RFC 9187




Independent Submission                                          J. Touch
Request for Comments: 9187                        Independent Consultant
Category: Informational                                     January 2022
ISSN: 2070-1721


            Sequence Number Extension for Windowed Protocols

Abstract



   Sliding window protocols use finite sequence numbers to determine
   segment placement and order.  These sequence number spaces wrap
   around and are reused during the operation of such protocols.  This
   document describes a way to extend the size of these sequence numbers
   at the endpoints to avoid the impact of that wrap and reuse without
   transmitting additional information in the packet header.  The
   resulting extended sequence numbers can be used at the endpoints in
   encryption and authentication algorithms to ensure input bit patterns
   do not repeat over the lifetime of a connection.

Status of This Memo



   This document is not an Internet Standards Track specification; it is
   published for informational purposes.

   This is a contribution to the RFC Series, independently of any other
   RFC stream.  The RFC Editor has chosen to publish this document at
   its discretion and makes no statement about its value for
   implementation or deployment.  Documents approved for publication by
   the RFC Editor are not candidates for any level of Internet Standard;
   see Section 2 of RFC 7841.

   Information about the current status of this document, any errata,
   and how to provide feedback on it may be obtained at
   https://www.rfc-editor.org/info/rfc9187.

Copyright Notice



   Copyright (c) 2022 IETF Trust and the persons identified as the
   document authors.  All rights reserved.

   This document is subject to BCP 78 and the IETF Trust's Legal
   Provisions Relating to IETF Documents
   (https://trustee.ietf.org/license-info) in effect on the date of
   publication of this document.  Please review these documents
   carefully, as they describe your rights and restrictions with respect
   to this document.

Table of Contents



   1.  Introduction
   2.  Background
   3.  Related Discussion
   4.  Using SNE in Protocol Design
   5.  Example Code
   6.  Validation Suite
   7.  Security Considerations
   8.  IANA Considerations
   9.  Informative References
   Acknowledgments

   Author's Address



1.  Introduction



   Protocols use sequence numbers to maintain ordering and, in sliding
   window systems, to control the amount of outstanding unacknowledged
   information.  These sequence numbers are finite and thus commonly
   wrap around during long connections, reusing past values.

   It can be useful for protocols to keep track of this wrap around in a
   separate counter, such that the sequence number and counter together
   form an equivalent number space that need not wrap.  This technique
   was introduced as "Sequence Number Extension" in the TCP
   Authentication Option (TCP-AO) [RFC5925].  The example provided there
   was intended to introduce the concept, but the pseudocode provided is
   not complete.

   This document presents the formal requirements for Sequence Number
   Extension (SNE), a code example, and a check sequence that can be
   used to validate this and alternate implementations.  Sequence
   numbers are used in a variety of protocols to support loss detection,
   reordering, flow control, and congestion control.  Limitations in the
   size of a sequence number protocol field can limit the ways in which
   these capabilities can be supported.

   Under certain conditions, it is possible for both endpoints of a
   protocol to keep track of sequence number rollover and effectively
   extend the sequence number space without requiring modification of
   the sequence number field used within protocol messages.  These
   conditions assume that the received sequence numbers never vary by
   more than half the size of the space of the field used in messages,
   i.e., they never hop forward or backward by more than half that
   space.  This constraint is typical in sliding window protocols, such
   as TCP.  However, although both ends can track rollover
   unambiguously, doing so can be surprisingly complex.  This document
   provides examples and test cases to simplify that process.

   This document is intended for protocol designers who seek to use
   larger sequence numbers at the endpoints without needing to extend
   the sequence number field used in messages, such as for
   authentication protocols, e.g., TCP-AO [RFC5925].  Use of extended
   sequence numbers should be part of a protocol specification so that
   both endpoints can ensure they comply with the requirements needed to
   enable their use in both locations.

   The remainder of this document describes how SNE can be supported and
   provides the pseudocode to demonstrate how received messages can
   unambiguously determine the appropriate extension value, as long as
   the reordering is constrained.  Section 2 provides background on the
   concept.  Section 3 discusses currently known uses of SNE.  Section 4
   discusses how SNE is used in protocol design and how it differs from
   in-band use of sequence numbers.  Section 5 provides a framework for
   testing SNE implementations, including example code for the SNE
   function, and Section 6 provides a sequence that can be used by that
   code for validation.  Section 7 concludes with a discussion of
   security issues.

2.  Background



   Protocols use sequence numbers to maintain message order.  The
   transmitter typically increments them either once per message or by
   the length of the message.  The receiver uses them to reorder
   messages and detect gaps due to inferred loss.

   Sequence numbers are represented within those messages (e.g., in the
   headers) as values of a finite, unsigned number space.  This space is
   typically represented in a fixed-length bit string, whose values
   range from 0..(2^N)-1, inclusive.

   The use of finite representations has repercussions on the use of
   these values at both the transmitter and receiver.  Without
   additional constraints, when the number space "wraps around", it
   would be impossible for the receiver to distinguish between the uses
   of the same value.

   As a consequence, additional constraints are required.  Transmitters
   are typically required to limit reuse until they can assume that
   receivers would successfully differentiate the two uses of the same
   value.  The receiver always interprets values it sees based on the
   assumption that successive values never differ by just under half the
   number space.  A receiver cannot detect an error in that sequence,
   but it will incorrectly interpret numbers if reordering violates this
   constraint.

   The constraint requires that "forward" values advance the values by
   less than half the sequence number space, ensuring that receivers
   never experience a series of values that violate that rule.

   We define a sequence space as follows:

   *  An unsigned integer within the range of 0..(2^N)-1, i.e., for N
      bits.

   *  An operation that increments values in that space by K, where K <
      2^(N-1), i.e., less than half the range.  This operation is used
      exclusively by the transmitter.

   *  An operation that compares two values in that space to determine
      their order, e.g., where X < Y implies that X comes before Y.

   We assume that both sides begin with the same initial value, which
   can be anywhere in the space.  That value is either assumed (e.g., 0)
   before the protocol begins or coordinated before other messages are
   exchanged (as with TCP Initial Sequence Numbers (ISNs) [RFC0793]).
   It is assumed that the receiver always receives values that are
   always within (2^N)-1, but the successive received values never jump
   forward or backward by more than 2^(N-1)-1, i.e., just under half the
   range.

   No other operations are supported.  The transmitter is not permitted
   to "backup", such that values are always used in "increment" order.
   The receiver cannot experience loss or gaps larger than 2^(N-1)-1,
   which is typically enforced either by assumption or by explicit
   endpoint coordination.

   An SNE is a separate number space that can be combined with the
   sequence number to create a larger number space that need not wrap
   around during a connection.

   On the transmit side, SNE is trivially accomplished by incrementing a
   local counter once each time the sequence number increment "wraps"
   around or by keeping a larger local sequence number whose least-
   significant part is the message sequence number and most-significant
   part can be considered the SNE.  The transmitter typically does not
   need to maintain an SNE except when used in local computations, such
   as for Message Authentication Codes (MACs) in TCP-AO [RFC5925].

   The goal of this document is to demonstrate that SNE can be
   accomplished on the receiver side without transmitting additional
   information in messages.  It defines the stateful function
   compute_sne() as follows:

         SNE = compute_sne(seqno)

   The compute_sne() function accepts the sequence number seen in a
   received message and computes the corresponding SNE.  The function
   includes persistent local state that tracks the largest currently
   received SNE and seqno combination.  The concatenation of SNE and
   seqno emulates the equivalent larger sequence number space that can
   avoid wrap around.

   Note that the function defined here is capable of receiving any
   series of seqno values and computing their correct corresponding SNE,
   as long as the series never "jumps" more than half the number space
   "backward" from the largest value seen "forward".

3.  Related Discussion



   The DNS uses sequence numbers to determine when a Start of Authority
   (SOA) serial number is more recent than a previous one, even
   considering sequence space wrap [RFC1034][RFC1035].  The use of
   wrapped sequence numbers for sliding windows in network protocols was
   first described as a sequence number space [IEN74].

   A more recent discussion describes this as "serial number arithmetic"
   and defines a comparison operator it claimed was missing in IEN-74
   [RFC1982].  That document defines two operations: addition
   (presumably shifting the window forward) and comparison (defining the
   order of two values).  Addition is defined in that document as
   limited to values within the range of 0..windowsize/2-1.  Comparison
   is defined in that document by a set of equations therein, but that
   document does not provide a way for a receiver to compute the correct
   equivalent SNE, especially including the potential for sequence
   number reordering, as is demonstrated in this document.

4.  Using SNE in Protocol Design



   As noted in the introduction, message sequence numbers enable
   reordering, loss detection, flow control, and congestion control.
   They are also used to differentiate otherwise potentially identical
   messages that might repeat as part of a sequence or stream.

   The size of the sequence number field used within transferred
   messages defines the ability of a protocol to tolerate reordering and
   gaps, notably limited to half the space of that field.  For example,
   a field of 8 bits can reorder and detect losses of smaller than 2^7,
   i.e., 127 messages.  When used for these purposes -- reordering, loss
   detection, flow control, and congestion control -- the size of the
   field defines the limits of those capabilities.

   Sequence numbers are also used to differentiate messages; when used
   this way, they can be problematic if they repeat for otherwise
   identical messages.  Protocols using sequence numbers tolerate that
   repetition because they are aware of the rollover of these sequence
   number spaces at both endpoints.  In some cases, it can be useful to
   track this rollover and use the rollover count as an extension to the
   sequence number, e.g., to differentiate authentication MACs.  This
   SNE is never transmitted in messages; the existing rules of sequence
   numbers ensure both ends can keep track unambiguously -- both for new
   messages and reordered messages.

   The constraints required to use SNE have already been presented as
   background in Section 2.  The transmitter must never send messages
   out of sequence beyond half the range of the sequence number field
   used in messages.  A receiver uses this assumption to interpret
   whether received numbers are part of pre-wrap sequences or post-wrap
   sequences.  Note that a receiver cannot enforce or detect if the
   transmitter has violated these assumptions on its own; it relies on
   explicit coordination to ensure this property is maintained, such as
   the exchange of acknowledgements.

   SNEs are intended for use when it is helpful for both ends to
   unambiguously determine whether the sequence number in a message has
   wrapped and whether a received message is pre-wrap or post-wrap for
   each such wrap.  This can be used by both endpoints to ensure all
   messages of arbitrarily long sequences can be differentiated, e.g.,
   ensuring unique MACs.

   SNE does not extend the actual sequence space of a protocol or (thus)
   its tolerance to reordering or gaps.  It also cannot improve its
   dynamic range for flow control or congestion control, although there
   are other somewhat related methods that can, such as window scaling
   [RFC7323] (which increases range at the expense of granularity).

   SNE is not needed if messages are already unique over the entirety of
   a transfer sequence, e.g., either because the sequence number field
   used in its messages never wrap around or because other fields
   provide that disambiguation, such as timestamps.

5.  Example Code



   The following C code is provided as a verified example of SNE from 16
   to 32 bits.  The code includes both the framework used for validation
   and the compute_sne() function, the latter of which can be used
   operationally.

   A correct test will indicate "OK" for each test.  An incorrect test
   will indicate "ERROR" where applicable.

   <CODE BEGINS> file "compute_sne.c"
   #include <stdio.h>
   #include <sys/param.h>

   #define distance(x,y)   (((x)<(y))?((y)-(x)):((x)-(y)))

   #define SNEDEBUG 1

   // This is the core code, stand-alone, to compute SNE from seqno
   // >> replace this function with your own code to test alternates
   unsigned long compute_sne(unsigned long seqno) {
       // INPUT: 32-bit unsigned sequence number (low bits)
       // OUTPUT: 32-bit unsigned SNE (high bits)

       // variables used in this code example to compute SNE:

       static unsigned long
         RCV_SNE = 0;        // high-watermark SNE
       static int
         RCV_SNE_FLAG = 1;   // set during first half rollover
                             // (prevents re-rollover)
       static unsigned long
         RCV_PREV_SEQ = 0;   // high-watermark SEQ
       unsigned long
         holdSNE;            // temp copy of output

       holdSNE = RCV_SNE;                // use current SNE to start
       if (distance(seqno,RCV_PREV_SEQ) < 0x80000000) {
           // both in same SNE range?
           if ((seqno >= 0x80000000) && (RCV_PREV_SEQ < 0x80000000)) {
               // jumps fwd over N/2?
               RCV_SNE_FLAG = 0;         // reset wrap increment flag
           }
           RCV_PREV_SEQ = MAX(seqno,RCV_PREV_SEQ);
                                         // move prev forward if needed
       } else {
               // both in diff SNE ranges
               if (seqno < 0x80000000) {
                   // jumps forward over zero?
                   RCV_PREV_SEQ = seqno; // update prev
                   if (RCV_SNE_FLAG == 0) {
                       // first jump over zero? (wrap)
                       RCV_SNE_FLAG = 1;
                                 // set flag so we increment once
                       RCV_SNE = RCV_SNE + 1;
                                 // increment window
                       holdSNE = RCV_SNE;
                                 // use updated SNE value
                   }
               } else {
                   // jump backward over zero
                   holdSNE = RCV_SNE - 1;
                                 // use pre-rollover SNE value
               }
       }
       #ifdef SNEDEBUG
       fprintf(stderr,"state RCV_SNE_FLAG =        %1d\n",
         RCV_SNE_FLAG);
       fprintf(stderr,"state      RCV_SNE = %08lx\n", RCV_SNE);
       fprintf(stderr,"state RCV_PREV_SEQ = %08lx\n", RCV_PREV_SEQ);
       #endif
       return holdSNE;
   }

   int main() {
       // variables used as input and output:
       unsigned long SEG_SEQ;        // input - received SEQ
       unsigned long SNE;            // output - SNE corresponding
                                     // to received SEQ

       // variables used to validate the computed SNE:
       unsigned long SEG_HIGH;       // input - xmitter side SNE
                                     // -> SNE should match this value
       unsigned long long BIG_PREV;  // prev 64-bit total seqno
       unsigned long long BIG_THIS = 0;  // current 64-bit total seqno
                                     // -> THIS, PREV should never jump
                                     //    more than half the SEQ space

      char *prompt = "Input hex numbers only (0x is optional)\n\n")
                     "\tHex input\n"
                     "\t(2 hex numbers separated by whitespace,"
                     "each with 8 or fewer digits)";

       fprintf(stderr,"%s\n",prompt);

       while (scanf("%lx %lx",&SEG_HIGH,&SEG_SEQ) == 2) {
           BIG_PREV = BIG_THIS;
           BIG_THIS = (((unsigned long long)SEG_HIGH) << 32)
                     | ((unsigned long long)SEG_SEQ);

           // given SEG_SEQ, compute SNE
           SNE = compute_sne(SEG_SEQ);

           fprintf(stderr,"           SEG_SEQ = %08lx\n", SEG_SEQ);
           fprintf(stderr,"               SNE = %08lx\n", SNE);
           fprintf(stderr,"          SEG_HIGH = %08lx %s\n",SEG_HIGH,
                   (SEG_HIGH == SNE)? " - OK" : " - ERROR !!!!!!!");
           fprintf(stderr,"\t\tthe jump was %16llx %s %s\n",
                   distance(BIG_PREV,BIG_THIS),
                   ((BIG_PREV < BIG_THIS)?"+":"-"),
                   (((distance(BIG_PREV,BIG_THIS)) > 0x7FFFFFFF)
                    ? "ILLEGAL JUMP" : "."));
           fprintf(stderr,"\n");
           fprintf(stderr,"\n");

           fprintf(stderr,"%s\n",prompt);

       }
   }
   <CODE ENDS>

6.  Validation Suite



   The following numbers are used to validate SNE variants and are shown
   in the order they legitimately could be received.  Each line
   represents a single 64-bit number, represented as two hexadecimal
   32-bit numbers with a space between.  The numbers are formatted for
   use in the example code provided in Section 5.

   A correctly operating extended sequence number system can receive the
   least-significant half (the right side) and compute the correct most-
   significant half (the left side) correctly.  It specifically tests
   both forward and backward jumps in received values that represent
   legitimate reordering.

   00000000 00000000
   00000000 30000000
   00000000 90000000
   00000000 70000000
   00000000 a0000000
   00000001 00000001
   00000000 e0000000
   00000001 00000000
   00000001 7fffffff
   00000001 00000000
   00000001 50000000
   00000001 80000000
   00000001 00000001
   00000001 40000000
   00000001 90000000
   00000001 b0000000
   00000002 0fffffff
   00000002 20000000
   00000002 90000000
   00000002 70000000
   00000002 A0000000
   00000003 00004000
   00000002 D0000000
   00000003 20000000
   00000003 90000000
   00000003 70000000
   00000003 A0000000
   00000004 00004000
   00000003 D0000000

7.  Security Considerations



   Sequence numbers and their extensions can be useful in a variety of
   security contexts.  Because the extension part (most-significant
   half) is determined by the previously exchanged sequence values
   (least-significant half), the extension should not be considered as
   adding entropy for the purposes of message authentication or
   encryption.

8.  IANA Considerations



   This document has no IANA actions.

9.  Informative References



   [IEN74]    Plummmer, W., "Sequence Number Arithmetic", IEN-74,
              September 1978.

   [RFC0793]  Postel, J., "Transmission Control Protocol", STD 7,
              RFC 793, DOI 10.17487/RFC0793, September 1981,
              <https://www.rfc-editor.org/info/rfc793>.

   [RFC1034]  Mockapetris, P., "Domain names - concepts and facilities",
              STD 13, RFC 1034, DOI 10.17487/RFC1034, November 1987,
              <https://www.rfc-editor.org/info/rfc1034>.

   [RFC1035]  Mockapetris, P., "Domain names - implementation and
              specification", STD 13, RFC 1035, DOI 10.17487/RFC1035,
              November 1987, <https://www.rfc-editor.org/info/rfc1035>.

   [RFC1982]  Elz, R. and R. Bush, "Serial Number Arithmetic", RFC 1982,
              DOI 10.17487/RFC1982, August 1996,
              <https://www.rfc-editor.org/info/rfc1982>.

   [RFC5925]  Touch, J., Mankin, A., and R. Bonica, "The TCP
              Authentication Option", RFC 5925, DOI 10.17487/RFC5925,
              June 2010, <https://www.rfc-editor.org/info/rfc5925>.

   [RFC7323]  Borman, D., Braden, B., Jacobson, V., and R.
              Scheffenegger, Ed., "TCP Extensions for High Performance",
              RFC 7323, DOI 10.17487/RFC7323, September 2014,
              <https://www.rfc-editor.org/info/rfc7323>.

Acknowledgments

   The need for this document was first noted by Juhamatti Kuusisaari in
   April 2020 during discussions of the pseudocode in RFC 5925.

Author's Address



   Joe Touch
   Manhattan Beach, CA 90266
   United States of America

   Phone: +1 (310) 560-0334
   Email: touch@strayalpha.com