Record Route and/or Time Stamp option is received in an
        ICMP Echo Request, this option (these options) SHOULD be
        updated to include the current host and included in the IP
        header of the Echo Reply message, without "truncation".
        Thus, the recorded route will be for the entire round trip.

        If a Source Route option is received in an ICMP Echo
        Request, the return route MUST be reversed and used as a
        Source Route option for the Echo Reply message.

     3.2.2.7  Information Request/Reply: RFC-792

        A host SHOULD NOT implement these messages.

        DISCUSSION:
             The Information Request/Reply pair was intended to
             support self-configuring systems such as diskless
             workstations, to allow them to discover their IP
             network numbers at boot time.  However, the RARP and
             BOOTP protocols provide better mechanisms for a host to
             discover its own IP address.

     3.2.2.8  Timestamp and Timestamp Reply: RFC-792

        A host MAY implement Timestamp and Timestamp Reply.  If they
        are implemented, the following rules MUST be followed.

RFC1122 INTERNET LAYER October 1989 

        o    The ICMP Timestamp server function returns a Timestamp
             Reply to every Timestamp message that is received.  If
             this function is implemented, it SHOULD be designed for
             minimum variability in delay (e.g., implemented in the
             kernel to avoid delay in scheduling a user process).

        The following cases for Timestamp are to be handled
        according to the corresponding rules for ICMP Echo:

        o    An ICMP Timestamp Request message to an IP broadcast or
             IP multicast address MAY be silently discarded.

        o    The IP source address in an ICMP Timestamp Reply MUST
             be the same as the specific-destination address of the
             corresponding Timestamp Request message.

        o    If a Source-route option is received in an ICMP Echo
             Request, the return route MUST be reversed and used as
             a Source Route option for the Timestamp Reply message.

        o    If a Record Route and/or Timestamp option is received
             in a Timestamp Request, this (these) option(s) SHOULD
             be updated to include the current host and included in
             the IP header of the Timestamp Reply message.

        o    Incoming Timestamp Reply messages MUST be passed up to
             the ICMP user interface.

        The preferred form for a timestamp value (the "standard
        value") is in units of milliseconds since midnight Universal
        Time.  However, it may be difficult to provide this value
        with millisecond resolution.  For example, many systems use
        clocks that update only at line frequency, 50 or 60 times
        per second.  Therefore, some latitude is allowed in a
        "standard value":

        (a)  A "standard value" MUST be updated at least 15 times
             per second (i.e., at most the six low-order bits of the
             value may be undefined).

        (b)  The accuracy of a "standard value" MUST approximate
             that of operator-set CPU clocks, i.e., correct within a
             few minutes.

RFC1122 INTERNET LAYER October 1989 

     3.2.2.9  Address Mask Request/Reply: RFC-950

        A host MUST support the first, and MAY implement all three,
        of the following methods for determining the address mask(s)
        corresponding to its IP address(es):

        (1)  static configuration information;

        (2)  obtaining the address mask(s) dynamically as a side-
             effect of the system initialization process (see
             [INTRO:1]); and

        (3)  sending ICMP Address Mask Request(s) and receiving ICMP
             Address Mask Reply(s).

        The choice of method to be used in a particular host MUST be
        configurable.

        When method (3), the use of Address Mask messages, is
        enabled, then:

        (a)  When it initializes, the host MUST broadcast an Address
             Mask Request message on the connected network
             corresponding to the IP address.  It MUST retransmit
             this message a small number of times if it does not
             receive an immediate Address Mask Reply.

        (b)  Until it has received an Address Mask Reply, the host
             SHOULD assume a mask appropriate for the address class
             of the IP address, i.e., assume that the connected
             network is not subnetted.

        (c)  The first Address Mask Reply message received MUST be
             used to set the address mask corresponding to the
             particular local IP address.  This is true even if the
             first Address Mask Reply message is "unsolicited", in
             which case it will have been broadcast and may arrive
             after the host has ceased to retransmit Address Mask
             Requests.  Once the mask has been set by an Address
             Mask Reply, later Address Mask Reply messages MUST be
             (silently) ignored.

        Conversely, if Address Mask messages are disabled, then no
        ICMP Address Mask Requests will be sent, and any ICMP
        Address Mask Replies received for that local IP address MUST
        be (silently) ignored.

        A host SHOULD make some reasonableness check on any address

RFC1122 INTERNET LAYER October 1989 

        mask it installs; see IMPLEMENTATION section below.

        A system MUST NOT send an Address Mask Reply unless it is an
        authoritative agent for address masks.  An authoritative
        agent may be a host or a gateway, but it MUST be explicitly
        configured as a address mask agent.  Receiving an address
        mask via an Address Mask Reply does not give the receiver
        authority and MUST NOT be used as the basis for issuing
        Address Mask Replies.

        With a statically configured address mask, there SHOULD be
        an additional configuration flag that determines whether the
        host is to act as an authoritative agent for this mask,
        i.e., whether it will answer Address Mask Request messages
        using this mask.

        If it is configured as an agent, the host MUST broadcast an
        Address Mask Reply for the mask on the appropriate interface
        when it initializes.

        See "System Initialization" in [INTRO:1] for more
        information about the use of Address Mask Request/Reply
        messages.

        DISCUSSION
             Hosts that casually send Address Mask Replies with
             invalid address masks have often been a serious
             nuisance.  To prevent this, Address Mask Replies ought
             to be sent only by authoritative agents that have been
             selected by explicit administrative action.

             When an authoritative agent receives an Address Mask
             Request message, it will send a unicast Address Mask
             Reply to the source IP address.  If the network part of
             this address is zero (see (a) and (b) in 3.2.1.3), the
             Reply will be broadcast.

             Getting no reply to its Address Mask Request messages,
             a host will assume there is no agent and use an
             unsubnetted mask, but the agent may be only temporarily
             unreachable.  An agent will broadcast an unsolicited
             Address Mask Reply whenever it initializes, in order to
             update the masks of all hosts that have initialized in
             the meantime.

        IMPLEMENTATION:
             The following reasonableness check on an address mask
             is suggested: the mask is not all 1 bits, and it is

RFC1122 INTERNET LAYER October 1989 

             either zero or else the 8 highest-order bits are on.

  3.2.3  Internet Group Management Protocol IGMP

     IGMP [IP:4] is a protocol used between hosts and gateways on a
     single network to establish hosts' membership in particular
     multicast groups.  The gateways use this information, in
     conjunction with a multicast routing protocol, to support IP
     multicasting across the Internet.

     At this time, implementation of IGMP is OPTIONAL; see Section
     3.3.7 for more information.  Without IGMP, a host can still
     participate in multicasting local to its connected networks.

3.3 SPECIFIC ISSUES 

  3.3.1  Routing Outbound Datagrams

     The IP layer chooses the correct next hop for each datagram it
     sends.  If the destination is on a connected network, the
     datagram is sent directly to the destination host; otherwise,
     it has to be routed to a gateway on a connected network.

     3.3.1.1  Local/Remote Decision

        To decide if the destination is on a connected network, the
        following algorithm MUST be used [see IP:3]:

        (a)  The address mask (particular to a local IP address for
             a multihomed host) is a 32-bit mask that selects the
             network number and subnet number fields of the
             corresponding IP address.

        (b)  If the IP destination address bits extracted by the
             address mask match the IP source address bits extracted
             by the same mask, then the destination is on the
             corresponding connected network, and the datagram is to
             be transmitted directly to the destination host.

        (c)  If not, then the destination is accessible only through
             a gateway.  Selection of a gateway is described below
             (3.3.1.2).

        A special-case destination address is handled as follows:

        *    For a limited broadcast or a multicast address, simply
             pass the datagram to the link layer for the appropriate
             interface.

RFC1122 INTERNET LAYER October 1989 

        *    For a (network or subnet) directed broadcast, the
             datagram can use the standard routing algorithms.

        The host IP layer MUST operate correctly in a minimal
        network environment, and in particular, when there are no
        gateways.  For example, if the IP layer of a host insists on
        finding at least one gateway to initialize, the host will be
        unable to operate on a single isolated broadcast net.

     3.3.1.2  Gateway Selection

        To efficiently route a series of datagrams to the same
        destination, the source host MUST keep a "route cache" of
        mappings to next-hop gateways.  A host uses the following
        basic algorithm on this cache to route a datagram; this
        algorithm is designed to put the primary routing burden on
        the gateways [IP:11].

        (a)  If the route cache contains no information for a
             particular destination, the host chooses a "default"
             gateway and sends the datagram to it.  It also builds a
             corresponding Route Cache entry.

        (b)  If that gateway is not the best next hop to the
             destination, the gateway will forward the datagram to
             the best next-hop gateway and return an ICMP Redirect
             message to the source host.

        (c)  When it receives a Redirect, the host updates the
             next-hop gateway in the appropriate route cache entry,
             so later datagrams to the same destination will go
             directly to the best gateway.

        Since the subnet mask appropriate to the destination address
        is generally not known, a Network Redirect message SHOULD be
        treated identically to a Host Redirect message; i.e., the
        cache entry for the destination host (only) would be updated
        (or created, if an entry for that host did not exist) for
        the new gateway.

        DISCUSSION:
             This recommendation is to protect against gateways that
             erroneously send Network Redirects for a subnetted
             network, in violation of the gateway requirements
             [INTRO:2].

        When there is no route cache entry for the destination host
        address (and the destination is not on the connected

RFC1122 INTERNET LAYER October 1989 

        network), the IP layer MUST pick a gateway from its list of
        "default" gateways.  The IP layer MUST support multiple
        default gateways.

        As an extra feature, a host IP layer MAY implement a table
        of "static routes".  Each such static route MAY include a
        flag specifying whether it may be overridden by ICMP
        Redirects.

        DISCUSSION:
             A host generally needs to know at least one default
             gateway to get started.  This information can be
             obtained from a configuration file or else from the
             host startup sequence, e.g., the BOOTP protocol (see
             [INTRO:1]).

             It has been suggested that a host can augment its list
             of default gateways by recording any new gateways it
             learns about.  For example, it can record every gateway
             to which it is ever redirected.  Such a feature, while
             possibly useful in some circumstances, may cause
             problems in other cases (e.g., gateways are not all
             equal), and it is not recommended.

             A static route is typically a particular preset mapping
             from destination host or network into a particular
             next-hop gateway; it might also depend on the Type-of-
             Service (see next section).  Static routes would be set
             up by system administrators to override the normal
             automatic routing mechanism, to handle exceptional
             situations.  However, any static routing information is
             a potential source of failure as configurations change
             or equipment fails.

     3.3.1.3  Route Cache

        Each route cache entry needs to include the following
        fields:

        (1)  Local IP address (for a multihomed host)

        (2)  Destination IP address

        (3)  Type(s)-of-Service

        (4)  Next-hop gateway IP address

        Field (2) MAY be the full IP address of the destination

RFC1122 INTERNET LAYER October 1989 

        host, or only the destination network number.  Field (3),
        the TOS, SHOULD be included.

        See Section 3.3.4.2 for a discussion of the implications of
        multihoming for the lookup procedure in this cache.

        DISCUSSION:
             Including the Type-of-Service field in the route cache
             and considering it in the host route algorithm will
             provide the necessary mechanism for the future when
             Type-of-Service routing is commonly used in the
             Internet.  See Section 3.2.1.6.

             Each route cache entry defines the endpoints of an
             Internet path.  Although the connecting path may change
             dynamically in an arbitrary way, the transmission
             characteristics of the path tend to remain
             approximately constant over a time period longer than a
             single typical host-host transport connection.
             Therefore, a route cache entry is a natural place to
             cache data on the properties of the path.  Examples of
             such properties might be the maximum unfragmented
             datagram size (see Section 3.3.3), or the average
             round-trip delay measured by a transport protocol.
             This data will generally be both gathered and used by a
             higher layer protocol, e.g., by TCP, or by an
             application using UDP.  Experiments are currently in
             progress on caching path properties in this manner.

             There is no consensus on whether the route cache should
             be keyed on destination host addresses alone, or allow
             both host and network addresses.  Those who favor the
             use of only host addresses argue that:

             (1)  As required in Section 3.3.1.2, Redirect messages
                  will generally result in entries keyed on
                  destination host addresses; the simplest and most
                  general scheme would be to use host addresses
                  always.

             (2)  The IP layer may not always know the address mask
                  for a network address in a complex subnetted
                  environment.

             (3)  The use of only host addresses allows the
                  destination address to be used as a pure 32-bit
                  number, which may allow the Internet architecture
                  to be more easily extended in the future without

RFC1122 INTERNET LAYER October 1989 

                  any change to the hosts.

             The opposing view is that allowing a mixture of
             destination hosts and networks in the route cache:

             (1)  Saves memory space.

             (2)  Leads to a simpler data structure, easily
                  combining the cache with the tables of default and
                  static routes (see below).

             (3)  Provides a more useful place to cache path
                  properties, as discussed earlier.

        IMPLEMENTATION:
             The cache needs to be large enough to include entries
             for the maximum number of destination hosts that may be
             in use at one time.

             A route cache entry may also include control
             information used to choose an entry for replacement.
             This might take the form of a "recently used" bit, a
             use count, or a last-used timestamp, for example.  It
             is recommended that it include the time of last
             modification of the entry, for diagnostic purposes.

             An implementation may wish to reduce the overhead of
             scanning the route cache for every datagram to be
             transmitted.  This may be accomplished with a hash
             table to speed the lookup, or by giving a connection-
             oriented transport protocol a "hint" or temporary
             handle on the appropriate cache entry, to be passed to
             the IP layer with each subsequent datagram.

             Although we have described the route cache, the lists
             of default gateways, and a table of static routes as
             conceptually distinct, in practice they may be combined
             into a single "routing table" data structure.

     3.3.1.4  Dead Gateway Detection

        The IP layer MUST be able to detect the failure of a "next-
        hop" gateway that is listed in its route cache and to choose
        an alternate gateway (see Section 3.3.1.5).

        Dead gateway detection is covered in some detail in RFC-816
        [IP:11]. Experience to date has not produced a complete

RFC1122 INTERNET LAYER October 1989 

        algorithm which is totally satisfactory, though it has
        identified several forbidden paths and promising techniques.

        *    A particular gateway SHOULD NOT be used indefinitely in
             the absence of positive indications that it is
             functioning.

        *    Active probes such as "pinging" (i.e., using an ICMP
             Echo Request/Reply exchange) are expensive and scale
             poorly.  In particular, hosts MUST NOT actively check
             the status of a first-hop gateway by simply pinging the
             gateway continuously.

        *    Even when it is the only effective way to verify a
             gateway's status, pinging MUST be used only when
             traffic is being sent to the gateway and when there is
             no other positive indication to suggest that the
             gateway is functioning.

        *    To avoid pinging, the layers above and/or below the
             Internet layer SHOULD be able to give "advice" on the
             status of route cache entries when either positive
             (gateway OK) or negative (gateway dead) information is
             available.

        DISCUSSION:
             If an implementation does not include an adequate
             mechanism for detecting a dead gateway and re-routing,
             a gateway failure may cause datagrams to apparently
             vanish into a "black hole".  This failure can be
             extremely confusing for users and difficult for network
             personnel to debug.

             The dead-gateway detection mechanism must not cause
             unacceptable load on the host, on connected networks,
             or on first-hop gateway(s).  The exact constraints on
             the timeliness of dead gateway detection and on
             acceptable load may vary somewhat depending on the
             nature of the host's mission, but a host generally
             needs to detect a failed first-hop gateway quickly
             enough that transport-layer connections will not break
             before an alternate gateway can be selected.

             Passing advice from other layers of the protocol stack
             complicates the interfaces between the layers, but it
             is the preferred approach to dead gateway detection.
             Advice can come from almost any part of the IP/TCP

RFC1122 INTERNET LAYER October 1989 

             architecture, but it is expected to come primarily from
             the transport and link layers.  Here are some possible
             sources for gateway advice:

             o    TCP or any connection-oriented transport protocol
                  should be able to give negative advice, e.g.,
                  triggered by excessive retransmissions.

             o    TCP may give positive advice when (new) data is
                  acknowledged.  Even though the route may be
                  asymmetric, an ACK for new data proves that the
                  acknowleged data must have been transmitted
                  successfully.

             o    An ICMP Redirect message from a particular gateway
                  should be used as positive advice about that
                  gateway.

             o    Link-layer information that reliably detects and
                  reports host failures (e.g., ARPANET Destination
                  Dead messages) should be used as negative advice.

             o    Failure to ARP or to re-validate ARP mappings may
                  be used as negative advice for the corresponding
                  IP address.

             o    Packets arriving from a particular link-layer
                  address are evidence that the system at this
                  address is alive.  However, turning this
                  information into advice about gateways requires
                  mapping the link-layer address into an IP address,
                  and then checking that IP address against the
                  gateways pointed to by the route cache.  This is
                  probably prohibitively inefficient.

             Note that positive advice that is given for every
             datagram received may cause unacceptable overhead in
             the implementation.

             While advice might be passed using required arguments
             in all interfaces to the IP layer, some transport and
             application layer protocols cannot deduce the correct
             advice.  These interfaces must therefore allow a
             neutral value for advice, since either always-positive
             or always-negative advice leads to incorrect behavior.

             There is another technique for dead gateway detection
             that has been commonly used but is not recommended.

RFC1122 INTERNET LAYER October 1989 

             This technique depends upon the host passively
             receiving ("wiretapping") the Interior Gateway Protocol
             (IGP) datagrams that the gateways are broadcasting to
             each other.  This approach has the drawback that a host
             needs to recognize all the interior gateway protocols
             that gateways may use (see [INTRO:2]).  In addition, it
             only works on a broadcast network.

             At present, pinging (i.e., using ICMP Echo messages) is
             the mechanism for gateway probing when absolutely
             required.  A successful ping guarantees that the
             addressed interface and its associated machine are up,
             but it does not guarantee that the machine is a gateway
             as opposed to a host.  The normal inference is that if
             a Redirect or other evidence indicates that a machine
             was a gateway, successful pings will indicate that the
             machine is still up and hence still a gateway.
             However, since a host silently discards packets that a
             gateway would forward or redirect, this assumption
             could sometimes fail.  To avoid this problem, a new
             ICMP message under development will ask "are you a
             gateway?"

        IMPLEMENTATION:
             The following specific algorithm has been suggested:

             o    Associate a "reroute timer" with each gateway
                  pointed to by the route cache.  Initialize the
                  timer to a value Tr, which must be small enough to
                  allow detection of a dead gateway before transport
                  connections time out.

             o    Positive advice would reset the reroute timer to
                  Tr.  Negative advice would reduce or zero the
                  reroute timer.

             o    Whenever the IP layer used a particular gateway to
                  route a datagram, it would check the corresponding
                  reroute timer.  If the timer had expired (reached
                  zero), the IP layer would send a ping to the
                  gateway, followed immediately by the datagram.

             o    The ping (ICMP Echo) would be sent again if
                  necessary, up to N times.  If no ping reply was
                  received in N tries, the gateway would be assumed
                  to have failed, and a new first-hop gateway would
                  be chosen for all cache entries pointing to the
                  failed gateway.

RFC1122 INTERNET LAYER October 1989 

             Note that the size of Tr is inversely related to the
             amount of advice available.  Tr should be large enough
             to insure that:

             *    Any pinging will be at a low level (e.g., <10%) of
                  all packets sent to a gateway from the host, AND

             *    pinging is infrequent (e.g., every 3 minutes)

             Since the recommended algorithm is concerned with the
             gateways pointed to by route cache entries, rather than
             the cache entries themselves, a two level data
             structure (perhaps coordinated with ARP or similar
             caches) may be desirable for implementing a route
             cache.

     3.3.1.5  New Gateway Selection

        If the failed gateway is not the current default, the IP
        layer can immediately switch to a default gateway.  If it is
        the current default that failed, the IP layer MUST select a
        different default gateway (assuming more than one default is
        known) for the failed route and for establishing new routes.

        DISCUSSION:
             When a gateway does fail, the other gateways on the
             connected network will learn of the failure through
             some inter-gateway routing protocol.  However, this
             will not happen instantaneously, since gateway routing
             protocols typically have a settling time of 30-60
             seconds.  If the host switches to an alternative
             gateway before the gateways have agreed on the failure,
             the new target gateway will probably forward the
             datagram to the failed gateway and send a Redirect back
             to the host pointing to the failed gateway (!).  The
             result is likely to be a rapid oscillation in the
             contents of the host's route cache during the gateway
             settling period.  It has been proposed that the dead-
             gateway logic should include some hysteresis mechanism
             to prevent such oscillations.  However, experience has
             not shown any harm from such oscillations, since
             service cannot be restored to the host until the
             gateways' routing information does settle down.

        IMPLEMENTATION:
             One implementation technique for choosing a new default
             gateway is to simply round-robin among the default
             gateways in the host's list.  Another is to rank the

RFC1122 INTERNET LAYER October 1989 

             gateways in priority order, and when the current
             default gateway is not the highest priority one, to
             "ping" the higher-priority gateways slowly to detect
             when they return to service.  This pinging can be at a
             very low rate, e.g., 0.005 per second.

     3.3.1.6  Initialization

        The following information MUST be configurable:

        (1)  IP address(es).

        (2)  Address mask(s).

        (3)  A list of default gateways, with a preference level.

        A manual method of entering this configuration data MUST be
        provided.  In addition, a variety of methods can be used to
        determine this information dynamically; see the section on
        "Host Initialization" in [INTRO:1].

        DISCUSSION:
             Some host implementations use "wiretapping" of gateway
             protocols on a broadcast network to learn what gateways
             exist.  A standard method for default gateway discovery
             is under development.

  3.3.2  Reassembly

     The IP layer MUST implement reassembly of IP datagrams.

     We designate the largest datagram size that can be reassembled
     by EMTU_R ("Effective MTU to receive"); this is sometimes
     called the "reassembly buffer size".  EMTU_R MUST be greater
     than or equal to 576, SHOULD be either configurable or
     indefinite, and SHOULD be greater than or equal to the MTU of
     the connected network(s).

     DISCUSSION:
          A fixed EMTU_R limit should not be built into the code
          because some application layer protocols require EMTU_R
          values larger than 576.

     IMPLEMENTATION:
          An implementation may use a contiguous reassembly buffer
          for each datagram, or it may use a more complex data
          structure that places no definite limit on the reassembled
          datagram size; in the latter case, EMTU_R is said to be

RFC1122 INTERNET LAYER October 1989 

          "indefinite".

          Logically, reassembly is performed by simply copying each
          fragment into the packet buffer at the proper offset.
          Note that fragments may overlap if successive
          retransmissions use different packetizing but the same
          reassembly Id.

          The tricky part of reassembly is the bookkeeping to
          determine when all bytes of the datagram have been
          reassembled.  We recommend Clark's algorithm [IP:10] that
          requires no additional data space for the bookkeeping.
          However, note that, contrary to [IP:10], the first
          fragment header needs to be saved for inclusion in a
          possible ICMP Time Exceeded (Reassembly Timeout) message.

     There MUST be a mechanism by which the transport layer can
     learn MMS_R, the maximum message size that can be received and
     reassembled in an IP datagram (see GET_MAXSIZES calls in
     Section 3.4).  If EMTU_R is not indefinite, then the value of
     MMS_R is given by:

        MMS_R = EMTU_R - 20

     since 20 is the minimum size of an IP header.

     There MUST be a reassembly timeout.  The reassembly timeout
     value SHOULD be a fixed value, not set from the remaining TTL.
     It is recommended that the value lie between 60 seconds and 120
     seconds.  If this timeout expires, the partially-reassembled
     datagram MUST be discarded and an ICMP Time Exceeded message
     sent to the source host (if fragment zero has been received).

     DISCUSSION:
          The IP specification says that the reassembly timeout
          should be the remaining TTL from the IP header, but this
          does not work well because gateways generally treat TTL as
          a simple hop count rather than an elapsed time.  If the
          reassembly timeout is too small, datagrams will be
          discarded unnecessarily, and communication may fail.  The
          timeout needs to be at least as large as the typical
          maximum delay across the Internet.  A realistic minimum
          reassembly timeout would be 60 seconds.

          It has been suggested that a cache might be kept of
          round-trip times measured by transport protocols for
          various destinations, and that these values might be used
          to dynamically determine a reasonable reassembly timeout

RFC1122 INTERNET LAYER October 1989 

          value.  Further investigation of this approach is
          required.

          If the reassembly timeout is set too high, buffer
          resources in the receiving host will be tied up too long,
          and the MSL (Maximum Segment Lifetime) [TCP:1] will be
          larger than necessary.  The MSL controls the maximum rate
          at which fragmented datagrams can be sent using distinct
          values of the 16-bit Ident field; a larger MSL lowers the
          maximum rate.  The TCP specification [TCP:1] arbitrarily
          assumes a value of 2 minutes for MSL.  This sets an upper
          limit on a reasonable reassembly timeout value.

  3.3.3  Fragmentation

     Optionally, the IP layer MAY implement a mechanism to fragment
     outgoing datagrams intentionally.

     We designate by EMTU_S ("Effective MTU for sending") the
     maximum IP datagram size that may be sent, for a particular
     combination of IP source and destination addresses and perhaps
     TOS.

     A host MUST implement a mechanism to allow the transport layer
     to learn MMS_S, the maximum transport-layer message size that
     may be sent for a given {source, destination, TOS} triplet (see
     GET_MAXSIZES call in Section 3.4).  If no local fragmentation
     is performed, the value of MMS_S will be:

        MMS_S = EMTU_S - <IP header size>

     and EMTU_S must be less than or equal to the MTU of the network
     interface corresponding to the source address of the datagram.
     Note that <IP header size> in this equation will be 20, unless
     the IP reserves space to insert IP options for its own purposes
     in addition to any options inserted by the transport layer.

     A host that does not implement local fragmentation MUST ensure
     that the transport layer (for TCP) or the application layer
     (for UDP) obtains MMS_S from the IP layer and does not send a
     datagram exceeding MMS_S in size.

     It is generally desirable to avoid local fragmentation and to
     choose EMTU_S low enough to avoid fragmentation in any gateway
     along the path.  In the absence of actual knowledge of the
     minimum MTU along the path, the IP layer SHOULD use
     EMTU_S <= 576 whenever the destination address is not on a
     connected network, and otherwise use the connected network's

RFC1122 INTERNET LAYER October 1989 

     MTU.

     The MTU of each physical interface MUST be configurable.

     A host IP layer implementation MAY have a configuration flag
     "All-Subnets-MTU", indicating that the MTU of the connected
     network is to be used for destinations on different subnets
     within the same network, but not for other networks.  Thus,
     this flag causes the network class mask, rather than the subnet
     address mask, to be used to choose an EMTU_S.  For a multihomed
     host, an "All-Subnets-MTU" flag is needed for each network
     interface.

     DISCUSSION:
          Picking the correct datagram size to use when sending data
          is a complex topic [IP:9].

          (a)  In general, no host is required to accept an IP
               datagram larger than 576 bytes (including header and
               data), so a host must not send a larger datagram
               without explicit knowledge or prior arrangement with
               the destination host.  Thus, MMS_S is only an upper
               bound on the datagram size that a transport protocol
               may send; even when MMS_S exceeds 556, the transport
               layer must limit its messages to 556 bytes in the
               absence of other knowledge about the destination
               host.

          (b)  Some transport protocols (e.g., TCP) provide a way to
               explicitly inform the sender about the largest
               datagram the other end can receive and reassemble
               [IP:7].  There is no corresponding mechanism in the
               IP layer.

               A transport protocol that assumes an EMTU_R larger
               than 576 (see Section 3.3.2), can send a datagram of
               this larger size to another host that implements the
               same protocol.

          (c)  Hosts should ideally limit their EMTU_S for a given
               destination to the minimum MTU of all the networks
               along the path, to avoid any fragmentation.  IP
               fragmentation, while formally correct, can create a
               serious transport protocol performance problem,
               because loss of a single fragment means all the
               fragments in the segment must be retransmitted
               [IP:9].

RFC1122 INTERNET LAYER October 1989 

          Since nearly all networks in the Internet currently
          support an MTU of 576 or greater, we strongly recommend
          the use of 576 for datagrams sent to non-local networks.

          It has been suggested that a host could determine the MTU
          over a given path by sending a zero-offset datagram
          fragment and waiting for the receiver to time out the
          reassembly (which cannot complete!) and return an ICMP
          Time Exceeded message.  This message would include the
          largest remaining fragment header in its body.  More
          direct mechanisms are being experimented with, but have
          not yet been adopted (see e.g., RFC-1063).

  3.3.4  Local Multihoming

     3.3.4.1  Introduction

        A multihomed host has multiple IP addresses, which we may
        think of as "logical interfaces".  These logical interfaces
        may be associated with one or more physical interfaces, and
        these physical interfaces may be connected to the same or
        different networks.

        Here are some important cases of multihoming:

        (a)  Multiple Logical Networks

             The Internet architects envisioned that each physical
             network would have a single unique IP network (or
             subnet) number.  However, LAN administrators have
             sometimes found it useful to violate this assumption,
             operating a LAN with multiple logical networks per
             physical connected network.

             If a host connected to such a physical network is
             configured to handle traffic for each of N different
             logical networks, then the host will have N logical
             interfaces.  These could share a single physical
             interface, or might use N physical interfaces to the
             same network.

        (b)  Multiple Logical Hosts

             When a host has multiple IP addresses that all have the
             same <Network-number> part (and the same <Subnet-
             number> part, if any), the logical interfaces are known
             as "logical hosts".  These logical interfaces might
             share a single physical interface or might use separate

RFC1122 INTERNET LAYER October 1989 

             physical interfaces to the same physical network.

        (c)  Simple Multihoming

             In this case, each logical interface is mapped into a
             separate physical interface and each physical interface
             is connected to a different physical network.  The term
             "multihoming" was originally applied only to this case,
             but it is now applied more generally.

             A host with embedded gateway functionality will
             typically fall into the simple multihoming case.  Note,
             however, that a host may be simply multihomed without
             containing an embedded gateway, i.e., without
             forwarding datagrams from one connected network to
             another.

             This case presents the most difficult routing problems.
             The choice of interface (i.e., the choice of first-hop
             network) may significantly affect performance or even
             reachability of remote parts of the Internet.

        Finally, we note another possibility that is NOT
        multihoming:  one logical interface may be bound to multiple
        physical interfaces, in order to increase the reliability or
        throughput between directly connected machines by providing
        alternative physical paths between them.  For instance, two
        systems might be connected by multiple point-to-point links.
        We call this "link-layer multiplexing".  With link-layer
        multiplexing, the protocols above the link layer are unaware
        that multiple physical interfaces are present; the link-
        layer device driver is responsible for multiplexing and
        routing packets across the physical interfaces.

        In the Internet protocol architecture, a transport protocol
        instance ("entity") has no address of its own, but instead
        uses a single Internet Protocol (IP) address.  This has
        implications for the IP, transport, and application layers,
        and for the interfaces between them.  In particular, the
        application software may have to be aware of the multiple IP
        addresses of a multihomed host; in other cases, the choice
        can be made within the network software.

     3.3.4.2  Multihoming Requirements

        The following general rules apply to the selection of an IP
        source address for sending a datagram from a multihomed

RFC1122 INTERNET LAYER October 1989 

        host.

        (1)  If the datagram is sent in response to a received
             datagram, the source address for the response SHOULD be
             the specific-destination address of the request.  See
             Sections 4.1.3.5 and 4.2.3.7 and the "General Issues"
             section of [INTRO:1] for more specific requirements on
             higher layers.

             Otherwise, a source address must be selected.

        (2)  An application MUST be able to explicitly specify the
             source address for initiating a connection or a
             request.

        (3)  In the absence of such a specification, the networking
             software MUST choose a source address.  Rules for this
             choice are described below.

        There are two key requirement issues related to multihoming:

        (A)  A host MAY silently discard an incoming datagram whose
             destination address does not correspond to the physical
             interface through which it is received.

        (B)  A host MAY restrict itself to sending (non-source-
             routed) IP datagrams only through the physical
             interface that corresponds to the IP source address of
             the datagrams.

        DISCUSSION:
             Internet host implementors have used two different
             conceptual models for multihoming, briefly summarized
             in the following discussion.  This document takes no
             stand on which model is preferred; each seems to have a
             place.  This ambivalence is reflected in the issues (A)
             and (B) being optional.

             o    Strong ES Model

                  The Strong ES (End System, i.e., host) model
                  emphasizes the host/gateway (ES/IS) distinction,
                  and would therefore substitute MUST for MAY in
                  issues (A) and (B) above.  It tends to model a
                  multihomed host as a set of logical hosts within
                  the same physical host.

RFC1122 INTERNET LAYER October 1989 

                  With respect to (A), proponents of the Strong ES
                  model note that automatic Internet routing
                  mechanisms could not route a datagram to a
                  physical interface that did not correspond to the
                  destination address.

                  Under the Strong ES model, the route computation
                  for an outgoing datagram is the mapping:

                     route(src IP addr, dest IP addr, TOS)
                                                    -> gateway

                  Here the source address is included as a parameter
                  in order to select a gateway that is directly
                  reachable on the corresponding physical interface.
                  Note that this model logically requires that in
                  general there be at least one default gateway, and
                  preferably multiple defaults, for each IP source
                  address.

             o    Weak ES Model

                  This view de-emphasizes the ES/IS distinction, and
                  would therefore substitute MUST NOT for MAY in
                  issues (A) and (B).  This model may be the more
                  natural one for hosts that wiretap gateway routing
                  protocols, and is necessary for hosts that have
                  embedded gateway functionality.

                  The Weak ES Model may cause the Redirect mechanism
                  to fail.  If a datagram is sent out a physical
                  interface that does not correspond to the
                  destination address, the first-hop gateway will
                  not realize when it needs to send a Redirect.  On
                  the other hand, if the host has embedded gateway
                  functionality, then it has routing information
                  without listening to Redirects.

                  In the Weak ES model, the route computation for an
                  outgoing datagram is the mapping:

                     route(dest IP addr, TOS) -> gateway, interface

RFC1122 INTERNET LAYER October 1989 

     3.3.4.3  Choosing a Source Address

        DISCUSSION:
             When it sends an initial connection request (e.g., a
             TCP "SYN" segment) or a datagram service request (e.g.,
             a UDP-based query), the transport layer on a multihomed
             host needs to know which source address to use.  If the
             application does not specify it, the transport layer
             must ask the IP layer to perform the conceptual
             mapping:

                 GET_SRCADDR(remote IP addr, TOS)
                                           -> local IP address

             Here TOS is the Type-of-Service value (see Section
             3.2.1.6), and the result is the desired source address.
             The following rules are suggested for implementing this
             mapping:

             (a)  If the remote Internet address lies on one of the
                  (sub-) nets to which the host is directly
                  connected, a corresponding source address may be
                  chosen, unless the corresponding interface is
                  known to be down.

             (b)  The route cache may be consulted, to see if there
                  is an active route to the specified destination
                  network through any network interface; if so, a
                  local IP address corresponding to that interface
                  may be chosen.

             (c)  The table of static routes, if any (see Section
                  3.3.1.2) may be similarly consulted.

             (d)  The default gateways may be consulted.  If these
                  gateways are assigned to different interfaces, the
                  interface corresponding to the gateway with the
                  highest preference may be chosen.

             In the future, there may be a defined way for a
             multihomed host to ask the gateways on all connected
             networks for advice about the best network to use for a
             given destination.

        IMPLEMENTATION:
             It will be noted that this process is essentially the
             same as datagram routing (see Section 3.3.1), and
             therefore hosts may be able to combine the

RFC1122 INTERNET LAYER October 1989 

             implementation of the two functions.

  3.3.5  Source Route Forwarding

     Subject to restrictions given below, a host MAY be able to act
     as an intermediate hop in a source route, forwarding a source-
     routed datagram to the next specified hop.

     However, in performing this gateway-like function, the host
     MUST obey all the relevant rules for a gateway forwarding
     source-routed datagrams [INTRO:2].  This includes the following
     specific provisions, which override the corresponding host
     provisions given earlier in this document:

     (A)  TTL (ref. Section 3.2.1.7)

          The TTL field MUST be decremented and the datagram perhaps
          discarded as specified for a gateway in [INTRO:2].

     (B)  ICMP Destination Unreachable (ref. Section 3.2.2.1)

          A host MUST be able to generate Destination Unreachable
          messages with the following codes:

          4    (Fragmentation Required but DF Set) when a source-
               routed datagram cannot be fragmented to fit into the
               target network;

          5    (Source Route Failed) when a source-routed datagram
               cannot be forwarded, e.g., because of a routing
               problem or because the next hop of a strict source
               route is not on a connected network.

     (C)  IP Source Address (ref. Section 3.2.1.3)

          A source-routed datagram being forwarded MAY (and normally
          will) have a source address that is not one of the IP
          addresses of the forwarding host.

     (D)  Record Route Option (ref. Section 3.2.1.8d)

          A host that is forwarding a source-routed datagram
          containing a Record Route option MUST update that option,
          if it has room.

     (E)  Timestamp Option (ref. Section 3.2.1.8e)

          A host that is forwarding a source-routed datagram

RFC1122 INTERNET LAYER October 1989 

          containing a Timestamp Option MUST add the current
          timestamp to that option, according to the rules for this
          option.

     To define the rules restricting host forwarding of source-
     routed datagrams, we use the term "local source-routing" if the
     next hop will be through the same physical interface through
     which the datagram arrived; otherwise, it is "non-local
     source-routing".

     o    A host is permitted to perform local source-routing
          without restriction.

     o    A host that supports non-local source-routing MUST have a
          configurable switch to disable forwarding, and this switch
          MUST default to disabled.

     o    The host MUST satisfy all gateway requirements for
          configurable policy filters [INTRO:2] restricting non-
          local forwarding.

     If a host receives a datagram with an incomplete source route
     but does not forward it for some reason, the host SHOULD return
     an ICMP Destination Unreachable (code 5, Source Route Failed)
     message, unless the datagram was itself an ICMP error message.

  3.3.6  Broadcasts

     Section 3.2.1.3 defined the four standard IP broadcast address
     forms:

       Limited Broadcast:  {-1, -1}

       Directed Broadcast:  {<Network-number>,-1}

       Subnet Directed Broadcast:
                          {<Network-number>,<Subnet-number>,-1}

       All-Subnets Directed Broadcast: {<Network-number>,-1,-1}

     A host MUST recognize any of these forms in the destination
     address of an incoming datagram.

     There is a class of hosts* that use non-standard broadcast
     address forms, substituting 0 for -1.  All hosts SHOULD

_________________________

  • 4.2BSD Unix and its derivatives, but not 4.3BSD.

RFC1122 INTERNET LAYER October 1989 

     recognize and accept any of these non-standard broadcast
     addresses as the destination address of an incoming datagram.
     A host MAY optionally have a configuration option to choose the
     0 or the -1 form of broadcast address, for each physical
     interface, but this option SHOULD default to the standard (-1)
     form.

     When a host sends a datagram to a link-layer broadcast address,
     the IP destination address MUST be a legal IP broadcast or IP
     multicast address.

     A host SHOULD silently discard a datagram that is received via
     a link-layer broadcast (see Section 2.4) but does not specify
     an IP multicast or broadcast destination address.

     Hosts SHOULD use the Limited Broadcast address to broadcast to
     a connected network.

     DISCUSSION:
          Using the Limited Broadcast address instead of a Directed
          Broadcast address may improve system robustness.  Problems
          are often caused by machines that do not understand the
          plethora of broadcast addresses (see Section 3.2.1.3), or
          that may have different ideas about which broadcast
          addresses are in use.  The prime example of the latter is
          machines that do not understand subnetting but are
          attached to a subnetted net.  Sending a Subnet Broadcast
          for the connected network will confuse those machines,
          which will see it as a message to some other host.

          There has been discussion on whether a datagram addressed
          to the Limited Broadcast address ought to be sent from all
          the interfaces of a multihomed host.  This specification
          takes no stand on the issue.

  3.3.7  IP Multicasting

     A host SHOULD support local IP multicasting on all connected
     networks for which a mapping from Class D IP addresses to
     link-layer addresses has been specified (see below).  Support
     for local IP multicasting includes sending multicast datagrams,
     joining multicast groups and receiving multicast datagrams, and
     leaving multicast groups.  This implies support for all of
     [IP:4] except the IGMP protocol itself, which is OPTIONAL.

RFC1122 INTERNET LAYER October 1989 

     DISCUSSION:
          IGMP provides gateways that are capable of multicast
          routing with the information required to support IP
          multicasting across multiple networks.  At this time,
          multicast-routing gateways are in the experimental stage
          and are not widely available.  For hosts that are not
          connected to networks with multicast-routing gateways or
          that do not need to receive multicast datagrams
          originating on other networks, IGMP serves no purpose and
          is therefore optional for now.  However, the rest of
          [IP:4] is currently recommended for the purpose of
          providing IP-layer access to local network multicast
          addressing, as a preferable alternative to local broadcast
          addressing.  It is expected that IGMP will become
          recommended at some future date, when multicast-routing
          gateways have become more widely available.

     If IGMP is not implemented, a host SHOULD still join the "all-
     hosts" group (224.0.0.1) when the IP layer is initialized and
     remain a member for as long as the IP layer is active.

     DISCUSSION:
          Joining the "all-hosts" group will support strictly local
          uses of multicasting, e.g., a gateway discovery protocol,
          even if IGMP is not implemented.

     The mapping of IP Class D addresses to local addresses is
     currently specified for the following types of networks:

     o    Ethernet/IEEE 802.3, as defined in [IP:4].

     o    Any network that supports broadcast but not multicast,
          addressing: all IP Class D addresses map to the local
          broadcast address.

     o    Any type of point-to-point link (e.g., SLIP or HDLC
          links): no mapping required.  All IP multicast datagrams
          are sent as-is, inside the local framing.

     Mappings for other types of networks will be specified in the
     future.

     A host SHOULD provide a way for higher-layer protocols or
     applications to determine which of the host's connected
     network(s) support IP multicast addressing.

RFC1122 INTERNET LAYER October 1989 

  3.3.8  Error Reporting

     Wherever practical, hosts MUST return ICMP error datagrams on
     detection of an error, except in those cases where returning an
     ICMP error message is specifically prohibited.

     DISCUSSION:
          A common phenomenon in datagram networks is the "black
          hole disease": datagrams are sent out, but nothing comes
          back.  Without any error datagrams, it is difficult for
          the user to figure out what the problem is.

3.4 INTERNET/TRANSPORT LAYER INTERFACE 

  The interface between the IP layer and the transport layer MUST
  provide full access to all the mechanisms of the IP layer,
  including options, Type-of-Service, and Time-to-Live.  The
  transport layer MUST either have mechanisms to set these interface
  parameters, or provide a path to pass them through from an
  application, or both.

  DISCUSSION:
       Applications are urged to make use of these mechanisms where
       applicable, even when the mechanisms are not currently
       effective in the Internet (e.g., TOS).  This will allow these
       mechanisms to be immediately useful when they do become
       effective, without a large amount of retrofitting of host
       software.

  We now describe a conceptual interface between the transport layer
  and the IP layer, as a set of procedure calls.  This is an
  extension of the information in Section 3.3 of RFC-791 [IP:1].

  *    Send Datagram

            SEND(src, dst, prot, TOS, TTL, BufPTR, len, Id, DF, opt
                 => result )

       where the parameters are defined in RFC-791.  Passing an Id
       parameter is optional; see Section 3.2.1.5.

  *    Receive Datagram

            RECV(BufPTR, prot
                 => result, src, dst, SpecDest, TOS, len, opt)

RFC1122 INTERNET LAYER October 1989 

       All the parameters are defined in RFC-791, except for:

            SpecDest = specific-destination address of datagram
                        (defined in Section 3.2.1.3)

       The result parameter dst contains the datagram's destination
       address.  Since this may be a broadcast or multicast address,
       the SpecDest parameter (not shown in RFC-791) MUST be passed.
       The parameter opt contains all the IP options received in the
       datagram; these MUST also be passed to the transport layer.

  *    Select Source Address

            GET_SRCADDR(remote, TOS)  -> local

            remote = remote IP address
            TOS = Type-of-Service
            local = local IP address

       See Section 3.3.4.3.

  *    Find Maximum Datagram Sizes

            GET_MAXSIZES(local, remote, TOS) -> MMS_R, MMS_S

            MMS_R = maximum receive transport-message size.
            MMS_S = maximum send transport-message size.
           (local, remote, TOS defined above)

       See Sections 3.3.2 and 3.3.3.

  *    Advice on Delivery Success

            ADVISE_DELIVPROB(sense, local, remote, TOS)

       Here the parameter sense is a 1-bit flag indicating whether
       positive or negative advice is being given; see the
       discussion in Section 3.3.1.4. The other parameters were
       defined earlier.

  *    Send ICMP Message

            SEND_ICMP(src, dst, TOS, TTL, BufPTR, len, Id, DF, opt)
                 -> result

RFC1122 INTERNET LAYER October 1989 

            (Parameters defined in RFC-791).

       Passing an Id parameter is optional; see Section 3.2.1.5.
       The transport layer MUST be able to send certain ICMP
       messages:  Port Unreachable or any of the query-type
       messages.  This function could be considered to be a special
       case of the SEND() call, of course; we describe it separately
       for clarity.

  *    Receive ICMP Message

            RECV_ICMP(BufPTR ) -> result, src, dst, len, opt

            (Parameters defined in RFC-791).

       The IP layer MUST pass certain ICMP messages up to the
       appropriate transport-layer routine.  This function could be
       considered to be a special case of the RECV() call, of
       course; we describe it separately for clarity.

       For an ICMP error message, the data that is passed up MUST
       include the original Internet header plus all the octets of
       the original message that are included in the ICMP message.
       This data will be used by the transport layer to locate the
       connection state information, if any.

       In particular, the following ICMP messages are to be passed
       up:

       o    Destination Unreachable

       o    Source Quench

       o    Echo Reply (to ICMP user interface, unless the Echo
            Request originated in the IP layer)

       o    Timestamp Reply (to ICMP user interface)

       o    Time Exceeded

  DISCUSSION:
       In the future, there may be additions to this interface to
       pass path data (see Section 3.3.1.3) between the IP and
       transport layers.

RFC1122 INTERNET LAYER October 1989

3.5 INTERNET LAYER REQUIREMENTS SUMMARY 

                                             |        | | | |S| |
                                             |        | | | |H| |F
                                             |        | | | |O|M|o
                                             |        | |S| |U|U|o
                                             |        | |H| |L|S|t
                                             |        |M|O| |D|T|n
                                             |        |U|U|M| | |o
                                             |        |S|L|A|N|N|t
                                             |        |T|D|Y|O|O|t

FEATURE |SECTION | | | |T|T|e

|--------|-|-|-|-|-|-- 

                                             |        | | | | | |

Implement IP and ICMP |3.1 |x| | | | | Handle remote multihoming in application layer |3.1 |x| | | | | Support local multihoming |3.1 | | |x| | | Meet gateway specs if forward datagrams |3.1 |x| | | | | Configuration switch for embedded gateway |3.1 |x| | | | |1 Config switch default to non-gateway |3.1 |x| | | | |1 Auto-config based on number of interfaces |3.1 | | | | |x|1 Able to log discarded datagrams |3.1 | |x| | | | Record in counter |3.1 | |x| | | | 

                                             |        | | | | | |

Silently discard Version != 4 |3.2.1.1 |x| | | | | Verify IP checksum, silently discard bad dgram |3.2.1.2 |x| | | | | Addressing: | | | | | | | 

 Subnet addressing (RFC-950)                    |3.2.1.3 |x| | | | |
 Src address must be host's own IP address      |3.2.1.3 |x| | | | |
 Silently discard datagram with bad dest addr   |3.2.1.3 |x| | | | |
 Silently discard datagram with bad src addr    |3.2.1.3 |x| | | | |

Support reassembly |3.2.1.4 |x| | | | | Retain same Id field in identical datagram |3.2.1.5 | | |x| | | 

                                             |        | | | | | |

TOS: | | | | | | | 

 Allow transport layer to set TOS               |3.2.1.6 |x| | | | |
 Pass received TOS up to transport layer        |3.2.1.6 | |x| | | |
 Use RFC-795 link-layer mappings for TOS        |3.2.1.6 | | | |x| |

TTL: | | | | | | | 

 Send packet with TTL of 0                      |3.2.1.7 | | | | |x|
 Discard received packets with TTL < 2          |3.2.1.7 | | | | |x|
 Allow transport layer to set TTL               |3.2.1.7 |x| | | | |
 Fixed TTL is configurable                      |3.2.1.7 |x| | | | |
                                             |        | | | | | |

IP Options: | | | | | | | 

 Allow transport layer to send IP options       |3.2.1.8 |x| | | | |
 Pass all IP options rcvd to higher layer       |3.2.1.8 |x| | | | |

RFC1122 INTERNET LAYER October 1989 

 IP layer silently ignore unknown options       |3.2.1.8 |x| | | | |
 Security option                                |3.2.1.8a| | |x| | |
 Send Stream Identifier option                  |3.2.1.8b| | | |x| |
 Silently ignore Stream Identifer option        |3.2.1.8b|x| | | | |
 Record Route option                            |3.2.1.8d| | |x| | |
 Timestamp option                               |3.2.1.8e| | |x| | |

Source Route Option: | | | | | | | 

 Originate & terminate Source Route options     |3.2.1.8c|x| | | | |
 Datagram with completed SR passed up to TL     |3.2.1.8c|x| | | | |
 Build correct (non-redundant) return route     |3.2.1.8c|x| | | | |
 Send multiple SR options in one header         |3.2.1.8c| | | | |x|
                                             |        | | | | | |

ICMP: | | | | | | | 

 Silently discard ICMP msg with unknown type    |3.2.2   |x| | | | |
 Include more than 8 octets of orig datagram    |3.2.2   | | |x| | |
  Included octets same as received           |3.2.2   |x| | | | |
 Demux ICMP Error to transport protocol         |3.2.2   |x| | | | |
 Send ICMP error message with TOS=0             |3.2.2   | |x| | | |
 Send ICMP error message for:                   |        | | | | | |

- ICMP error msg |3.2.2 | | | | |x| - IP b'cast or IP m'cast |3.2.2 | | | | |x| - Link-layer b'cast |3.2.2 | | | | |x| - Non-initial fragment |3.2.2 | | | | |x| - Datagram with non-unique src address |3.2.2 | | | | |x| 

 Return ICMP error msgs (when not prohibited)   |3.3.8   |x| | | | |
                                             |        | | | | | |
 Dest Unreachable:                              |        | | | | | |
Generate Dest Unreachable (code 2/3)         |3.2.2.1 | |x| | | |
Pass ICMP Dest Unreachable to higher layer   |3.2.2.1 |x| | | | |
Higher layer act on Dest Unreach             |3.2.2.1 | |x| | | |
  Interpret Dest Unreach as only hint        |3.2.2.1 |x| | | | |
 Redirect:                                      |        | | | | | |
Host send Redirect                           |3.2.2.2 | | | |x| |
Update route cache when recv Redirect        |3.2.2.2 |x| | | | |
Handle both Host and Net Redirects           |3.2.2.2 |x| | | | |
Discard illegal Redirect                     |3.2.2.2 | |x| | | |
 Source Quench:                                 |        | | | | | |
Send Source Quench if buffering exceeded     |3.2.2.3 | | |x| | |
Pass Source Quench to higher layer           |3.2.2.3 |x| | | | |
Higher layer act on Source Quench            |3.2.2.3 | |x| | | |
 Time Exceeded: pass to higher layer            |3.2.2.4 |x| | | | |
 Parameter Problem:                             |        | | | | | |
Send Parameter Problem messages              |3.2.2.5 | |x| | | |
Pass Parameter Problem to higher layer       |3.2.2.5 |x| | | | |
Report Parameter Problem to user             |3.2.2.5 | | |x| | |
                                             |        | | | | | |
 ICMP Echo Request or Reply:                    |        | | | | | |
Echo server and Echo client                  |3.2.2.6 |x| | | | |

RFC1122 INTERNET LAYER October 1989 

Echo client                                  |3.2.2.6 | |x| | | |
Discard Echo Request to broadcast address    |3.2.2.6 | | |x| | |
Discard Echo Request to multicast address    |3.2.2.6 | | |x| | |
Use specific-dest addr as Echo Reply src     |3.2.2.6 |x| | | | |
Send same data in Echo Reply                 |3.2.2.6 |x| | | | |
Pass Echo Reply to higher layer              |3.2.2.6 |x| | | | |
Reflect Record Route, Time Stamp options     |3.2.2.6 | |x| | | |
Reverse and reflect Source Route option      |3.2.2.6 |x| | | | |
                                             |        | | | | | |
 ICMP Information Request or Reply:             |3.2.2.7 | | | |x| |
 ICMP Timestamp and Timestamp Reply:            |3.2.2.8 | | |x| | |
Minimize delay variability                   |3.2.2.8 | |x| | | |1
Silently discard b'cast Timestamp            |3.2.2.8 | | |x| | |1
Silently discard m'cast Timestamp            |3.2.2.8 | | |x| | |1
Use specific-dest addr as TS Reply src       |3.2.2.8 |x| | | | |1
Reflect Record Route, Time Stamp options     |3.2.2.6 | |x| | | |1
Reverse and reflect Source Route option      |3.2.2.8 |x| | | | |1
Pass Timestamp Reply to higher layer         |3.2.2.8 |x| | | | |1
Obey rules for "standard value"              |3.2.2.8 |x| | | | |1
                                             |        | | | | | |
 ICMP Address Mask Request and Reply:           |        | | | | | |
Addr Mask source configurable                |3.2.2.9 |x| | | | |
Support static configuration of addr mask    |3.2.2.9 |x| | | | |
Get addr mask dynamically during booting     |3.2.2.9 | | |x| | |
Get addr via ICMP Addr Mask Request/Reply    |3.2.2.9 | | |x| | |
  Retransmit Addr Mask Req if no Reply       |3.2.2.9 |x| | | | |3
  Assume default mask if no Reply            |3.2.2.9 | |x| | | |3
  Update address mask from first Reply only  |3.2.2.9 |x| | | | |3
Reasonableness check on Addr Mask            |3.2.2.9 | |x| | | |
Send unauthorized Addr Mask Reply msgs       |3.2.2.9 | | | | |x|
  Explicitly configured to be agent          |3.2.2.9 |x| | | | |
Static config=> Addr-Mask-Authoritative flag |3.2.2.9 | |x| | | |
  Broadcast Addr Mask Reply when init.       |3.2.2.9 |x| | | | |3
                                             |        | | | | | |

ROUTING OUTBOUND DATAGRAMS: | | | | | | | 

 Use address mask in local/remote decision      |3.3.1.1 |x| | | | |
 Operate with no gateways on conn network       |3.3.1.1 |x| | | | |
 Maintain "route cache" of next-hop gateways    |3.3.1.2 |x| | | | |
 Treat Host and Net Redirect the same           |3.3.1.2 | |x| | | |
 If no cache entry, use default gateway         |3.3.1.2 |x| | | | |
Support multiple default gateways            |3.3.1.2 |x| | | | |
 Provide table of static routes                 |3.3.1.2 | | |x| | |
Flag: route overridable by Redirects         |3.3.1.2 | | |x| | |
 Key route cache on host, not net address       |3.3.1.3 | | |x| | |
 Include TOS in route cache                     |3.3.1.3 | |x| | | |
                                             |        | | | | | |
 Able to detect failure of next-hop gateway     |3.3.1.4 |x| | | | |
 Assume route is good forever                   |3.3.1.4 | | | |x| |

RFC1122 INTERNET LAYER October 1989 

 Ping gateways continuously                     |3.3.1.4 | | | | |x|
 Ping only when traffic being sent              |3.3.1.4 |x| | | | |
 Ping only when no positive indication          |3.3.1.4 |x| | | | |
 Higher and lower layers give advice            |3.3.1.4 | |x| | | |
 Switch from failed default g'way to another    |3.3.1.5 |x| | | | |
 Manual method of entering config info          |3.3.1.6 |x| | | | |
                                             |        | | | | | |

REASSEMBLY and FRAGMENTATION: | | | | | | | 

 Able to reassemble incoming datagrams          |3.3.2   |x| | | | |
At least 576 byte datagrams                  |3.3.2   |x| | | | |
EMTU_R configurable or indefinite            |3.3.2   | |x| | | |
 Transport layer able to learn MMS_R            |3.3.2   |x| | | | |
 Send ICMP Time Exceeded on reassembly timeout  |3.3.2   |x| | | | |
Fixed reassembly timeout value               |3.3.2   | |x| | | |
                                             |        | | | | | |
 Pass MMS_S to higher layers                    |3.3.3   |x| | | | |
 Local fragmentation of outgoing packets        |3.3.3   | | |x| | |
 Else don't send bigger than MMS_S           |3.3.3   |x| | | | |
 Send max 576 to off-net destination            |3.3.3   | |x| | | |
 All-Subnets-MTU configuration flag             |3.3.3   | | |x| | |
                                             |        | | | | | |

MULTIHOMING: | | | | | | | 

 Reply with same addr as spec-dest addr         |3.3.4.2 | |x| | | |
 Allow application to choose local IP addr      |3.3.4.2 |x| | | | |
 Silently discard d'gram in "wrong" interface   |3.3.4.2 | | |x| | |
 Only send d'gram through "right" interface     |3.3.4.2 | | |x| | |4
                                             |        | | | | | |

SOURCE-ROUTE FORWARDING: | | | | | | | 

 Forward datagram with Source Route option      |3.3.5   | | |x| | |1
Obey corresponding gateway rules             |3.3.5   |x| | | | |1
  Update TTL by gateway rules                |3.3.5   |x| | | | |1
  Able to generate ICMP err code 4, 5        |3.3.5   |x| | | | |1
  IP src addr not local host                 |3.3.5   | | |x| | |1
  Update Timestamp, Record Route options     |3.3.5   |x| | | | |1
Configurable switch for non-local SRing      |3.3.5   |x| | | | |1
  Defaults to OFF                            |3.3.5   |x| | | | |1
Satisfy gwy access rules for non-local SRing |3.3.5   |x| | | | |1
If not forward, send Dest Unreach (cd 5)     |3.3.5   | |x| | | |2
                                             |        | | | | | |

BROADCAST: | | | | | | | 

 Broadcast addr as IP source addr               |3.2.1.3 | | | | |x|
 Receive 0 or -1 broadcast formats OK           |3.3.6   | |x| | | |
 Config'ble option to send 0 or -1 b'cast       |3.3.6   | | |x| | |
Default to -1 broadcast                      |3.3.6   | |x| | | |
 Recognize all broadcast address formats        |3.3.6   |x| | | | |
 Use IP b'cast/m'cast addr in link-layer b'cast |3.3.6   |x| | | | |
 Silently discard link-layer-only b'cast dg's   |3.3.6   | |x| | | |
 Use Limited Broadcast addr for connected net   |3.3.6   | |x| | | |

RFC1122 INTERNET LAYER October 1989 

                                             |        | | | | | |

MULTICAST: | | | | | | | 

 Support local IP multicasting (RFC-1112)       |3.3.7   | |x| | | |
 Support IGMP (RFC-1112)                        |3.3.7   | | |x| | |
 Join all-hosts group at startup                |3.3.7   | |x| | | |
 Higher layers learn i'face m'cast capability   |3.3.7   | |x| | | |
                                             |        | | | | | |

INTERFACE: | | | | | | | 

 Allow transport layer to use all IP mechanisms |3.4     |x| | | | |
 Pass interface ident up to transport layer     |3.4     |x| | | | |
 Pass all IP options up to transport layer      |3.4     |x| | | | |
 Transport layer can send certain ICMP messages |3.4     |x| | | | |
 Pass spec'd ICMP messages up to transp. layer  |3.4     |x| | | | |
 Include IP hdr+8 octets or more from orig.  |3.4     |x| | | | |
 Able to leap tall buildings at a single bound  |3.5     | |x| | | |

Footnotes:

(1) Only if feature is implemented.

(2) This requirement is overruled if datagram is an ICMP error message.

(3) Only if feature is implemented and is configured "on".

(4) Unless has embedded gateway functionality or is source routed.

RFC1122 TRANSPORT LAYER -- UDP October 1989

TRANSPORT PROTOCOLS

4.1 USER DATAGRAM PROTOCOL -- UDP 

  4.1.1  INTRODUCTION

     The User Datagram Protocol UDP [UDP:1] offers only a minimal
     transport service -- non-guaranteed datagram delivery -- and
     gives applications direct access to the datagram service of the
     IP layer.  UDP is used by applications that do not require the
     level of service of TCP or that wish to use communications
     services (e.g., multicast or broadcast delivery) not available
     from TCP.

     UDP is almost a null protocol; the only services it provides
     over IP are checksumming of data and multiplexing by port
     number.  Therefore, an application program running over UDP
     must deal directly with end-to-end communication problems that
     a connection-oriented protocol would have handled -- e.g.,
     retransmission for reliable delivery, packetization and
     reassembly, flow control, congestion avoidance, etc., when
     these are required.  The fairly complex coupling between IP and
     TCP will be mirrored in the coupling between UDP and many
     applications using UDP.

  4.1.2  PROTOCOL WALK-THROUGH

     There are no known errors in the specification of UDP.

  4.1.3  SPECIFIC ISSUES

     4.1.3.1  Ports

        UDP well-known ports follow the same rules as TCP well-known
        ports; see Section 4.2.2.1 below.

        If a datagram arrives addressed to a UDP port for which
        there is no pending LISTEN call, UDP SHOULD send an ICMP
        Port Unreachable message.

     4.1.3.2  IP Options

        UDP MUST pass any IP option that it receives from the IP
        layer transparently to the application layer.

        An application MUST be able to specify IP options to be sent
        in its UDP datagrams, and UDP MUST pass these options to the
        IP layer.

RFC1122 TRANSPORT LAYER -- UDP October 1989 

        DISCUSSION:
             At present, the only options that need be passed
             through UDP are Source Route, Record Route, and Time
             Stamp.  However, new options may be defined in the
             future, and UDP need not and should not make any
             assumptions about the format or content of options it
             passes to or from the application; an exception to this
             might be an IP-layer security option.

             An application based on UDP will need to obtain a
             source route from a request datagram and supply a
             reversed route for sending the corresponding reply.

     4.1.3.3  ICMP Messages

        UDP MUST pass to the application layer all ICMP error
        messages that it receives from the IP layer.  Conceptually
        at least, this may be accomplished with an upcall to the
        ERROR_REPORT routine (see Section 4.2.4.1).

        DISCUSSION:
             Note that ICMP error messages resulting from sending a
             UDP datagram are received asynchronously.  A UDP-based
             application that wants to receive ICMP error messages
             is responsible for maintaining the state necessary to
             demultiplex these messages when they arrive; for
             example, the application may keep a pending receive
             operation for this purpose.  The application is also
             responsible to avoid confusion from a delayed ICMP
             error message resulting from an earlier use of the same
             port(s).

     4.1.3.4  UDP Checksums

        A host MUST implement the facility to generate and validate
        UDP checksums.  An application MAY optionally be able to
        control whether a UDP checksum will be generated, but it
        MUST default to checksumming on.

        If a UDP datagram is received with a checksum that is non-
        zero and invalid, UDP MUST silently discard the datagram.
        An application MAY optionally be able to control whether UDP
        datagrams without checksums should be discarded or passed to
        the application.

        DISCUSSION:
             Some applications that normally run only across local
             area networks have chosen to turn off UDP checksums for

RFC1122 TRANSPORT LAYER -- UDP October 1989 

             efficiency.  As a result, numerous cases of undetected
             errors have been reported.  The advisability of ever
             turning off UDP checksumming is very controversial.

        IMPLEMENTATION:
             There is a common implementation error in UDP
             checksums.  Unlike the TCP checksum, the UDP checksum
             is optional; the value zero is transmitted in the
             checksum field of a UDP header to indicate the absence
             of a checksum.  If the transmitter really calculates a
             UDP checksum of zero, it must transmit the checksum as
             all 1's (65535).  No special action is required at the
             receiver, since zero and 65535 are equivalent in 1's
             complement arithmetic.

     4.1.3.5  UDP Multihoming

        When a UDP datagram is received, its specific-destination
        address MUST be passed up to the application layer.

        An application program MUST be able to specify the IP source
        address to be used for sending a UDP datagram or to leave it
        unspecified (in which case the networking software will
        choose an appropriate source address).  There SHOULD be a
        way to communicate the chosen source address up to the
        application layer (e.g, so that the application can later
        receive a reply datagram only from the corresponding
        interface).

        DISCUSSION:
             A request/response application that uses UDP should use
             a source address for the response that is the same as
             the specific destination address of the request.  See
             the "General Issues" section of [INTRO:1].

     4.1.3.6  Invalid Addresses

        A UDP datagram received with an invalid IP source address
        (e.g., a broadcast or multicast address) must be discarded
        by UDP or by the IP layer (see Section 3.2.1.3).

        When a host sends a UDP datagram, the source address MUST be
        (one of) the IP address(es) of the host.

  4.1.4  UDP/APPLICATION LAYER INTERFACE

     The application interface to UDP MUST provide the full services
     of the IP/transport interface described in Section 3.4 of this

RFC1122 TRANSPORT LAYER -- UDP October 1989 

     document.  Thus, an application using UDP needs the functions
     of the GET_SRCADDR(), GET_MAXSIZES(), ADVISE_DELIVPROB(), and
     RECV_ICMP() calls described in Section 3.4.  For example,
     GET_MAXSIZES() can be used to learn the effective maximum UDP
     maximum datagram size for a particular {interface,remote
     host,TOS} triplet.

     An application-layer program MUST be able to set the TTL and
     TOS values as well as IP options for sending a UDP datagram,
     and these values must be passed transparently to the IP layer.
     UDP MAY pass the received TOS up to the application layer.

  4.1.5  UDP REQUIREMENTS SUMMARY

                                             |        | | | |S| |
                                             |        | | | |H| |F
                                             |        | | | |O|M|o
                                             |        | |S| |U|U|o
                                             |        | |H| |L|S|t
                                             |        |M|O| |D|T|n
                                             |        |U|U|M| | |o
                                             |        |S|L|A|N|N|t
                                             |        |T|D|Y|O|O|t

FEATURE |SECTION | | | |T|T|e

|--------|-|-|-|-|-|-- 

                                             |        | | | | | |
UDP                                          |        | | | | | |

|--------|-|-|-|-|-|-- 

                                             |        | | | | | |

UDP send Port Unreachable |4.1.3.1 | |x| | | | 

                                             |        | | | | | |

IP Options in UDP | | | | | | | 

- Pass rcv'd IP options to applic layer         |4.1.3.2 |x| | | | |
- Applic layer can specify IP options in Send   |4.1.3.2 |x| | | | |
- UDP passes IP options down to IP layer        |4.1.3.2 |x| | | | |
                                             |        | | | | | |

Pass ICMP msgs up to applic layer |4.1.3.3 |x| | | | | 

                                             |        | | | | | |

UDP checksums: | | | | | | | 

- Able to generate/check checksum               |4.1.3.4 |x| | | | |
- Silently discard bad checksum                 |4.1.3.4 |x| | | | |
- Sender Option to not generate checksum        |4.1.3.4 | | |x| | |

- Default is to checksum |4.1.3.4 |x| | | | | 

- Receiver Option to require checksum           |4.1.3.4 | | |x| | |
                                             |        | | | | | |

UDP Multihoming | | | | | | | 

- Pass spec-dest addr to application            |4.1.3.5 |x| | | | |

RFC1122 TRANSPORT LAYER -- UDP October 1989 

- Applic layer can specify Local IP addr        |4.1.3.5 |x| | | | |
- Applic layer specify wild Local IP addr       |4.1.3.5 |x| | | | |
- Applic layer notified of Local IP addr used   |4.1.3.5 | |x| | | |
                                             |        | | | | | |

Bad IP src addr silently discarded by UDP/IP |4.1.3.6 |x| | | | | Only send valid IP source address |4.1.3.6 |x| | | | | UDP Application Interface Services | | | | | | | Full IP interface of 3.4 for application |4.1.4 |x| | | | | 

- Able to spec TTL, TOS, IP opts when send dg   |4.1.4   |x| | | | |
- Pass received TOS up to applic layer          |4.1.4   | | |x| | |

RFC1122 TRANSPORT LAYER -- TCP October 1989

4.2 TRANSMISSION CONTROL PROTOCOL -- TCP 

  4.2.1  INTRODUCTION

     The Transmission Control Protocol TCP [TCP:1] is the primary
     virtual-circuit transport protocol for the Internet suite.  TCP
     provides reliable, in-sequence delivery of a full-duplex stream
     of octets (8-bit bytes).  TCP is used by those applications
     needing reliable, connection-oriented transport service, e.g.,
     mail (SMTP), file transfer (FTP), and virtual terminal service
     (Telnet); requirements for these application-layer protocols
     are described in [INTRO:1].

  4.2.2  PROTOCOL WALK-THROUGH

     4.2.2.1  Well-Known Ports: RFC-793 Section 2.7

        DISCUSSION:
             TCP reserves port numbers in the range 0-255 for
             "well-known" ports, used to access services that are
             standardized across the Internet.  The remainder of the
             port space can be freely allocated to application
             processes.  Current well-known port definitions are
             listed in the RFC entitled "Assigned Numbers"
             [INTRO:6].  A prerequisite for defining a new well-
             known port is an RFC documenting the proposed service
             in enough detail to allow new implementations.

             Some systems extend this notion by adding a third
             subdivision of the TCP port space: reserved ports,
             which are generally used for operating-system-specific
             services.  For example, reserved ports might fall
             between 256 and some system-dependent upper limit.
             Some systems further choose to protect well-known and
             reserved ports by permitting only privileged users to
             open TCP connections with those port values.  This is
             perfectly reasonable as long as the host does not
             assume that all hosts protect their low-numbered ports
             in this manner.

     4.2.2.2  Use of Push: RFC-793 Section 2.8

        When an application issues a series of SEND calls without
        setting the PUSH flag, the TCP MAY aggregate the data
        internally without sending it.  Similarly, when a series of
        segments is received without the PSH bit, a TCP MAY queue
        the data internally without passing it to the receiving
        application.

RFC1122 TRANSPORT LAYER -- TCP October 1989 

        The PSH bit is not a record marker and is independent of
        segment boundaries.  The transmitter SHOULD collapse
        successive PSH bits when it packetizes data, to send the
        largest possible segment.

        A TCP MAY implement PUSH flags on SEND calls.  If PUSH flags
        are not implemented, then the sending TCP: (1) must not
        buffer data indefinitely, and (2) MUST set the PSH bit in
        the last buffered segment (i.e., when there is no more
        queued data to be sent).

        The discussion in RFC-793 on pages 48, 50, and 74
        erroneously implies that a received PSH flag must be passed
        to the application layer.  Passing a received PSH flag to
        the application layer is now OPTIONAL.

        An application program is logically required to set the PUSH
        flag in a SEND call whenever it needs to force delivery of
        the data to avoid a communication deadlock.  However, a TCP
        SHOULD send a maximum-sized segment whenever possible, to
        improve performance (see Section 4.2.3.4).

        DISCUSSION:
             When the PUSH flag is not implemented on SEND calls,
             i.e., when the application/TCP interface uses a pure
             streaming model, responsibility for aggregating any
             tiny data fragments to form reasonable sized segments
             is partially borne by the application layer.

             Generally, an interactive application protocol must set
             the PUSH flag at least in the last SEND call in each
             command or response sequence.  A bulk transfer protocol
             like FTP should set the PUSH flag on the last segment
             of a file or when necessary to prevent buffer deadlock.

             At the receiver, the PSH bit forces buffered data to be
             delivered to the application (even if less than a full
             buffer has been received). Conversely, the lack of a
             PSH bit can be used to avoid unnecessary wakeup calls
             to the application process; this can be an important
             performance optimization for large timesharing hosts.
             Passing the PSH bit to the receiving application allows
             an analogous optimization within the application.

     4.2.2.3  Window Size: RFC-793 Section 3.1

        The window size MUST be treated as an unsigned number, or
        else large window sizes will appear like negative windows

RFC1122 TRANSPORT LAYER -- TCP October 1989 

        and TCP will not work.  It is RECOMMENDED that
        implementations reserve 32-bit fields for the send and
        receive window sizes in the connection record and do all
        window computations with 32 bits.

        DISCUSSION:
             It is known that the window field in the TCP header is
             too small for high-speed, long-delay paths.
             Experimental TCP options have been defined to extend
             the window size; see for example [TCP:11].  In
             anticipation of the adoption of such an extension, TCP
             implementors should treat windows as 32 bits.

     4.2.2.4  Urgent Pointer: RFC-793 Section 3.1

        The second sentence is in error: the urgent pointer points
        to the sequence number of the LAST octet (not LAST+1) in a
        sequence of urgent data.  The description on page 56 (last
        sentence) is correct.

        A TCP MUST support a sequence of urgent data of any length.

        A TCP MUST inform the application layer asynchronously
        whenever it receives an Urgent pointer and there was
        previously no pending urgent data, or whenever the Urgent
        pointer advances in the data stream.  There MUST be a way
        for the application to learn how much urgent data remains to
        be read from the connection, or at least to determine
        whether or not more urgent data remains to be read.

        DISCUSSION:
             Although the Urgent mechanism may be used for any
             application, it is normally used to send "interrupt"-
             type commands to a Telnet program (see "Using Telnet
             Synch Sequence" section in [INTRO:1]).

             The asynchronous or "out-of-band" notification will
             allow the application to go into "urgent mode", reading
             data from the TCP connection.  This allows control
             commands to be sent to an application whose normal
             input buffers are full of unprocessed data.

        IMPLEMENTATION:
             The generic ERROR-REPORT() upcall described in Section
             4.2.4.1 is a possible mechanism for informing the
             application of the arrival of urgent data.

RFC1122 TRANSPORT LAYER -- TCP October 1989 

     4.2.2.5  TCP Options: RFC-793 Section 3.1

        A TCP MUST be able to receive a TCP option in any segment.
        A TCP MUST ignore without error any TCP option it does not
        implement, assuming that the option has a length field (all
        TCP options defined in the future will have length fields).
        TCP MUST be prepared to handle an illegal option length
        (e.g., zero) without crashing; a suggested procedure is to
        reset the connection and log the reason.

     4.2.2.6  Maximum Segment Size Option: RFC-793 Section 3.1

        TCP MUST implement both sending and receiving the Maximum
        Segment Size option [TCP:4].

        TCP SHOULD send an MSS (Maximum Segment Size) option in
        every SYN segment when its receive MSS differs from the
        default 536, and MAY send it always.

        If an MSS option is not received at connection setup, TCP
        MUST assume a default send MSS of 536 (576-40) [TCP:4].

        The maximum size of a segment that TCP really sends, the
        "effective send MSS," MUST be the smaller of the send MSS
        (which reflects the available reassembly buffer size at the
        remote host) and the largest size permitted by the IP layer:

           Eff.snd.MSS =

              min(SendMSS+20, MMS_S) - TCPhdrsize - IPoptionsize

        where:

        *    SendMSS is the MSS value received from the remote host,
             or the default 536 if no MSS option is received.

        *    MMS_S is the maximum size for a transport-layer message
             that TCP may send.

        *    TCPhdrsize is the size of the TCP header; this is
             normally 20, but may be larger if TCP options are to be
             sent.

        *    IPoptionsize is the size of any IP options that TCP
             will pass to the IP layer with the current message.

        The MSS value to be sent in an MSS option must be less than

RFC1122 TRANSPORT LAYER -- TCP October 1989 

        or equal to:

           MMS_R - 20

        where MMS_R is the maximum size for a transport-layer
        message that can be received (and reassembled).  TCP obtains
        MMS_R and MMS_S from the IP layer; see the generic call
        GET_MAXSIZES in Section 3.4.

        DISCUSSION:
             The choice of TCP segment size has a strong effect on
             performance.  Larger segments increase throughput by
             amortizing header size and per-datagram processing
             overhead over more data bytes; however, if the packet
             is so large that it causes IP fragmentation, efficiency
             drops sharply if any fragments are lost [IP:9].

             Some TCP implementations send an MSS option only if the
             destination host is on a non-connected network.
             However, in general the TCP layer may not have the
             appropriate information to make this decision, so it is
             preferable to leave to the IP layer the task of
             determining a suitable MTU for the Internet path.  We
             therefore recommend that TCP always send the option (if
             not 536) and that the IP layer determine MMS_R as
             specified in 3.3.3 and 3.4.  A proposed IP-layer
             mechanism to measure the MTU would then modify the IP
             layer without changing TCP.

     4.2.2.7  TCP Checksum: RFC-793 Section 3.1

        Unlike the UDP checksum (see Section 4.1.3.4), the TCP
        checksum is never optional.  The sender MUST generate it and
        the receiver MUST check it.

     4.2.2.8  TCP Connection State Diagram: RFC-793 Section 3.2,
        page 23

        There are several problems with this diagram:

        (a)  The arrow from SYN-SENT to SYN-RCVD should be labeled
             with "snd SYN,ACK", to agree with the text on page 68
             and with Figure 8.

        (b)  There could be an arrow from SYN-RCVD state to LISTEN
             state, conditioned on receiving a RST after a passive
             open (see text page 70).

RFC1122 TRANSPORT LAYER -- TCP October 1989 

        (c)  It is possible to go directly from FIN-WAIT-1 to the
             TIME-WAIT state (see page 75 of the spec).

     4.2.2.9  Initial Sequence Number Selection: RFC-793 Section
        3.3, page 27

        A TCP MUST use the specified clock-driven selection of
        initial sequence numbers.

     4.2.2.10  Simultaneous Open Attempts: RFC-793 Section 3.4, page
        32

        There is an error in Figure 8: the packet on line 7 should
        be identical to the packet on line 5.

        A TCP MUST support simultaneous open attempts.

        DISCUSSION:
             It sometimes surprises implementors that if two
             applications attempt to simultaneously connect to each
             other, only one connection is generated instead of two.
             This was an intentional design decision; don't try to
             "fix" it.

     4.2.2.11  Recovery from Old Duplicate SYN: RFC-793 Section 3.4,
        page 33

        Note that a TCP implementation MUST keep track of whether a
        connection has reached SYN_RCVD state as the result of a
        passive OPEN or an active OPEN.

     4.2.2.12  RST Segment: RFC-793 Section 3.4

        A TCP SHOULD allow a received RST segment to include data.

        DISCUSSION
             It has been suggested that a RST segment could contain
             ASCII text that encoded and explained the cause of the
             RST.  No standard has yet been established for such
             data.

     4.2.2.13  Closing a Connection: RFC-793 Section 3.5

        A TCP connection may terminate in two ways: (1) the normal
        TCP close sequence using a FIN handshake, and (2) an "abort"
        in which one or more RST segments are sent and the
        connection state is immediately discarded.  If a TCP

RFC1122 TRANSPORT LAYER -- TCP October 1989 

        connection is closed by the remote site, the local
        application MUST be informed whether it closed normally or
        was aborted.

        The normal TCP close sequence delivers buffered data
        reliably in both directions.  Since the two directions of a
        TCP connection are closed independently, it is possible for
        a connection to be "half closed," i.e., closed in only one
        direction, and a host is permitted to continue sending data
        in the open direction on a half-closed connection.

        A host MAY implement a "half-duplex" TCP close sequence, so
        that an application that has called CLOSE cannot continue to
        read data from the connection.  If such a host issues a
        CLOSE call while received data is still pending in TCP, or
        if new data is received after CLOSE is called, its TCP
        SHOULD send a RST to show that data was lost.

        When a connection is closed actively, it MUST linger in
        TIME-WAIT state for a time 2xMSL (Maximum Segment Lifetime).
        However, it MAY accept a new SYN from the remote TCP to
        reopen the connection directly from TIME-WAIT state, if it:

        (1)  assigns its initial sequence number for the new
             connection to be larger than the largest sequence
             number it used on the previous connection incarnation,
             and

        (2)  returns to TIME-WAIT state if the SYN turns out to be
             an old duplicate.

        DISCUSSION:
             TCP's full-duplex data-preserving close is a feature
             that is not included in the analogous ISO transport
             protocol TP4.

             Some systems have not implemented half-closed
             connections, presumably because they do not fit into
             the I/O model of their particular operating system.  On
             these systems, once an application has called CLOSE, it
             can no longer read input data from the connection; this
             is referred to as a "half-duplex" TCP close sequence.

             The graceful close algorithm of TCP requires that the
             connection state remain defined on (at least)  one end
             of the connection, for a timeout period of 2xMSL, i.e.,
             4 minutes.  During this period, the (remote socket,

RFC1122 TRANSPORT LAYER -- TCP October 1989 

             local socket) pair that defines the connection is busy
             and cannot be reused.  To shorten the time that a given
             port pair is tied up, some TCPs allow a new SYN to be
             accepted in TIME-WAIT state.

     4.2.2.14  Data Communication: RFC-793 Section 3.7, page 40

        Since RFC-793 was written, there has been extensive work on
        TCP algorithms to achieve efficient data communication.
        Later sections of the present document describe required and
        recommended TCP algorithms to determine when to send data
        (Section 4.2.3.4), when to send an acknowledgment (Section
        4.2.3.2), and when to update the window (Section 4.2.3.3).

        DISCUSSION:
             One important performance issue is "Silly Window
             Syndrome" or "SWS" [TCP:5], a stable pattern of small
             incremental window movements resulting in extremely
             poor TCP performance.  Algorithms to avoid SWS are
             described below for both the sending side (Section
             4.2.3.4) and the receiving side (Section 4.2.3.3).

             In brief, SWS is caused by the receiver advancing the
             right window edge whenever it has any new buffer space
             available to receive data and by the sender using any
             incremental window, no matter how small, to send more
             data [TCP:5].  The result can be a stable pattern of
             sending tiny data segments, even though both sender and
             receiver have a large total buffer space for the
             connection.  SWS can only occur during the transmission
             of a large amount of data; if the connection goes
             quiescent, the problem will disappear.  It is caused by
             typical straightforward implementation of window
             management, but the sender and receiver algorithms
             given below will avoid it.

             Another important TCP performance issue is that some
             applications, especially remote login to character-at-
             a-time hosts, tend to send streams of one-octet data
             segments.  To avoid deadlocks, every TCP SEND call from
             such applications must be "pushed", either explicitly
             by the application or else implicitly by TCP.  The
             result may be a stream of TCP segments that contain one
             data octet each, which makes very inefficient use of
             the Internet and contributes to Internet congestion.
             The Nagle Algorithm described in Section 4.2.3.4
             provides a simple and effective solution to this
             problem.  It does have the effect of clumping

RFC1122 TRANSPORT LAYER -- TCP October 1989 

             characters over Telnet connections; this may initially
             surprise users accustomed to single-character echo, but
             user acceptance has not been a problem.

             Note that the Nagle algorithm and the send SWS
             avoidance algorithm play complementary roles in
             improving performance.  The Nagle algorithm discourages
             sending tiny segments when the data to be sent
             increases in small increments, while the SWS avoidance
             algorithm discourages small segments resulting from the
             right window edge advancing in small increments.

             A careless implementation can send two or more
             acknowledgment segments per data segment received.  For
             example, suppose the receiver acknowledges every data
             segment immediately.  When the application program
             subsequently consumes the data and increases the
             available receive buffer space again, the receiver may
             send a second acknowledgment segment to update the
             window at the sender.  The extreme case occurs with
             single-character segments on TCP connections using the
             Telnet protocol for remote login service.  Some
             implementations have been observed in which each
             incoming 1-character segment generates three return
             segments: (1) the acknowledgment, (2) a one byte
             increase in the window, and (3) the echoed character,
             respectively.

     4.2.2.15  Retransmission Timeout: RFC-793 Section 3.7, page 41

        The algorithm suggested in RFC-793 for calculating the
        retransmission timeout is now known to be inadequate; see
        Section 4.2.3.1 below.

        Recent work by Jacobson [TCP:7] on Internet congestion and
        TCP retransmission stability has produced a transmission
        algorithm combining "slow start" with "congestion
        avoidance".  A TCP MUST implement this algorithm.

        If a retransmitted packet is identical to the original
        packet (which implies not only that the data boundaries have
        not changed, but also that the window and acknowledgment
        fields of the header have not changed), then the same IP
        Identification field MAY be used (see Section 3.2.1.5).

        IMPLEMENTATION:
             Some TCP implementors have chosen to "packetize" the
             data stream, i.e., to pick segment boundaries when

RFC1122 TRANSPORT LAYER -- TCP October 1989 

             segments are originally sent and to queue these
             segments in a "retransmission queue" until they are
             acknowledged.  Another design (which may be simpler) is
             to defer packetizing until each time data is
             transmitted or retransmitted, so there will be no
             segment retransmission queue.

             In an implementation with a segment retransmission
             queue, TCP performance may be enhanced by repacketizing
             the segments awaiting acknowledgment when the first
             retransmission timeout occurs.  That is, the
             outstanding segments that fitted would be combined into
             one maximum-sized segment, with a new IP Identification
             value.  The TCP would then retain this combined segment
             in the retransmit queue until it was acknowledged.
             However, if the first two segments in the
             retransmission queue totalled more than one maximum-
             sized segment, the TCP would retransmit only the first
             segment using the original IP Identification field.

     4.2.2.16  Managing the Window: RFC-793 Section 3.7, page 41

        A TCP receiver SHOULD NOT shrink the window, i.e., move the
        right window edge to the left.  However, a sending TCP MUST
        be robust against window shrinking, which may cause the
        "useable window" (see Section 4.2.3.4) to become negative.

        If this happens, the sender SHOULD NOT send new data, but
        SHOULD retransmit normally the old unacknowledged data
        between SND.UNA and SND.UNA+SND.WND.  The sender MAY also
        retransmit old data beyond SND.UNA+SND.WND, but SHOULD NOT
        time out the connection if data beyond the right window edge
        is not acknowledged.  If the window shrinks to zero, the TCP
        MUST probe it in the standard way (see next Section).

        DISCUSSION:
             Many TCP implementations become confused if the window
             shrinks from the right after data has been sent into a
             larger window.  Note that TCP has a heuristic to select
             the latest window update despite possible datagram
             reordering; as a result, it may ignore a window update
             with a smaller window than previously offered if
             neither the sequence number nor the acknowledgment
             number is increased.

RFC1122 TRANSPORT LAYER -- TCP October 1989 

     4.2.2.17  Probing Zero Windows: RFC-793 Section 3.7, page 42

        Probing of zero (offered) windows MUST be supported.

        A TCP MAY keep its offered receive window closed
        indefinitely.  As long as the receiving TCP continues to
        send acknowledgments in response to the probe segments, the
        sending TCP MUST allow the connection to stay open.

        DISCUSSION:
             It is extremely important to remember that ACK
             (acknowledgment) segments that contain no data are not
             reliably transmitted by TCP.  If zero window probing is
             not supported, a connection may hang forever when an
             ACK segment that re-opens the window is lost.

             The delay in opening a zero window generally occurs
             when the receiving application stops taking data from
             its TCP.  For example, consider a printer daemon
             application, stopped because the printer ran out of
             paper.

        The transmitting host SHOULD send the first zero-window
        probe when a zero window has existed for the retransmission
        timeout period (see Section 4.2.2.15), and SHOULD increase
        exponentially the interval between successive probes.

        DISCUSSION:
             This procedure minimizes delay if the zero-window
             condition is due to a lost ACK segment containing a
             window-opening update.  Exponential backoff is
             recommended, possibly with some maximum interval not
             specified here.  This procedure is similar to that of
             the retransmission algorithm, and it may be possible to
             combine the two procedures in the implementation.

     4.2.2.18  Passive OPEN Calls:  RFC-793 Section 3.8

        Every passive OPEN call either creates a new connection
        record in LISTEN state, or it returns an error; it MUST NOT
        affect any previously created connection record.

        A TCP that supports multiple concurrent users MUST provide
        an OPEN call that will functionally allow an application to
        LISTEN on a port while a connection block with the same
        local port is in SYN-SENT or SYN-RECEIVED state.

        DISCUSSION:

RFC1122 TRANSPORT LAYER -- TCP October 1989 

             Some applications (e.g., SMTP servers) may need to
             handle multiple connection attempts at about the same
             time.  The probability of a connection attempt failing
             is reduced by giving the application some means of
             listening for a new connection at the same time that an
             earlier connection attempt is going through the three-
             way handshake.

        IMPLEMENTATION:
             Acceptable implementations of concurrent opens may
             permit multiple passive OPEN calls, or they may allow
             "cloning" of LISTEN-state connections from a single
             passive OPEN call.

     4.2.2.19  Time to Live: RFC-793 Section 3.9, page 52

        RFC-793 specified that TCP was to request the IP layer to
        send TCP segments with TTL = 60.  This is obsolete; the TTL
        value used to send TCP segments MUST be configurable.  See
        Section 3.2.1.7 for discussion.

     4.2.2.20  Event Processing: RFC-793 Section 3.9

        While it is not strictly required, a TCP SHOULD be capable
        of queueing out-of-order TCP segments.  Change the "may" in
        the last sentence of the first paragraph on page 70 to
        "should".

        DISCUSSION:
             Some small-host implementations have omitted segment
             queueing because of limited buffer space.  This
             omission may be expected to adversely affect TCP
             throughput, since loss of a single segment causes all
             later segments to appear to be "out of sequence".

        In general, the processing of received segments MUST be
        implemented to aggregate ACK segments whenever possible.
        For example, if the TCP is processing a series of queued
        segments, it MUST process them all before sending any ACK
        segments.

        Here are some detailed error corrections and notes on the
        Event Processing section of RFC-793.

        (a)  CLOSE Call, CLOSE-WAIT state, p. 61: enter LAST-ACK
             state, not CLOSING.

        (b)  LISTEN state, check for SYN (pp. 65, 66): With a SYN

RFC1122 TRANSPORT LAYER -- TCP October 1989 

             bit, if the security/compartment or the precedence is
             wrong for the segment, a reset is sent.  The wrong form
             of reset is shown in the text; it should be:

               <SEQ=0><ACK=SEG.SEQ+SEG.LEN><CTL=RST,ACK>

        (c)  SYN-SENT state, Check for SYN, p. 68: When the
             connection enters ESTABLISHED state, the following
             variables must be set:
                SND.WND <- SEG.WND
                SND.WL1 <- SEG.SEQ
                SND.WL2 <- SEG.ACK

        (d)  Check security and precedence, p. 71: The first heading
             "ESTABLISHED STATE" should really be a list of all
             states other than SYN-RECEIVED: ESTABLISHED, FIN-WAIT-
             1, FIN-WAIT-2, CLOSE-WAIT, CLOSING, LAST-ACK, and
             TIME-WAIT.

        (e)  Check SYN bit, p. 71:  "In SYN-RECEIVED state and if
             the connection was initiated with a passive OPEN, then
             return this connection to the LISTEN state and return.
             Otherwise...".

        (f)  Check ACK field, SYN-RECEIVED state, p. 72: When the
             connection enters ESTABLISHED state, the variables
             listed in (c) must be set.

        (g)  Check ACK field, ESTABLISHED state, p. 72: The ACK is a
             duplicate if SEG.ACK =< SND.UNA (the = was omitted).
             Similarly, the window should be updated if: SND.UNA =<
             SEG.ACK =< SND.NXT.

        (h)  USER TIMEOUT, p. 77:

             It would be better to notify the application of the
             timeout rather than letting TCP force the connection
             closed.  However, see also Section 4.2.3.5.

     4.2.2.21  Acknowledging Queued Segments: RFC-793 Section 3.9

        A TCP MAY send an ACK segment acknowledging RCV.NXT when a
        valid segment arrives that is in the window but not at the
        left window edge.

RFC1122 TRANSPORT LAYER -- TCP October 1989 

        DISCUSSION:
             RFC-793 (see page 74) was ambiguous about whether or
             not an ACK segment should be sent when an out-of-order
             segment was received, i.e., when SEG.SEQ was unequal to
             RCV.NXT.

             One reason for ACKing out-of-order segments might be to
             support an experimental algorithm known as "fast
             retransmit".   With this algorithm, the sender uses the
             "redundant" ACK's to deduce that a segment has been
             lost before the retransmission timer has expired.  It
             counts the number of times an ACK has been received
             with the same value of SEG.ACK and with the same right
             window edge.  If more than a threshold number of such
             ACK's is received, then the segment containing the
             octets starting at SEG.ACK is assumed to have been lost
             and is retransmitted, without awaiting a timeout.  The
             threshold is chosen to compensate for the maximum
             likely segment reordering in the Internet.  There is
             not yet enough experience with the fast retransmit
             algorithm to determine how useful it is.

  4.2.3  SPECIFIC ISSUES

     4.2.3.1  Retransmission Timeout Calculation

        A host TCP MUST implement Karn's algorithm and Jacobson's
        algorithm for computing the retransmission timeout ("RTO").

        o    Jacobson's algorithm for computing the smoothed round-
             trip ("RTT") time incorporates a simple measure of the
             variance [TCP:7].

        o    Karn's algorithm for selecting RTT measurements ensures
             that ambiguous round-trip times will not corrupt the
             calculation of the smoothed round-trip time [TCP:6].

        This implementation also MUST include "exponential backoff"
        for successive RTO values for the same segment.
        Retransmission of SYN segments SHOULD use the same algorithm
        as data segments.

        DISCUSSION:
             There were two known problems with the RTO calculations
             specified in RFC-793.  First, the accurate measurement
             of RTTs is difficult when there are retransmissions.
             Second, the algorithm to compute the smoothed round-
             trip time is inadequate [TCP:7], because it incorrectly

RFC1122 TRANSPORT LAYER -- TCP October 1989 

             assumed that the variance in RTT values would be small
             and constant.  These problems were solved by Karn's and
             Jacobson's algorithm, respectively.

             The performance increase resulting from the use of
             these improvements varies from noticeable to dramatic.
             Jacobson's algorithm for incorporating the measured RTT
             variance is especially important on a low-speed link,
             where the natural variation of packet sizes causes a
             large variation in RTT.  One vendor found link
             utilization on a 9.6kb line went from 10% to 90% as a
             result of implementing Jacobson's variance algorithm in
             TCP.

        The following values SHOULD be used to initialize the
        estimation parameters for a new connection:

        (a)  RTT = 0 seconds.

        (b)  RTO = 3 seconds.  (The smoothed variance is to be
             initialized to the value that will result in this RTO).

        The recommended upper and lower bounds on the RTO are known
        to be inadequate on large internets.  The lower bound SHOULD
        be measured in fractions of a second (to accommodate high
        speed LANs) and the upper bound should be 2*MSL, i.e., 240
        seconds.

        DISCUSSION:
             Experience has shown that these initialization values
             are reasonable, and that in any case the Karn and
             Jacobson algorithms make TCP behavior reasonably
             insensitive to the initial parameter choices.

     4.2.3.2  When to Send an ACK Segment

        A host that is receiving a stream of TCP data segments can
        increase efficiency in both the Internet and the hosts by
        sending fewer than one ACK (acknowledgment) segment per data
        segment received; this is known as a "delayed ACK" [TCP:5].

        A TCP SHOULD implement a delayed ACK, but an ACK should not
        be excessively delayed; in particular, the delay MUST be
        less than 0.5 seconds, and in a stream of full-sized
        segments there SHOULD be an ACK for at least every second
        segment.

        DISCUSSION:

RFC1122 TRANSPORT LAYER -- TCP October 1989 

             A delayed ACK gives the application an opportunity to
             update the window and perhaps to send an immediate
             response.  In particular, in the case of character-mode
             remote login, a delayed ACK can reduce the number of
             segments sent by the server by a factor of 3 (ACK,
             window update, and echo character all combined in one
             segment).

             In addition, on some large multi-user hosts, a delayed
             ACK can substantially reduce protocol processing
             overhead by reducing the total number of packets to be
             processed [TCP:5].  However, excessive delays on ACK's
             can disturb the round-trip timing and packet "clocking"
             algorithms [TCP:7].

     4.2.3.3  When to Send a Window Update

        A TCP MUST include a SWS avoidance algorithm in the receiver
        [TCP:5].

        IMPLEMENTATION:
             The receiver's SWS avoidance algorithm determines when
             the right window edge may be advanced; this is
             customarily known as "updating the window".  This
             algorithm combines with the delayed ACK algorithm (see
             Section 4.2.3.2) to determine when an ACK segment
             containing the current window will really be sent to
             the receiver.  We use the notation of RFC-793; see
             Figures 4 and 5 in that document.

             The solution to receiver SWS is to avoid advancing the
             right window edge RCV.NXT+RCV.WND in small increments,
             even if data is received from the network in small
             segments.

             Suppose the total receive buffer space is RCV.BUFF.  At
             any given moment, RCV.USER octets of this total may be
             tied up with data that has been received and
             acknowledged but which the user process has not yet
             consumed.  When the connection is quiescent, RCV.WND =
             RCV.BUFF and RCV.USER = 0.

             Keeping the right window edge fixed as data arrives and
             is acknowledged requires that the receiver offer less
             than its full buffer space, i.e., the receiver must
             specify a RCV.WND that keeps RCV.NXT+RCV.WND constant
             as RCV.NXT increases.  Thus, the total buffer space
             RCV.BUFF is generally divided into three parts:

RFC1122 TRANSPORT LAYER -- TCP October 1989 

             |<------- RCV.BUFF ---------------->|
                  1             2            3
         ----|---------|------------------|------|----
                    RCV.NXT               ^
                                       (Fixed)

         1 - RCV.USER =  data received but not yet consumed;
         2 - RCV.WND =   space advertised to sender;
         3 - Reduction = space available but not yet
                         advertised.

             The suggested SWS avoidance algorithm for the receiver
             is to keep RCV.NXT+RCV.WND fixed until the reduction
             satisfies:

                  RCV.BUFF - RCV.USER - RCV.WND  >=

                         min( Fr * RCV.BUFF, Eff.snd.MSS )

             where Fr is a fraction whose recommended value is 1/2,
             and Eff.snd.MSS is the effective send MSS for the
             connection (see Section 4.2.2.6).  When the inequality
             is satisfied, RCV.WND is set to RCV.BUFF-RCV.USER.

             Note that the general effect of this algorithm is to
             advance RCV.WND in increments of Eff.snd.MSS (for
             realistic receive buffers:  Eff.snd.MSS < RCV.BUFF/2).
             Note also that the receiver must use its own
             Eff.snd.MSS, assuming it is the same as the sender's.

     4.2.3.4  When to Send Data

        A TCP MUST include a SWS avoidance algorithm in the sender.

        A TCP SHOULD implement the Nagle Algorithm [TCP:9] to
        coalesce short segments.  However, there MUST be a way for
        an application to disable the Nagle algorithm on an
        individual connection.  In all cases, sending data is also
        subject to the limitation imposed by the Slow Start
        algorithm (Section 4.2.2.15).

        DISCUSSION:
             The Nagle algorithm is generally as follows:

                  If there is unacknowledged data (i.e., SND.NXT >
                  SND.UNA), then the sending TCP buffers all user

RFC1122 TRANSPORT LAYER -- TCP October 1989 

                  data (regardless of the PSH bit), until the
                  outstanding data has been acknowledged or until
                  the TCP can send a full-sized segment (Eff.snd.MSS
                  bytes; see Section 4.2.2.6).

             Some applications (e.g., real-time display window
             updates) require that the Nagle algorithm be turned
             off, so small data segments can be streamed out at the
             maximum rate.

        IMPLEMENTATION:
             The sender's SWS avoidance algorithm is more difficult
             than the receivers's, because the sender does not know
             (directly) the receiver's total buffer space RCV.BUFF.
             An approach which has been found to work well is for
             the sender to calculate Max(SND.WND), the maximum send
             window it has seen so far on the connection, and to use
             this value as an estimate of RCV.BUFF.  Unfortunately,
             this can only be an estimate; the receiver may at any
             time reduce the size of RCV.BUFF.  To avoid a resulting
             deadlock, it is necessary to have a timeout to force
             transmission of data, overriding the SWS avoidance
             algorithm.  In practice, this timeout should seldom
             occur.

             The "useable window" [TCP:5] is:

                  U = SND.UNA + SND.WND - SND.NXT

             i.e., the offered window less the amount of data sent
             but not acknowledged.  If D is the amount of data
             queued in the sending TCP but not yet sent, then the
             following set of rules is recommended.

             Send data:

             (1)  if a maximum-sized segment can be sent, i.e, if:

                       min(D,U) >= Eff.snd.MSS;

             (2)  or if the data is pushed and all queued data can
                  be sent now, i.e., if:

                      [SND.NXT = SND.UNA and] PUSHED and D <= U

                  (the bracketed condition is imposed by the Nagle
                  algorithm);

RFC1122 TRANSPORT LAYER -- TCP October 1989 

             (3)  or if at least a fraction Fs of the maximum window
                  can be sent, i.e., if:

                      [SND.NXT = SND.UNA and]

                              min(D.U) >= Fs * Max(SND.WND);

             (4)  or if data is PUSHed and the override timeout
                  occurs.

             Here Fs is a fraction whose recommended value is 1/2.
             The override timeout should be in the range 0.1 - 1.0
             seconds.  It may be convenient to combine this timer
             with the timer used to probe zero windows (Section
             4.2.2.17).

             Finally, note that the SWS avoidance algorithm just
             specified is to be used instead of the sender-side
             algorithm contained in [TCP:5].

     4.2.3.5  TCP Connection Failures

        Excessive retransmission of the same segment by TCP
        indicates some failure of the remote host or the Internet
        path.  This failure may be of short or long duration.  The
        following procedure MUST be used to handle excessive
        retransmissions of data segments [IP:11]:

        (a)  There are two thresholds R1 and R2 measuring the amount
             of retransmission that has occurred for the same
             segment.  R1 and R2 might be measured in time units or
             as a count of retransmissions.

        (b)  When the number of transmissions of the same segment
             reaches or exceeds threshold R1, pass negative advice
             (see Section 3.3.1.4) to the IP layer, to trigger
             dead-gateway diagnosis.

        (c)  When the number of transmissions of the same segment
             reaches a threshold R2 greater than R1, close the
             connection.

        (d)  An application MUST be able to set the value for R2 for
             a particular connection.  For example, an interactive
             application might set R2 to "infinity," giving the user
             control over when to disconnect.

RFC1122 TRANSPORT LAYER -- TCP October 1989 

        (d)  TCP SHOULD inform the application of the delivery
             problem (unless such information has been disabled by
             the application; see Section 4.2.4.1), when R1 is
             reached and before R2.  This will allow a remote login
             (User Telnet) application program to inform the user,
             for example.

        The value of R1 SHOULD correspond to at least 3
        retransmissions, at the current RTO.  The value of R2 SHOULD
        correspond to at least 100 seconds.

        An attempt to open a TCP connection could fail with
        excessive retransmissions of the SYN segment or by receipt
        of a RST segment or an ICMP Port Unreachable.  SYN
        retransmissions MUST be handled in the general way just
        described for data retransmissions, including notification
        of the application layer.

        However, the values of R1 and R2 may be different for SYN
        and data segments.  In particular, R2 for a SYN segment MUST
        be set large enough to provide retransmission of the segment
        for at least 3 minutes.  The application can close the
        connection (i.e., give up on the open attempt) sooner, of
        course.

        DISCUSSION:
             Some Internet paths have significant setup times, and
             the number of such paths is likely to increase in the
             future.

     4.2.3.6  TCP Keep-Alives

        Implementors MAY include "keep-alives" in their TCP
        implementations, although this practice is not universally
        accepted.  If keep-alives are included, the application MUST
        be able to turn them on or off for each TCP connection, and
        they MUST default to off.

        Keep-alive packets MUST only be sent when no data or
        acknowledgement packets have been received for the
        connection within an interval.  This interval MUST be
        configurable and MUST default to no less than two hours.

        It is extremely important to remember that ACK segments that
        contain no data are not reliably transmitted by TCP.
        Consequently, if a keep-alive mechanism is implemented it
        MUST NOT interpret failure to respond to any specific probe
        as a dead connection.

RFC1122 TRANSPORT LAYER -- TCP October 1989 

        An implementation SHOULD send a keep-alive segment with no
        data; however, it MAY be configurable to send a keep-alive
        segment containing one garbage octet, for compatibility with
        erroneous TCP implementations.

        DISCUSSION:
             A "keep-alive" mechanism periodically probes the other
             end of a connection when the connection is otherwise
             idle, even when there is no data to be sent.  The TCP
             specification does not include a keep-alive mechanism
             because it could:  (1) cause perfectly good connections
             to break during transient Internet failures; (2)
             consume unnecessary bandwidth ("if no one is using the
             connection, who cares if it is still good?"); and (3)
             cost money for an Internet path that charges for
             packets.

             Some TCP implementations, however, have included a
             keep-alive mechanism.  To confirm that an idle
             connection is still active, these implementations send
             a probe segment designed to elicit a response from the
             peer TCP.  Such a segment generally contains SEG.SEQ =
             SND.NXT-1 and may or may not contain one garbage octet
             of data.  Note that on a quiet connection SND.NXT =
             RCV.NXT, so that this SEG.SEQ will be outside the
             window.  Therefore, the probe causes the receiver to
             return an acknowledgment segment, confirming that the
             connection is still live.  If the peer has dropped the
             connection due to a network partition or a crash, it
             will respond with a RST instead of an acknowledgment
             segment.

             Unfortunately, some misbehaved TCP implementations fail
             to respond to a segment with SEG.SEQ = SND.NXT-1 unless
             the segment contains data.  Alternatively, an
             implementation could determine whether a peer responded
             correctly to keep-alive packets with no garbage data
             octet.

             A TCP keep-alive mechanism should only be invoked in
             server applications that might otherwise hang
             indefinitely and consume resources unnecessarily if a
             client crashes or aborts a connection during a network
             failure.

RFC1122 TRANSPORT LAYER -- TCP October 1989 

     4.2.3.7  TCP Multihoming

        If an application on a multihomed host does not specify the
        local IP address when actively opening a TCP connection,
        then the TCP MUST ask the IP layer to select a local IP
        address before sending the (first) SYN.  See the function
        GET_SRCADDR() in Section 3.4.

        At all other times, a previous segment has either been sent
        or received on this connection, and TCP MUST use the same
        local address is used that was used in those previous
        segments.

     4.2.3.8  IP Options

        When received options are passed up to TCP from the IP
        layer, TCP MUST ignore options that it does not understand.

        A TCP MAY support the Time Stamp and Record Route options.

        An application MUST be able to specify a source route when
        it actively opens a TCP connection, and this MUST take
        precedence over a source route received in a datagram.

        When a TCP connection is OPENed passively and a packet
        arrives with a completed IP Source Route option (containing
        a return route), TCP MUST save the return route and use it
        for all segments sent on this connection.  If a different
        source route arrives in a later segment, the later
        definition SHOULD override the earlier one.

     4.2.3.9  ICMP Messages

        TCP MUST act on an ICMP error message passed up from the IP
        layer, directing it to the connection that created the
        error.  The necessary demultiplexing information can be
        found in the IP header contained within the ICMP message.

        o    Source Quench

             TCP MUST react to a Source Quench by slowing
             transmission on the connection.  The RECOMMENDED
             procedure is for a Source Quench to trigger a "slow
             start," as if a retransmission timeout had occurred.

        o    Destination Unreachable -- codes 0, 1, 5

             Since these Unreachable messages indicate soft error

RFC1122 TRANSPORT LAYER -- TCP October 1989 

             conditions, TCP MUST NOT abort the connection, and it
             SHOULD make the information available to the
             application.

             DISCUSSION:
                  TCP could report the soft error condition directly
                  to the application layer with an upcall to the
                  ERROR_REPORT routine, or it could merely note the
                  message and report it to the application only when
                  and if the TCP connection times out.

        o    Destination Unreachable -- codes 2-4

             These are hard error conditions, so TCP SHOULD abort
             the connection.

        o    Time Exceeded -- codes 0, 1

             This should be handled the same way as Destination
             Unreachable codes 0, 1, 5 (see above).

        o    Parameter Problem

             This should be handled the same way as Destination
             Unreachable codes 0, 1, 5 (see above).

     4.2.3.10  Remote Address Validation

        A TCP implementation MUST reject as an error a local OPEN
        call for an invalid remote IP address (e.g., a broadcast or
        multicast address).

        An incoming SYN with an invalid source address must be
        ignored either by TCP or by the IP layer (see Section
        3.2.1.3).

        A TCP implementation MUST silently discard an incoming SYN
        segment that is addressed to a broadcast or multicast
        address.

     4.2.3.11  TCP Traffic Patterns

        IMPLEMENTATION:
             The TCP protocol specification [TCP:1] gives the
             implementor much freedom in designing the algorithms
             that control the message flow over the connection --
             packetizing, managing the window, sending

RFC1122 TRANSPORT LAYER -- TCP October 1989 

             acknowledgments, etc.  These design decisions are
             difficult because a TCP must adapt to a wide range of
             traffic patterns.  Experience has shown that a TCP
             implementor needs to verify the design on two extreme
             traffic patterns:

             o    Single-character Segments

                  Even if the sender is using the Nagle Algorithm,
                  when a TCP connection carries remote login traffic
                  across a low-delay LAN the receiver will generally
                  get a stream of single-character segments.  If
                  remote terminal echo mode is in effect, the
                  receiver's system will generally echo each
                  character as it is received.

             o    Bulk Transfer

                  When TCP is used for bulk transfer, the data
                  stream should be made up (almost) entirely of
                  segments of the size of the effective MSS.
                  Although TCP uses a sequence number space with
                  byte (octet) granularity, in bulk-transfer mode
                  its operation should be as if TCP used a sequence
                  space that counted only segments.

             Experience has furthermore shown that a single TCP can
             effectively and efficiently handle these two extremes.

             The most important tool for verifying a new TCP
             implementation is a packet trace program.  There is a
             large volume of experience showing the importance of
             tracing a variety of traffic patterns with other TCP
             implementations and studying the results carefully.

     4.2.3.12  Efficiency

        IMPLEMENTATION:
             Extensive experience has led to the following
             suggestions for efficient implementation of TCP:

             (a)  Don't Copy Data

                  In bulk data transfer, the primary CPU-intensive
                  tasks are copying data from one place to another
                  and checksumming the data.  It is vital to
                  minimize the number of copies of TCP data.  Since

RFC1122 TRANSPORT LAYER -- TCP October 1989 

                  the ultimate speed limitation may be fetching data
                  across the memory bus, it may be useful to combine
                  the copy with checksumming, doing both with a
                  single memory fetch.

             (b)  Hand-Craft the Checksum Routine

                  A good TCP checksumming routine is typically two
                  to five times faster than a simple and direct
                  implementation of the definition.  Great care and
                  clever coding are often required and advisable to
                  make the checksumming code "blazing fast".  See
                  [TCP:10].

             (c)  Code for the Common Case

                  TCP protocol processing can be complicated, but
                  for most segments there are only a few simple
                  decisions to be made.  Per-segment processing will
                  be greatly speeded up by coding the main line to
                  minimize the number of decisions in the most
                  common case.

  4.2.4  TCP/APPLICATION LAYER INTERFACE

     4.2.4.1  Asynchronous Reports

        There MUST be a mechanism for reporting soft TCP error
        conditions to the application.  Generically, we assume this
        takes the form of an application-supplied ERROR_REPORT
        routine that may be upcalled [INTRO:7] asynchronously from
        the transport layer:

           ERROR_REPORT(local connection name, reason, subreason)

        The precise encoding of the reason and subreason parameters
        is not specified here.  However, the conditions that are
        reported asynchronously to the application MUST include:

        *    ICMP error message arrived (see 4.2.3.9)

        *    Excessive retransmissions (see 4.2.3.5)

        *    Urgent pointer advance (see 4.2.2.4).

        However, an application program that does not want to
        receive such ERROR_REPORT calls SHOULD be able to

RFC1122 TRANSPORT LAYER -- TCP October 1989 

        effectively disable these calls.

        DISCUSSION:
             These error reports generally reflect soft errors that
             can be ignored without harm by many applications.  It
             has been suggested that these error report calls should
             default to "disabled," but this is not required.

     4.2.4.2  Type-of-Service

        The application layer MUST be able to specify the Type-of-
        Service (TOS) for segments that are sent on a connection.
        It not required, but the application SHOULD be able to
        change the TOS during the connection lifetime.  TCP SHOULD
        pass the current TOS value without change to the IP layer,
        when it sends segments on the connection.

        The TOS will be specified independently in each direction on
        the connection, so that the receiver application will
        specify the TOS used for ACK segments.

        TCP MAY pass the most recently received TOS up to the
        application.

        DISCUSSION
             Some applications (e.g., SMTP) change the nature of
             their communication during the lifetime of a
             connection, and therefore would like to change the TOS
             specification.

             Note also that the OPEN call specified in RFC-793
             includes a parameter ("options") in which the caller
             can specify IP options such as source route, record
             route, or timestamp.

     4.2.4.3  Flush Call

        Some TCP implementations have included a FLUSH call, which
        will empty the TCP send queue of any data for which the user
        has issued SEND calls but which is still to the right of the
        current send window.  That is, it flushes as much queued
        send data as possible without losing sequence number
        synchronization.  This is useful for implementing the "abort
        output" function of Telnet.

RFC1122 TRANSPORT LAYER -- TCP October 1989 

     4.2.4.4  Multihoming

        The user interface outlined in sections 2.7 and 3.8 of RFC-
        793 needs to be extended for multihoming.  The OPEN call
        MUST have an optional parameter:

            OPEN( ... [local IP address,] ... )

        to allow the specification of the local IP address.

        DISCUSSION:
             Some TCP-based applications need to specify the local
             IP address to be used to open a particular connection;
             FTP is an example.

        IMPLEMENTATION:
             A passive OPEN call with a specified "local IP address"
             parameter will await an incoming connection request to
             that address.  If the parameter is unspecified, a
             passive OPEN will await an incoming connection request
             to any local IP address, and then bind the local IP
             address of the connection to the particular address
             that is used.

             For an active OPEN call, a specified "local IP address"
             parameter will be used for opening the connection.  If
             the parameter is unspecified, the networking software
             will choose an appropriate local IP address (see
             Section 3.3.4.2) for the connection

  4.2.5  TCP REQUIREMENT SUMMARY

                                             |        | | | |S| |
                                             |        | | | |H| |F
                                             |        | | | |O|M|o
                                             |        | |S| |U|U|o
                                             |        | |H| |L|S|t
                                             |        |M|O| |D|T|n
                                             |        |U|U|M| | |o
                                             |        |S|L|A|N|N|t
                                             |        |T|D|Y|O|O|t

FEATURE |SECTION | | | |T|T|e

|--------|-|-|-|-|-|-- 

                                             |        | | | | | |

Push flag | | | | | | | 

 Aggregate or queue un-pushed data              |4.2.2.2 | | |x| | |
 Sender collapse successive PSH flags           |4.2.2.2 | |x| | | |
 SEND call can specify PUSH                     |4.2.2.2 | | |x| | |

RFC1122 TRANSPORT LAYER -- TCP October 1989 

If cannot: sender buffer indefinitely        |4.2.2.2 | | | | |x|
If cannot: PSH last segment                  |4.2.2.2 |x| | | | |
 Notify receiving ALP of PSH                    |4.2.2.2 | | |x| | |1
 Send max size segment when possible            |4.2.2.2 | |x| | | |
                                             |        | | | | | |

Window | | | | | | | 

 Treat as unsigned number                       |4.2.2.3 |x| | | | |
 Handle as 32-bit number                        |4.2.2.3 | |x| | | |
 Shrink window from right                       |4.2.2.16| | | |x| |
 Robust against shrinking window                |4.2.2.16|x| | | | |
 Receiver's window closed indefinitely          |4.2.2.17| | |x| | |
 Sender probe zero window                       |4.2.2.17|x| | | | |
First probe after RTO                        |4.2.2.17| |x| | | |
Exponential backoff                          |4.2.2.17| |x| | | |
 Allow window stay zero indefinitely            |4.2.2.17|x| | | | |
 Sender timeout OK conn with zero wind          |4.2.2.17| | | | |x|
                                             |        | | | | | |

Urgent Data | | | | | | | 

 Pointer points to last octet                   |4.2.2.4 |x| | | | |
 Arbitrary length urgent data sequence          |4.2.2.4 |x| | | | |
 Inform ALP asynchronously of urgent data       |4.2.2.4 |x| | | | |1
 ALP can learn if/how much urgent data Q'd      |4.2.2.4 |x| | | | |1
                                             |        | | | | | |

TCP Options | | | | | | | 

 Receive TCP option in any segment              |4.2.2.5 |x| | | | |
 Ignore unsupported options                     |4.2.2.5 |x| | | | |
 Cope with illegal option length                |4.2.2.5 |x| | | | |
 Implement sending & receiving MSS option       |4.2.2.6 |x| | | | |
 Send MSS option unless 536                     |4.2.2.6 | |x| | | |
 Send MSS option always                         |4.2.2.6 | | |x| | |
 Send-MSS default is 536                        |4.2.2.6 |x| | | | |
 Calculate effective send seg size              |4.2.2.6 |x| | | | |
                                             |        | | | | | |

TCP Checksums | | | | | | | 

 Sender compute checksum                        |4.2.2.7 |x| | | | |
 Receiver check checksum                        |4.2.2.7 |x| | | | |
                                             |        | | | | | |

Use clock-driven ISN selection |4.2.2.9 |x| | | | | 

                                             |        | | | | | |

Opening Connections | | | | | | | 

 Support simultaneous open attempts             |4.2.2.10|x| | | | |
 SYN-RCVD remembers last state                  |4.2.2.11|x| | | | |
 Passive Open call interfere with others        |4.2.2.18| | | | |x|
 Function: simultan. LISTENs for same port      |4.2.2.18|x| | | | |
 Ask IP for src address for SYN if necc.        |4.2.3.7 |x| | | | |
Otherwise, use local addr of conn.           |4.2.3.7 |x| | | | |
 OPEN to broadcast/multicast IP Address         |4.2.3.14| | | | |x|
 Silently discard seg to bcast/mcast addr       |4.2.3.14|x| | | | |

RFC1122 TRANSPORT LAYER -- TCP October 1989 

                                             |        | | | | | |

Closing Connections | | | | | | | 

 RST can contain data                           |4.2.2.12| |x| | | |
 Inform application of aborted conn             |4.2.2.13|x| | | | |
 Half-duplex close connections                  |4.2.2.13| | |x| | |
Send RST to indicate data lost               |4.2.2.13| |x| | | |
 In TIME-WAIT state for 2xMSL seconds           |4.2.2.13|x| | | | |
Accept SYN from TIME-WAIT state              |4.2.2.13| | |x| | |
                                             |        | | | | | |

Retransmissions | | | | | | | 

 Jacobson Slow Start algorithm                  |4.2.2.15|x| | | | |
 Jacobson Congestion-Avoidance algorithm        |4.2.2.15|x| | | | |
 Retransmit with same IP ident                  |4.2.2.15| | |x| | |
 Karn's algorithm                               |4.2.3.1 |x| | | | |
 Jacobson's RTO estimation alg.                 |4.2.3.1 |x| | | | |
 Exponential backoff                            |4.2.3.1 |x| | | | |
 SYN RTO calc same as data                      |4.2.3.1 | |x| | | |
 Recommended initial values and bounds          |4.2.3.1 | |x| | | |
                                             |        | | | | | |

Generating ACK's: | | | | | | | 

 Queue out-of-order segments                    |4.2.2.20| |x| | | |
 Process all Q'd before send ACK                |4.2.2.20|x| | | | |
 Send ACK for out-of-order segment              |4.2.2.21| | |x| | |
 Delayed ACK's                                  |4.2.3.2 | |x| | | |
Delay < 0.5 seconds                          |4.2.3.2 |x| | | | |
Every 2nd full-sized segment ACK'd           |4.2.3.2 |x| | | | |
 Receiver SWS-Avoidance Algorithm               |4.2.3.3 |x| | | | |
                                             |        | | | | | |

Sending data | | | | | | | 

 Configurable TTL                               |4.2.2.19|x| | | | |
 Sender SWS-Avoidance Algorithm                 |4.2.3.4 |x| | | | |
 Nagle algorithm                                |4.2.3.4 | |x| | | |
Application can disable Nagle algorithm      |4.2.3.4 |x| | | | |
                                             |        | | | | | |

Connection Failures: | | | | | | | 

 Negative advice to IP on R1 retxs              |4.2.3.5 |x| | | | |
 Close connection on R2 retxs                   |4.2.3.5 |x| | | | |
 ALP can set R2                                 |4.2.3.5 |x| | | | |1
 Inform ALP of  R1<=retxs<R2                    |4.2.3.5 | |x| | | |1
 Recommended values for R1, R2                  |4.2.3.5 | |x| | | |
 Same mechanism for SYNs                        |4.2.3.5 |x| | | | |
R2 at least 3 minutes for SYN                |4.2.3.5 |x| | | | |
                                             |        | | | | | |

Send Keep-alive Packets: |4.2.3.6 | | |x| | | 

 - Application can request                      |4.2.3.6 |x| | | | |
 - Default is "off"                             |4.2.3.6 |x| | | | |
 - Only send if idle for interval               |4.2.3.6 |x| | | | |
 - Interval configurable                        |4.2.3.6 |x| | | | |

RFC1122 TRANSPORT LAYER -- TCP October 1989 

 - Default at least 2 hrs.                      |4.2.3.6 |x| | | | |
 - Tolerant of lost ACK's                       |4.2.3.6 |x| | | | |
                                             |        | | | | | |

IP Options | | | | | | | 

 Ignore options TCP doesn't understand          |4.2.3.8 |x| | | | |
 Time Stamp support                             |4.2.3.8 | | |x| | |
 Record Route support                           |4.2.3.8 | | |x| | |
 Source Route:                                  |        | | | | | |
ALP can specify                              |4.2.3.8 |x| | | | |1
  Overrides src rt in datagram               |4.2.3.8 |x| | | | |
Build return route from src rt               |4.2.3.8 |x| | | | |
Later src route overrides                    |4.2.3.8 | |x| | | |
                                             |        | | | | | |

Receiving ICMP Messages from IP |4.2.3.9 |x| | | | | 

 Dest. Unreach (0,1,5) => inform ALP            |4.2.3.9 | |x| | | |
 Dest. Unreach (0,1,5) => abort conn            |4.2.3.9 | | | | |x|
 Dest. Unreach (2-4) => abort conn              |4.2.3.9 | |x| | | |
 Source Quench => slow start                    |4.2.3.9 | |x| | | |
 Time Exceeded => tell ALP, don't abort         |4.2.3.9 | |x| | | |
 Param Problem => tell ALP, don't abort         |4.2.3.9 | |x| | | |
                                             |        | | | | | |

Address Validation | | | | | | | 

 Reject OPEN call to invalid IP address         |4.2.3.10|x| | | | |
 Reject SYN from invalid IP address             |4.2.3.10|x| | | | |
 Silently discard SYN to bcast/mcast addr       |4.2.3.10|x| | | | |
                                             |        | | | | | |

TCP/ALP Interface Services | | | | | | | 

 Error Report mechanism                         |4.2.4.1 |x| | | | |
 ALP can disable Error Report Routine           |4.2.4.1 | |x| | | |
 ALP can specify TOS for sending                |4.2.4.2 |x| | | | |
Passed unchanged to IP                       |4.2.4.2 | |x| | | |
 ALP can change TOS during connection           |4.2.4.2 | |x| | | |
 Pass received TOS up to ALP                    |4.2.4.2 | | |x| | |
 FLUSH call                                     |4.2.4.3 | | |x| | |
 Optional local IP addr parm. in OPEN           |4.2.4.4 |x| | | | |

|--------|-|-|-|-|-|--

|--------|-|-|-|-|-|--

FOOTNOTES:

(1) "ALP" means Application-Layer program.

RFC1122 TRANSPORT LAYER -- TCP October 1989

REFERENCES

INTRODUCTORY REFERENCES

[INTRO:1] "Requirements for Internet Hosts -- Application and Support," 

 IETF Host Requirements Working Group, R. Braden, Ed., RFC-1123,
 October 1989.

[INTRO:2] "Requirements for Internet Gateways," R. Braden and J. 

 Postel, RFC-1009, June 1987.

[INTRO:3] "DDN Protocol Handbook," NIC-50004, NIC-50005, NIC-50006, 

 (three volumes), SRI International, December 1985.

[INTRO:4] "Official Internet Protocols," J. Reynolds and J. Postel, 

 RFC-1011, May 1987.

 This document is republished periodically with new RFC numbers; the
 latest version must be used.

[INTRO:5] "Protocol Document Order Information," O. Jacobsen and J. 

 Postel, RFC-980, March 1986.

[INTRO:6] "Assigned Numbers," J. Reynolds and J. Postel, RFC-1010, May 

 1987.

 This document is republished periodically with new RFC numbers; the
 latest version must be used.

[INTRO:7] "Modularity and Efficiency in Protocol Implementations," D. 

 Clark, RFC-817, July 1982.

[INTRO:8] "The Structuring of Systems Using Upcalls," D. Clark, 10th ACM 

 SOSP, Orcas Island, Washington, December 1985.

Secondary References:

[INTRO:9] "A Protocol for Packet Network Intercommunication," V. Cerf 

 and R. Kahn, IEEE Transactions on Communication, May 1974.

[INTRO:10] "The ARPA Internet Protocol," J. Postel, C. Sunshine, and D. 

 Cohen, Computer Networks, Vol. 5, No. 4, July 1981.

[INTRO:11] "The DARPA Internet Protocol Suite," B. Leiner, J. Postel, 

 R. Cole and D. Mills, Proceedings INFOCOM 85, IEEE, Washington DC,

RFC1122 TRANSPORT LAYER -- TCP October 1989 

 March 1985.  Also in: IEEE Communications Magazine, March 1985.
 Also available as ISI-RS-85-153.

[INTRO:12] "Final Text of DIS8473, Protocol for Providing the 

 Connectionless Mode Network Service," ANSI, published as RFC-994,
 March 1986.

[INTRO:13] "End System to Intermediate System Routing Exchange 

 Protocol," ANSI X3S3.3, published as RFC-995, April 1986.

LINK LAYER REFERENCES

[LINK:1] "Trailer Encapsulations," S. Leffler and M. Karels, RFC-893, 

 April 1984.

[LINK:2] "An Ethernet Address Resolution Protocol," D. Plummer, RFC-826, 

 November 1982.

[LINK:3] "A Standard for the Transmission of IP Datagrams over Ethernet 

 Networks," C. Hornig, RFC-894, April 1984.

[LINK:4] "A Standard for the Transmission of IP Datagrams over IEEE 802 

 "Networks," J. Postel and J. Reynolds, RFC-1042, February 1988.

 This RFC contains a great deal of information of importance to
 Internet implementers planning to use IEEE 802 networks.

IP LAYER REFERENCES

[IP:1] "Internet Protocol (IP)," J. Postel, RFC-791, September 1981.

[IP:2] "Internet Control Message Protocol (ICMP)," J. Postel, RFC-792, 

 September 1981.

[IP:3] "Internet Standard Subnetting Procedure," J. Mogul and J. Postel, 

 RFC-950, August 1985.

[IP:4] "Host Extensions for IP Multicasting," S. Deering, RFC-1112, 

 August 1989.

[IP:5] "Military Standard Internet Protocol," MIL-STD-1777, Department 

 of Defense, August 1983.

 This specification, as amended by RFC-963, is intended to describe

RFC1122 TRANSPORT LAYER -- TCP October 1989 

 the Internet Protocol but has some serious omissions (e.g., the
 mandatory subnet extension [IP:3] and the optional multicasting
 extension [IP:4]).  It is also out of date.  If there is a
 conflict, RFC-791, RFC-792, and RFC-950 must be taken as
 authoritative, while the present document is authoritative over
 all.

[IP:6] "Some Problems with the Specification of the Military Standard 

 Internet Protocol," D. Sidhu, RFC-963, November 1985.

[IP:7] "The TCP Maximum Segment Size and Related Topics," J. Postel, 

 RFC-879, November 1983.

 Discusses and clarifies the relationship between the TCP Maximum
 Segment Size option and the IP datagram size.

[IP:8] "Internet Protocol Security Options," B. Schofield, RFC-1108, 

 October 1989.

[IP:9] "Fragmentation Considered Harmful," C. Kent and J. Mogul, ACM 

 SIGCOMM-87, August 1987.  Published as ACM Comp Comm Review, Vol.
 17, no. 5.

 This useful paper discusses the problems created by Internet
 fragmentation and presents alternative solutions.

[IP:10] "IP Datagram Reassembly Algorithms," D. Clark, RFC-815, July 

 1982.

 This and the following paper should be read by every implementor.

[IP:11] "Fault Isolation and Recovery," D. Clark, RFC-816, July 1982.

SECONDARY IP REFERENCES:

[IP:12] "Broadcasting Internet Datagrams in the Presence of Subnets," J. 

 Mogul, RFC-922, October 1984.

[IP:13] "Name, Addresses, Ports, and Routes," D. Clark, RFC-814, July 

 1982.

[IP:14] "Something a Host Could Do with Source Quench: The Source Quench 

 Introduced Delay (SQUID)," W. Prue and J. Postel, RFC-1016, July
 1987.

 This RFC first described directed broadcast addresses.  However,
 the bulk of the RFC is concerned with gateways, not hosts.

RFC1122 TRANSPORT LAYER -- TCP October 1989

UDP REFERENCES:

[UDP:1] "User Datagram Protocol," J. Postel, RFC-768, August 1980.

TCP REFERENCES:

[TCP:1] "Transmission Control Protocol," J. Postel, RFC-793, September 

 1981.

[TCP:2] "Transmission Control Protocol," MIL-STD-1778, US Department of 

 Defense, August 1984.

 This specification as amended by RFC-964 is intended to describe
 the same protocol as RFC-793 [TCP:1].  If there is a conflict,
 RFC-793 takes precedence, and the present document is authoritative
 over both.

[TCP:3] "Some Problems with the Specification of the Military Standard 

 Transmission Control Protocol," D. Sidhu and T. Blumer, RFC-964,
 November 1985.

[TCP:4] "The TCP Maximum Segment Size and Related Topics," J. Postel, 

 RFC-879, November 1983.

[TCP:5] "Window and Acknowledgment Strategy in TCP," D. Clark, RFC-813, 

 July 1982.

[TCP:6] "Round Trip Time Estimation," P. Karn & C. Partridge, ACM 

 SIGCOMM-87, August 1987.

[TCP:7] "Congestion Avoidance and Control," V. Jacobson, ACM SIGCOMM-88, 

 August 1988.

SECONDARY TCP REFERENCES:

[TCP:8] "Modularity and Efficiency in Protocol Implementation," D. 

 Clark, RFC-817, July 1982.

RFC1122 TRANSPORT LAYER -- TCP October 1989

[TCP:9] "Congestion Control in IP/TCP," J. Nagle, RFC-896, January 1984.

[TCP:10] "Computing the Internet Checksum," R. Braden, D. Borman, and C. 

 Partridge, RFC-1071, September 1988.

[TCP:11] "TCP Extensions for Long-Delay Paths," V. Jacobson & R. Braden, 

 RFC-1072, October 1988.

Security Considerations

There are many security issues in the communication layers of host software, but a full discussion is beyond the scope of this RFC.

The Internet architecture generally provides little protection against spoofing of IP source addresses, so any security mechanism that is based upon verifying the IP source address of a datagram should be treated with suspicion. However, in restricted environments some source-address checking may be possible. For example, there might be a secure LAN whose gateway to the rest of the Internet discarded any incoming datagram with a source address that spoofed the LAN address. In this case, a host on the LAN could use the source address to test for local vs. remote source. This problem is complicated by source routing, and some have suggested that source-routed datagram forwarding by hosts (see Section 3.3.5) should be outlawed for security reasons.

Security-related issues are mentioned in sections concerning the IP Security option (Section 3.2.1.8), the ICMP Parameter Problem message (Section 3.2.2.5), IP options in UDP datagrams (Section 4.1.3.2), and reserved TCP ports (Section 4.2.2.1).

Author's Address

Robert Braden USC/Information Sciences Institute 4676 Admiralty Way Marina del Rey, CA 90292-6695

Phone: (213) 822 1511

EMail: [email protected]

Retrieved from "https://rfc-wiki.org/index.php?title=RFC1122&oldid=19174"