<?xml version='1.0' encoding='utf-8'?>
<!DOCTYPE rfc [
  <!ENTITY nbsp    "&#160;">
  <!ENTITY zwsp   "&#8203;">
  <!ENTITY nbhy   "&#8209;">
  <!ENTITY wj     "&#8288;">
]>
<?xml-stylesheet type="text/xsl" href="rfc2629.xslt" ?>
<!-- generated by https://github.com/cabo/kramdown-rfc version 1.7.43 (Ruby 3.2.3) -->
<rfc xmlns:xi="http://www.w3.org/2001/XInclude" ipr="trust200902" docName="draft-bonica-macneil-singh-ipv6-firewall-metadata-01" category="exp" submissionType="IETF" tocInclude="true" sortRefs="true" symRefs="true" version="3">
  <!-- xml2rfc v2v3 conversion 3.35.0 -->
  <front>
    <title abbrev="IPv6 Firewall Metadata">Verifiable Firewall and Security-Enforcement Metadata in IPv6</title>
    <seriesInfo name="Internet-Draft" value="draft-bonica-macneil-singh-ipv6-firewall-metadata-01"/>
    <author initials="R." surname="Bonica" fullname="Ron Bonica">
      <organization>Hewlett Packard Enterprise</organization>
      <address>
        <email>ronald.bonica@hpe.com</email>
      </address>
    </author>
    <author initials="E." surname="MacNeil" fullname="Erin MacNeil">
      <organization>Hewlett Packard Enterprise</organization>
      <address>
        <email>erin.macneil@hpe.com</email>
      </address>
    </author>
    <author initials="P." surname="Singh" fullname="Prashant Singh">
      <organization>Hewlett Packard Enterprise</organization>
      <address>
        <email>prashant.singh@hpe.com</email>
      </address>
    </author>
    <date year="2026" month="November" day="14"/>
    <area>Internet</area>
    <workgroup>IPv6 Maintenance Working Group</workgroup>
    <keyword>IPv6</keyword>
    <keyword>firewall</keyword>
    <keyword>metadata</keyword>
    <keyword>integrity</keyword>
    <keyword>ICV</keyword>
    <keyword>IOAM</keyword>
    <abstract>
      <?line 53?>

<t>This document defines a way to carry a verifiable record of security enforcement
inside IPv6 packets.  Each firewall (or other security enforcement point) that
handles a packet writes a small, authenticated note saying which rule it
matched and what action it took.  The notes are protected by a chained
Integrity Check Value (ICV), reusing the construction defined for In-situ OAM
(IOAM) data integrity.
A checker can then prove two separate things: that the notes are authentic (not
forged or changed on the way), and that the recorded actions match the intended
policy.  Because the two checks are independent, a firewall that honestly
diverges from policy -- for example because of a stale rule or an implementation
bug -- still produces an authentic note, and the divergence is reported rather
than hidden.  Operators thus gain precise per-hop evidence that every enforcement
point was traversed and applied the intended policy, so an EP that was skipped or
did not enforce is revealed.</t>
    </abstract>
  </front>
  <middle>
    <?line 70?>

<section anchor="introduction">
      <name>Introduction</name>
      <t>Operators want proof of how a packet was treated as it crossed a chain of
firewalls and other security enforcement points (EPs).  They want to answer
simple questions:</t>
      <ul spacing="normal">
        <li>
          <t>Did every EP apply the policy it was supposed to?</t>
        </li>
        <li>
          <t>Was every EP on the intended path actually traversed, or was one skipped or
bypassed?</t>
        </li>
        <li>
          <t>Did any EP accept, drop, or rate-limit the packet against policy?</t>
        </li>
        <li>
          <t>Was a recorded decision really made by the EP that claims it, or was it
forged or changed by someone on the path?</t>
        </li>
      </ul>
      <t>This document lets each EP add a small note to the packet.  The note says which
rule matched and what action the EP took.  Each EP also folds its note into a
running Integrity Check Value (ICV) using its own secret key.  The ICV is the
same chained construction used for IOAM data integrity
<xref target="I-D.ietf-ippm-ioam-data-integrity"/>; this document does not invent new
cryptography.</t>
      <t>The design separates two checks that are easy to mix up:</t>
      <ol spacing="normal" type="1"><li>
          <t>Is the note authentic? (Integrity.) The checker recomputes the ICV chain; if
it matches, the notes were written by the claimed EPs and were not changed on
the path.</t>
        </li>
        <li>
          <t>Does the action match policy? (Conformance.) The checker compares each
authentic note to the policy that was expected.</t>
        </li>
      </ol>
      <t>The two checks are independent: a note can be authentic yet still diverge from
policy.  An EP may honestly apply the wrong action -- a stale rule, a rollout
still in flight, or the same policy behaving differently across vendors, software
versions, or hardware -- and because the note stays authentic, that silent
divergence is reported instead of hidden.</t>
      <t>Current practice can steer a packet through a set of firewalls, and
proof-of-transit schemes can show it traversed them, but neither proves what each
firewall actually enforced -- the rule that matched and the action taken.  This
mechanism supplies that per-hop enforcement evidence, and from it a firewall that
was skipped or did not enforce is detected.</t>
      <t>This mechanism is intended for use within a single limited domain (<xref target="RFC8799"/>).</t>
    </section>
    <section anchor="conventions-and-definitions">
      <name>Conventions and Definitions</name>
      <t>The key words "<bcp14>MUST</bcp14>", "<bcp14>MUST NOT</bcp14>", "<bcp14>REQUIRED</bcp14>", "<bcp14>SHALL</bcp14>", "<bcp14>SHALL
NOT</bcp14>", "<bcp14>SHOULD</bcp14>", "<bcp14>SHOULD NOT</bcp14>", "<bcp14>RECOMMENDED</bcp14>", "<bcp14>NOT RECOMMENDED</bcp14>",
"<bcp14>MAY</bcp14>", and "<bcp14>OPTIONAL</bcp14>" in this document are to be interpreted as
described in BCP 14 <xref target="RFC2119"/> <xref target="RFC8174"/> when, and only when, they
appear in all capitals, as shown here.</t>
      <?line -18?>

<dl>
        <dt>Enforcement Point (EP):</dt>
        <dd>
          <t>A firewall or other on-path device that evaluates policy against a packet and
records the result.</t>
        </dd>
        <dt>Policy ID:</dt>
        <dd>
          <t>An identifier for the policy (or access list) every EP uses.</t>
        </dd>
        <dt>Policy Version:</dt>
        <dd>
          <t>The generation of that policy.  It changes each time the rule set is edited.</t>
        </dd>
        <dt>Filter ID:</dt>
        <dd>
          <t>An identifier for the exact rule (also called a rule ID) that matched, within
a given Policy ID and Policy Version. A policy is a set of rules, each with a
match and an action; the Filter ID names the one that fired.</t>
        </dd>
        <dt>ICV:</dt>
        <dd>
          <t>Integrity Check Value, computed with the chained AES-GMAC construction of
<xref target="I-D.ietf-ippm-ioam-data-integrity"/>.</t>
        </dd>
        <dt>Validator:</dt>
        <dd>
          <t>Any entity that checks the notes.  It may be an EP on the path or the final
decapsulating node.</t>
        </dd>
      </dl>
    </section>
    <section anchor="how-it-works">
      <name>How It Works</name>
      <t>A packet carries a small block of firewall metadata.  As the packet crosses each
EP, that EP writes one record (its note) and updates the ICV with its own key.
At the end, a validator checks the ICV and compares each record to policy.</t>
      <artwork><![CDATA[
   packet carrying the firewall-metadata option
       |           |           |                |
       v           v           v                v
   +-------+   +-------+   +-------+        +-----------+
   | EP 1  |   | EP 2  |   | EP 3  |  ...   |  Decap /  |
   | (fw)  |   | (fw)  |   | (fw)  |        | validator |
   +-------+   +-------+   +-------+        +-----------+
   write rec   write rec   write rec       check ICV
   + update    + update    + update        + compare to
     ICV         ICV         ICV             policy
]]></artwork>
      <t>Each EP needs only its own key to write its record.  Only a validator needs the
keys of the EPs whose records it checks.  The diagram shows validation at the
end for simplicity, but validation is a role, not a fixed place: a validator can
be any EP that holds the upstream keys (early, in-path checking) or just the
final node (one check at the end).</t>
    </section>
    <section anchor="what-the-identifiers-mean">
      <name>What the Identifiers Mean</name>
      <t>Policy ID, Policy Version, and Filter ID are compact identifiers that stand for
things that already exist in a firewall or access-list configuration.  A
controller assigns the identifiers and keeps the human-readable mapping.  The EP
stamps them; a validator maps them back to named rules.  Only the identifiers
travel in the packet; rule text and match fields never do.</t>
      <t>Firewall rules may be expressed in any syntax; what matters is that they are
semantically identical at every EP, so a given Filter ID maps to the same rule
everywhere on the path.</t>
      <t>An authenticated {Filter ID, Action} pair then proves which rule an EP matched
and what action it took, relative to the named Policy ID and Policy Version.</t>
    </section>
    <section anchor="encoding">
      <name>Encoding</name>
      <t>The metadata is carried in an IPv6 Hop-by-Hop Options header, the extension
header that RFC 8200 allows transit nodes to process (<xref target="RFC8200"/>).
The packet source adds it directly; if the packet originates elsewhere, the
domain ingress encapsulates it in an outer IPv6 header and adds it there, since
RFC 8200 forbids inserting an extension header in transit. Whoever adds it
pre-allocates every record slot; each configured transit EP fills a slot and
updates the ICV, and <bcp14>MUST NOT</bcp14> insert, remove, resize, or grow the header.</t>
      <t>The Option Type has action bits 00 (skip an unrecognized option) and its
change-en-route bit set to 1, giving the bit pattern 001xxxxx (value TBD).
The change-en-route bit marks IPv6-processing mutability of the Option Data only;
ICV coverage of the immutable fields and records is independent and defined in
<xref target="mask"/>.</t>
      <t><xref target="RFC9673"/> aims to make IPv6 Hop-by-Hop options useful to deploy and use.</t>
      <t>This document specifies carriage (c). Models (a) SRv6 steering (<xref target="RFC8754"/>,
<xref target="RFC8986"/>) and (b) encapsulation are destination-based alternatives left open
for working-group discussion:</t>
      <artwork><![CDATA[
  (a) SRv6 steering -- every EP is a segment endpoint (destination):

        ingress ==> EP1 ==> EP2 ==> EP3 ==> decap / validator
                    |       |       |          |
                    writes  writes  writes     checks ICV + policy
                    Rec1    Rec2    Rec3

  (b) Encapsulation -- only tunnel ends are destinations:

        ingress ========= tunnel ========> egress (EP + validator)
        adds outer IPv6                          one EP's record
        + option

  (c) Hop-by-Hop (this document) -- each EP processes the option in
      path as a transit node, which RFC 8200 permits for the
      Hop-by-Hop Options header:

        ... --> [fw EP1] --> [fw EP2] --> [fw EP3] --> ...
                 writes       writes       writes
]]></artwork>
    </section>
    <section anchor="packet-format">
      <name>Packet Format</name>
      <t>The header and record layouts follow IOAM conventions (<xref target="RFC9197"/>) -- fixed
format fields, 4-octet-friendly alignment, and a namespace identifier -- so the
format is familiar to IOAM implementers; the same field semantics have an
IOAM-native encoding in the companion <xref target="I-D.singh-macneil-ioam-firewall-metadata"/>.
All multi-byte fields are in network
byte order (big-endian).  The values in the figures match the reference run in
<xref target="examples"/>.</t>
      <t>The Hop-by-Hop Options header contains the firewall-metadata option with the
following fields. Record space is allocated before transmission; appending a
record means filling the next slot, not extending the header.</t>
      <artwork><![CDATA[
+----------------------+----------------------------------------+
| IPv6 Hop-by-Hop      | Next Header, Hdr Ext Len,              |
| Options header       | Option Type (TBD), Opt Data Len        |
+----------------------+----------------------------------------+
| Initial Header       | 8 octets  (see {{inithdr}})             |
+----------------------+----------------------------------------+
| Integrity Header     | 32 octets (MethodID, NonceLen, Rsvd,   |
|                      |            Nonce(12), ICV(16))         |
+----------------------+----------------------------------------+
| Per-Node Records     | N x 8 octets (see {{record}})          |
+----------------------+----------------------------------------+
| PadN                 | align option to a multiple of 8 octets |
+----------------------+----------------------------------------+
]]></artwork>
      <section anchor="inithdr">
        <name>Initial Header</name>
        <t>The Initial Header is 8 octets:</t>
        <artwork><![CDATA[
 0                   1                   2                   3
 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|          Namespace ID         |          Policy ID            |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Policy Version|   Capacity    | Flags|NextIdx |   Reserved    |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
]]></artwork>
        <dl>
          <dt>Namespace ID (16 bits):</dt>
          <dd>
            <t>Identifies the security-metadata namespace, mirroring the IOAM Namespace-ID
(<xref target="RFC9197"/>, Section 4.3). A namespace is a context that fixes how the
metadata is interpreted: it disambiguates otherwise non-unique Node IDs and
Policy IDs across overlapping domains, and it selects the record format. It is
also the extensibility mechanism: instead of adding optional fields to one
format, an operator can define a new namespace whose fixed record format
carries additional fields (for example a timestamp), and every EP and
validator keys its parsing off the Namespace ID. For the same reason there is
no separate format-version field. Placed first, as in IOAM. Set by the
ingress; immutable and covered by the ICV. The default value 0 means no
specific namespace.</t>
          </dd>
          <dt>Policy ID (16 bits):</dt>
          <dd>
            <t>Names the policy every EP uses. One value per path, set by the ingress; this
document assumes all EPs on the path evaluate the same policy (see
<xref target="one-policy"/>). Immutable; covered by the ICV.</t>
          </dd>
          <dt>Policy Version (8 bits):</dt>
          <dd>
            <t>Generation (epoch) of that policy, bumped by the control plane on every edit,
set once by the ingress. It binds the trace to a specific rule-set generation
and <bcp14>MUST</bcp14> be a domain-wide, monotonically increasing value per policy. This is
the ingress's single reference generation; if an EP ran a different generation
during a rollout (version skew), the per-record PV flag marks it, so the
mismatch is treated as expected skew rather than a true divergence
(see <xref target="one-policy"/>). Carrying each EP's actual generation is a namespace
extension. Immutable; covered by the ICV.</t>
          </dd>
          <dt>Capacity (8 bits):</dt>
          <dd>
            <t>The number of record slots the ingress pre-allocated for this option, counted
in records (the Namespace ID fixes the record size, 8 octets by default). Set
once by the ingress, sized so the option fits the domain PMTU (<xref target="RFC8200"/>).
Immutable; covered by the ICV. Capacity lets a validator tell a legitimately
full buffer (records present equal to Capacity) from a bypassed or stripped
record (fewer records present than Capacity) using only authenticated data.
Whether Capacity merits its own octet is left open for working-group review:
completeness is already enforced by the authenticated Node IDs against the
expected path (<xref target="checking"/>), so if that authenticated capacity and overflow
signal is judged not worth the space, this octet (byte 5) can be reclaimed as
Reserved.</t>
          </dd>
          <dt>Flags (3 bits):</dt>
          <dd>
            <t>Mutable control bits, excluded from the ICV so EPs may update them. One bit is
the Overflow (O) indicator: an EP with no free slot sets O to mark the metadata
incomplete. This indicator <bcp14>MUST</bcp14> be in the header, not in a per-node record,
because an EP that cannot write a record also cannot set a bit inside it. O is
only a hint; incomplete metadata is detected regardless by the completeness
check (<xref target="checking"/>). The remaining bits are reserved (<bcp14>MUST</bcp14> be zero).</t>
          </dd>
          <dt>NextIndex (5 bits):</dt>
          <dd>
            <t>The index of the next free record slot, equal to the count of records already
written. An EP writes its record at NextIndex and increments it; when NextIndex
equals Capacity there is no free slot and the EP sets O instead of writing. A
write-path cursor only: mutable and excluded from the ICV, so a validator
relies on the authenticated records and Capacity, never on this field.</t>
          </dd>
          <dt>Reserved (8 bits):</dt>
          <dd>
            <t>Set to zero on send, ignored on receipt. Excluded from the ICV. (In the
extensible record variant of <xref target="record-ext"/>, this octet instead carries the
Optional Data Present bitmap, which is then covered by the ICV.)</t>
          </dd>
        </dl>
      </section>
      <section anchor="record">
        <name>Per-Node Record</name>
        <t>Each EP appends one 8-octet record:</t>
        <artwork><![CDATA[
 0                   1                   2                   3
 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|            Node ID            |         Filter ID (20)        |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|(cont) |Action |State|Severity |             Flags             |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

  bytes 0-1 : Node ID      (16 bits)
  bytes 2-4 : Filter ID    (20 bits) || Action (4 bits)
  byte  5   : Policer Data -- State (3 bits) || Severity (5 bits)
  bytes 6-7 : Flags        (16 bits)
]]></artwork>
        <dl>
          <dt>Node ID (16 bits):</dt>
          <dd>
            <t>Identifies the EP that wrote the record. Selects the per-node key.</t>
          </dd>
          <dt>Filter ID (20 bits):</dt>
          <dd>
            <t>The firewall rule that matched, scoped to Policy ID and Policy Version. It is
the compact handle the controller assigns to that rule and maps back; it need
not equal the ACL sequence-id or priority, and its meaning is scoped to one
policy within a single domain -- the same value in another domain is unrelated.</t>
          </dd>
          <dt>Action (4 bits):</dt>
          <dd>
            <t>The action applied: 0 = ACCEPT, 1 = DROP, 2 = POLICE. Other values reserved.</t>
          </dd>
          <dt>Policer State (3 bits):</dt>
          <dd>
            <t>Policer state when Action = POLICE. This document uses 0 = GREEN (in profile, or
conformant -- within the configured committed rate and burst) and 2 = RED (out
of profile, or non-conformant -- exceeding them), the green/red marking of
<xref target="RFC2697"/>; value 1 (YELLOW) and the rest are reserved. The value is the
policer's state as it handled the packet: RED when the measured rate was over
the configured rate, GREEN when within it. It is stamped on a packet the policer
forwards (allows), so the rate condition travels onward with the packet. A RED
record therefore marks an allowed, forwarded packet that the policer handled
while over its configured rate -- not a dropped one (see <xref target="examples"/>, Case 5).
If the policer instead drops the excess traffic, those packets take the DROP
action and carry no metadata downstream -- so they never appear as RED records.</t>
          </dd>
          <dt>Severity (5 bits):</dt>
          <dd>
            <t>A carry-forward count of out-of-profile hops: a GREEN hop adds 0, a RED hop
adds 1. The count travels with the packet, so no EP needs per-flow state, and
the final value is how many hops policed this packet out of profile. Because it
accumulates across hops and arrives authenticated, an egress or collector can
use it as a distributed rate-limit / DDoS signal -- shedding or correlating
packets that were policed at many hops, catching distributed overload that no
single hop saw enough of to act on alone (see <xref target="inline"/>). A future version <bcp14>MAY</bcp14>
add Policer values and per-color weights so Severity becomes a weighted
intensity instead of a plain count.</t>
          </dd>
          <dt>Flags (16 bits):</dt>
          <dd>
            <t>Per-node status, covered by the ICV (so each bit is an authenticated per-node
statement). Two bits signal that the EP's policy context differs from the
Initial Header:
</t>
            <ul spacing="normal">
              <li>
                <t>Bit 0 -- PV (Policy Version Differs): the EP evaluated a different Policy
Version than the header (version skew, typically a transient rollout).</t>
              </li>
              <li>
                <t>Bit 1 -- PI (Policy ID Differs): the EP evaluated a different Policy ID
than the header (policy layering, typically intentional and structural).</t>
              </li>
            </ul>
            <t>When either is set, a validator treats an Action or Filter mismatch at that EP
as an expected context difference to review (see <xref target="one-policy"/>), not a true
execution divergence, so legitimate skew or layering does not raise a false
alarm. Keeping them separate lets a validator handle them differently (for
example, auto-suppress PV skew under a convergence timer while always expecting
PI layering). These bits say only that the context differs, not which Policy
ID or Version the EP used; the actual values are a namespace extension.
Remaining bits reserved (<bcp14>MUST</bcp14> be zero).</t>
          </dd>
        </dl>
        <t>Security-critical status (anything a validator acts on) <bcp14>MUST</bcp14> be an immutable
per-node field so that it is covered by the ICV. Mutable header fields (Flags,
NextIndex, Reserved) are not authenticated and <bcp14>MUST NOT</bcp14> carry security decisions.</t>
      </section>
      <section anchor="record-ext">
        <name>Record Extensibility: Fixed vs. Optional Data Present</name>
        <t>This document specifies the fixed 8-octet record above as the primary format,
and records an extensible variant as a co-equal alternative. The choice is left
open for working-group discussion.</t>
        <dl>
          <dt>Fixed (primary):</dt>
          <dd>
            <t>Every record is exactly 8 octets with fixed field offsets. It is the simplest
to implement (fixed offsets suit hardware), adds no per-packet length
computation, needs no extra covered field, and still supports format evolution
because a different Namespace ID can select a different fixed record format --
the IOAM approach. Its limitation is that an optional field (for example a
timestamp or session identifier) cannot be added to the same namespace in
place; it requires a new namespace/format.</t>
          </dd>
          <dt>Optional Data Present (alternative):</dt>
          <dd>
            <t>The Reserved octet of the Initial Header instead carries an 8-bit Optional Data
Present bitmap, set once by the ingress. Each set bit signals an optional field
appended to every record (for example a 64-bit timestamp or session
identifier), making the record self-describing on one namespace. Because the
bitmap defines the record layout it is security-relevant, so it is immutable
and covered by the ICV. Its costs are variable record length (every EP and
validator computes record size from the bitmap), an added covered field, an
extra IANA bit registry, and more implementation complexity.</t>
          </dd>
        </dl>
        <t>For an initial specification the fixed format is <bcp14>RECOMMENDED</bcp14>; the Optional Data
Present variant is preferable only where optional per-record fields are expected
frequently on a single namespace.</t>
      </section>
      <section anchor="wire">
        <name>Complete Option on the Wire</name>
        <t>The figure below assembles every field into one concrete option. The bytes are
the real values from Case 2 of <xref target="examples"/> (three EPs, all in sync): a
64-octet option carrying the 8-octet Initial Header, the 32-octet Integrity
Header, and three 8-octet records. All values are big-endian; offsets are in
hexadecimal.</t>
        <artwork><![CDATA[
Outer framing -- the option sits in an IPv6 HBH Options header:

 +--------------+--------------+--------------+----------------+
 | Next Header  | Hdr Ext Len  | Opt Type=TBD | Opt Data Len   |
 +--------------+--------------+--------------+----------------+
 |         Option Data (64 octets, detailed below)   | PadN     |
 +---------------------------------------------------+----------+

Option Data -- 64 octets (Case 2: 3 EPs, every EP ACCEPT / 0x1001):

 off  octets                     field
 ---  -----------------------    -----------------------------------
 Initial Header (8):
 00   00 00                      Namespace ID   = 0x0000   (covered)
 02   00 a1                      Policy ID      = 0x00a1   (covered)
 04   01                         Policy Version = 1        (covered)
 05   03                         Capacity       = 3        (covered)
 06   03                         Flags=0, NextIndex=3      (mutable)
 07   00                         Reserved       = 0        (mutable)

 Integrity Header (32):
 08   00                         Method ID      = 0
 09   0c                         Nonce Length   = 12
 0a   00 00                      Reserved
 0c   01 00 00 01 00 00 00 00    Nonce: KeyID=01 Encap=000001
 14   00 00 00 02                       Counter=...0000002
 18   45 cb cc fc 11 8d e8 cd    ICV (16) = carried ICV_3
 20   57 37 44 44 d4 cc 09 bc      45cbccfc118de8cd57374444d4cc09bc

 Per-Node Records (3 x 8):
 28   00 01 01 00 10 00 00 00    EP1: id=0001 flt=0x1001 ACPT GRN s0
 30   00 02 01 00 10 00 00 00    EP2: id=0002 flt=0x1001 ACPT GRN s0
 38   00 03 01 00 10 00 00 00    EP3: id=0003 flt=0x1001 ACPT GRN s0
]]></artwork>
        <t>Each record packs Node ID (2 octets), then Filter ID (20 bits) and Action (4
bits) as three octets <tt>01 00 10</tt> = (0x01001 &lt;&lt; 4) | 0x0, then Policer Data
(<tt>00</tt> = GREEN, Severity 0), then the 16-bit record Flags (<tt>00 00</tt>). The masked
Initial Header (mutable octets 06, 07 zeroed; octet 05 Capacity covered) is folded in only by the first
EP, in ICV_1; each later EP folds its record over the previous ICV, so the value
carried at offset 0x18 is the last link, ICV_3. The construction is defined next.</t>
        <t>The ICV reuses the chained AES-GMAC construction of
<xref target="I-D.ietf-ippm-ioam-data-integrity"/> (Method ID 0). Each EP i has a unique key
Key_i. Let MASK(H) be the Initial Header with its mutable bits cleared
(<xref target="mask"/>), Record_i the record written by EP i, and Nonce the 12-octet nonce
(KeyID(1) || EncapNodeID(3) || Counter(8)). For a path of N EPs:</t>
        <artwork><![CDATA[
  ICV_1 = GMAC(Key_1, Nonce, MASK(H) || Record_1)        (first EP)
  ICV_i = GMAC(Key_i, Nonce, ICV_{i-1} || Record_i)      for i = 2..N
]]></artwork>
        <t>The first EP computes ICV_1 over the masked header and its own record. Each later
EP i folds its record over the ICV it received (ICV_{i-1}) and overwrites the
field, so only the latest value, ICV_N, is carried -- that is the "carried ICV".
The Integrity Header carries Method ID (1), Nonce Length (1), Reserved (2), the
Nonce (12), and ICV_N (16). A validator holds the per-node keys, replays the
chain over the masked header and the carried records to recompute ICV_N, and
compares it to the carried value. These are the encapsulating-, transit-, and
validator-node procedures of <xref target="I-D.ietf-ippm-ioam-data-integrity"/> (Sections 5.3,
5.4, and 5.6; that document indexes the first node as 0).</t>
        <t>The chain, drawn as each EP folding its record into the single ICV field:</t>
        <artwork><![CDATA[
   MASK(H)+Rec1        ICV1+Rec2        ICV2+Rec3
        |                  |                |
        v                  v                v
   GMAC(Key1) --ICV1--> GMAC(Key2) --ICV2--> GMAC(Key3) --> ICV3
                                                           (carried)

   Only ICV3 travels on the wire. A validator holding Key1..KeyN
   replays ICV1 -> ICV2 -> ICV3 over the carried records and checks
   it equals ICV3.
]]></artwork>
      </section>
    </section>
    <section anchor="mask">
      <name>ICV Coverage Mask</name>
      <t>Only immutable fields are covered. Fields that legitimately change en route are
masked out (ANDed to zero) before the ICV is computed, exactly as in Section 5.2
of <xref target="I-D.ietf-ippm-ioam-data-integrity"/>.</t>
      <table>
        <thead>
          <tr>
            <th align="left">Initial Header field</th>
            <th align="center">Covered by ICV</th>
          </tr>
        </thead>
        <tbody>
          <tr>
            <td align="left">Namespace ID</td>
            <td align="center">yes</td>
          </tr>
          <tr>
            <td align="left">Policy ID</td>
            <td align="center">yes</td>
          </tr>
          <tr>
            <td align="left">Policy Version</td>
            <td align="center">yes</td>
          </tr>
          <tr>
            <td align="left">Capacity</td>
            <td align="center">yes</td>
          </tr>
          <tr>
            <td align="left">Flags</td>
            <td align="center">no</td>
          </tr>
          <tr>
            <td align="left">NextIndex</td>
            <td align="center">no</td>
          </tr>
          <tr>
            <td align="left">Reserved</td>
            <td align="center">no</td>
          </tr>
        </tbody>
      </table>
      <t>All per-node record fields are covered by the ICV.</t>
      <t>The AND-mask applied to the 8-octet Initial Header (fixed variant) is:</t>
      <artwork><![CDATA[
  FF FF  FF FF  FF  FF  00  00
  ^^^^^  ^^^^^  ^^  ^^  ^^  ^^
  |      |      |   |   |   Reserved           -> 00   (excluded)
  |      |      |   |   Flags || NextIndex  -> 00   (excluded)
  |      |      |   Capacity                -> FF   (covered)
  |      |      Policy Version             -> FF   (covered)
  |      Policy ID                         -> FFFF (covered)
  Namespace ID                            -> FFFF (covered)
]]></artwork>
      <t>(In the Optional Data Present variant of <xref target="record-ext"/>, the final octet is
covered: the mask becomes FF FF FF FF FF FF 00 FF.)</t>
      <t>Mask check. ICV_1 over the masked header plus the first record does not change
when only mutable bits change, and does change when a covered bit changes:</t>
      <artwork><![CDATA[
  base                ICV_1 = a48fd4ba...
  flip mutable bits   ICV_1 = a48fd4ba...  (SAME -> excluded)
  flip Policy ID      ICV_1 = a2e4c797...  (DIFFERS -> covered)
]]></artwork>
    </section>
    <section anchor="checking">
      <name>Checking the Metadata</name>
      <t>A validator does the following checks.</t>
      <artwork><![CDATA[
   carried records + carried ICV
             |
             v
   Step 1  recompute the chain, compare to the carried ICV
             |
        FAIL | PASS (authentic)
       +-----+-----------------------+
       v                             v
   tampered:                 Step 2  compare each record's
   admin response                    Action/Filter to the
                                     expected policy
                                     |
                                MATCH | MISMATCH
                               +------+-------------------+
                               v                          v
                          conformant              divergent: if PV/PI
                          (done)                  set, an expected
                                                  context difference
                                                  to review; else
                                                  admin response
]]></artwork>
      <t>Step 1 -- Integrity. Recompute the ICV chain over MASK(H) and the carried
records, then compare the final recomputed ICV to the carried ICV. If they are
equal, the records are authentic and unchanged; otherwise the metadata was
tampered with. This is the Validator procedure of
<xref target="I-D.ietf-ippm-ioam-data-integrity"/> (Section 5.6).</t>
      <t>Step 2 -- Conformance. For each real record, compare the Action and Filter ID to
the policy expected for the carried Policy ID and Policy Version. Any mismatch
flags honest divergence at that EP.</t>
      <t>Step 3 -- Completeness. Confirm that a record is present for every EP the
validator expected on the path. Because a chain built over a subset of EPs is
itself a valid chain, Step 1 cannot by itself reveal an EP that never stamped a
record. The validator compares the authenticated Node IDs against the expected
path, and their count against the immutable Capacity: a count equal to Capacity
is a legitimately full buffer, whereas a count below Capacity with an expected EP
absent is a bypassed or stripped record. Capacity is covered by the ICV while the
mutable NextIndex and the Overflow hint are not, so this uses only authenticated
data. A missing record is a positive detection: that EP was bypassed or failed to
enforce -- a
security event in its own right (for example, traffic that skipped a firewall),
not merely a gap. Removal of the entire option is outside this check and is left
to the encapsulating protocol (see Security Considerations).</t>
      <t>A record may pass Step 1 (authentic) and fail Step 2 (divergent).</t>
      <section anchor="one-policy">
        <name>One Policy and Generation per Path</name>
        <t>The Initial Header carries a single Policy ID and Policy Version, so this
specification assumes every EP on the path evaluates the same policy and
generation. Conformance (Step 2) compares each record's Action and Filter ID to
that one policy.</t>
        <t>Two situations break that assumption. EPs may legitimately apply different
policies (policy layering -- for example a perimeter firewall policy at the
domain edge and an internal segmentation policy deeper in, or a tenant policy at
one hop and a corporate policy at another), and during a rollout EPs may briefly
run different generations of the same policy (version skew). A single header
value cannot express either case.</t>
        <t>Each record therefore carries two authenticated hint flags (<xref target="record"/>): PV
(Policy Version Differs) and PI (Policy ID Differs). An EP sets PV when it ran a
different generation than the header (rollout skew) and PI when it ran a
different policy (layering). They let a validator treat that EP's Action or
Filter mismatch as an expected context difference to review, rather than a true
divergence, so an operator is not alarmed by legitimate skew or layering; keeping
them separate allows different handling (for example, auto-suppressing PV skew
under a convergence timer). The flags say only that the context differs, not
which Policy ID or generation the EP used; carrying those actual values costs
scarce record bits and is therefore a namespace extension (a record format,
selected by Namespace ID, that adds per-EP Policy ID and Policy Version). The
header Policy Version remains the ingress's reference generation.</t>
      </section>
      <section anchor="validation-is-a-role">
        <name>Validation Is a Role</name>
        <t>Validation is a role, not a fixed node. Any EP that holds the keys of the
upstream EPs 0..i can recompute the ICV prefix and check the records written so
far. The choice is a key-distribution policy:</t>
        <ul spacing="normal">
          <li>
            <t>End-only validation: only the decap holds all keys. Simplest distribution; a
single validation at the end that still runs all three checks -- integrity
(the chain), conformance (per-record action and filter), and completeness
(expected Node IDs).</t>
          </li>
          <li>
            <t>Distributed validation: capable EPs hold the upstream keys they need and check
inline. Earlier detection, at the cost of a wider key footprint.</t>
          </li>
        </ul>
        <t>An EP can play both roles in a single pass: it writes its own record and, if it
holds the upstream keys, validates the records before it (an inline-validating
EP). This is the node that inline Severity reactions (<xref target="inline"/>) and Case 4 rely
on.</t>
        <t>Whatever the key distribution, the decapsulating or egress node <bcp14>MUST</bcp14> validate the
metadata of packets that reach it; inline validation at earlier EPs is optional.
A packet dropped before egress is validated from the dropping EP's export
instead.</t>
      </section>
      <section anchor="response-to-a-detected-problem">
        <name>Response to a Detected Problem</name>
        <t>The protocol produces a trustworthy verdict; the response is set by the operator
and is local to the detecting entity. Typical actions, which <bcp14>MAY</bcp14> be combined:</t>
        <ul spacing="normal">
          <li>
            <t>log or raise an alarm;</t>
          </li>
          <li>
            <t>report the event to the management plane;</t>
          </li>
          <li>
            <t>collect additional diagnostics (for example the EP's firewall state or policer
counters) and attach them to the report;</t>
          </li>
          <li>
            <t>mirror or redirect the packet and its metadata to a collector for deeper
analysis;</t>
          </li>
          <li>
            <t>drop the packet, if so configured.</t>
          </li>
        </ul>
        <t>This applies equally to an integrity failure and a conformance mismatch: the
mechanism produces the verdict, and which action follows is administratively
configured. There is no on-wire "verdict" field.</t>
        <t>An EP always enforces its own policy independently; validating the metadata is an
additional capability, never a substitute for an EP's own inspection, and an EP's
own decision <bcp14>MUST NOT</bcp14> be delayed pending validation. When an EP that expects
metadata cannot verify it (missing, stripped, or failing the integrity check),
local policy selects the behaviour:</t>
        <ul spacing="normal">
          <li>
            <t>Verify-or-Drop (strict): unverifiable or missing expected metadata is treated
as absent, raising an alarm and, optionally, dropping the packet. Suited to
high-assurance zones.</t>
          </li>
          <li>
            <t>Permissive: the EP logs the event, enforces its own policy, and forwards.
Suited to migration, key rotation, or mixed domains.</t>
          </li>
          <li>
            <t>Non-participant: an EP never provisioned with the relevant keys simply forwards
the packet with the metadata intact.</t>
          </li>
        </ul>
        <t>When the response is DROP (or an EP's own policy drops the packet), the packet
does not continue, so the in-band note cannot be delivered downstream. The
dropping EP finalizes the partial chain and <bcp14>MAY</bcp14> export it out-of-band,
rate-limited, so the evidence is preserved for a key-holding collector to
validate; the dropping EP
need not -- and, as a non-validator, may be unable to -- validate the upstream
records itself. Responses that let the packet continue (ACCEPT, POLICE) keep the
note in-band to the decap.</t>
      </section>
    </section>
    <section anchor="examples">
      <name>Worked Examples</name>
      <t>These examples come from a runnable reference. Policy ID is 0x00a1, Policy
Version is 1, and the expected decision at every EP is ACCEPT / Filter ID
0x1001. ICVs are 16 octets; the "..." form shows the first bytes.</t>
      <section anchor="case-1-tampering-is-caught">
        <name>Case 1: Tampering Is Caught</name>
        <t>The packet holds a single ICV field -- the "carried ICV". The chain keeps only
the latest value: EP1 computes it first, and each later EP folds its own record
in and overwrites it, so the carried ICV is the value written by the last EP
(EP3). Three EPs each ACCEPT with Filter ID 0x1001.</t>
        <artwork><![CDATA[
  BUILD -- Rec_i = EP i's record, each folded into one ICV field:
  EP1: ICV_1 = GMAC(K1, Nonce, MASK(header) || Rec_1) -> b23daca9...
  EP2: ICV_2 = GMAC(K2, Nonce, ICV_1 || Rec_2) -> 04eb105e...
  EP3: ICV_3 = GMAC(K3, Nonce, ICV_2 || Rec_3) -> a01d8d8f...
]]></artwork>
        <t>An attacker flips EP2's recorded action ACCEPT -&gt; DROP but cannot update the
carried ICV, because it lacks EP2's key.</t>
        <artwork><![CDATA[
  CHECK (validator replays the chain over the carried records):
  ICV_1 = GMAC(K1, Nonce, MASK(header) || Rec_1) -> b23daca9...
  ICV_2 = GMAC(K2, Nonce, ICV_1 || Rec_2*) -> 31e121c7... (altered)
  ICV_3 = GMAC(K3, Nonce, ICV_2 || Rec_3) -> c3355628...
  recomp ICV_3 = c3355628 != carried = a01d8d8f -> TAMPERING CAUGHT
]]></artwork>
        <t>Integrity detects that the chain was altered, but a single carried ICV does not
by itself say where. Two things localize it: Step 2 conformance independently
flags EP2's record (DROP != expected ACCEPT); and a validator that knows the
expected records can replay a clean chain and compare step by step, where the
first mismatch is EP2. An ICV failure is the validator's verdict, not a value
any node writes -- there is no per-record "ICV failed" bit -- and the action on
failure (alarm, drop, collect, export) is administratively configured (see
Response to a Detected Problem).</t>
      </section>
      <section anchor="case-2-everything-in-sync">
        <name>Case 2: Everything in Sync</name>
        <t>Every EP applies the expected ACCEPT / 0x1001.  Each EP stamps one record and
folds it into the ICV; here every record is {Node ID, Filter 0x1001, Action
ACCEPT, Policer GREEN, Severity 0}.</t>
        <artwork><![CDATA[
  Initial Header: Namespace=0000 PolicyID=00a1 PolicyVer=1
  record EP1: ID=0001 Filter=0x1001 Action=ACCEPT Policer=GREEN Sev=0
  record EP2: ID=0002 Filter=0x1001 Action=ACCEPT Policer=GREEN Sev=0
  record EP3: ID=0003 Filter=0x1001 Action=ACCEPT Policer=GREEN Sev=0
  carried ICV = 45cbccfc...
  Step 1 Integrity   -> PASS (authentic, untampered)
  Step 2 Conformance -> MATCH (every EP applied ACCEPT/0x1001)
]]></artwork>
      </section>
      <section anchor="case-3-action-mismatch-drop-ends-the-path">
        <name>Case 3: Action Mismatch -&gt; DROP Ends the Path</name>
        <t>EP 2 honestly records DROP with the expected Filter ID 0x1001 -- a pure action
mismatch. A DROP stops the packet, so there is no EP 3 and the in-band chain
never reaches the egress.</t>
        <artwork><![CDATA[
  build:  EP1 {Filter 0x1001, ACCEPT} -> EP2 {Filter 0x1001, DROP}
          carried ICV (2 records) = dbbc39ae...
  EP2 drops per its own policy and MAY export the 2-record chain
  out-of-band (rate-limited) to a collector that holds the keys; EP2
  itself need not validate it.
  Collector:
    Step 1 Integrity   -> PASS (authentic)
    Step 2 Conformance -> MISMATCH at EP2: Action DROP vs exp ACCEPT
      (Filter 0x1001 matches; only the action diverges)
]]></artwork>
        <t>A plain dropping EP cannot run Step 1 -- it lacks the upstream keys -- so it
forwards the evidence to a key-holding collector. An inline-validating EP could
run both checks itself before exporting.</t>
      </section>
      <section anchor="case-4-filter-mismatch-caught-inline">
        <name>Case 4: Filter Mismatch -&gt; Caught Inline</name>
        <t>EP 2 takes the right action (ACCEPT) but matched a different rule, Filter ID
0x1005. The action is ACCEPT, so the packet continues. EP 3 acts as an inline
validator.</t>
        <artwork><![CDATA[
  build:  EP1 {Filter 0x1001, ACCEPT} -> EP2 {Filter 0x1005, ACCEPT}
          -> EP3 {Filter 0x1001, ACCEPT}    carried ICV = a018ba1e...
  EP3 inline validator (holds keys for EP1, EP2):
     recompute after EP1 = 7ac31d52...
     recompute after EP2 = 5daf39da...  (so-far matches -> PASS)
     conformance -> MISMATCH at EP2: Filter 0x1005 vs expected 0x1001
       (Action ACCEPT matches; only the rule that fired diverges)
     configured response = export to collector; keep forwarding
  Decap re-validates the full 3-record chain independently.
]]></artwork>
        <t>The outcome (ACCEPT) matched policy, but a different rule fired. This signature
-- right action, wrong rule -- is exactly what an allow/deny result alone would
hide. It typically indicates one of: a stale or drifted rule set on that EP,
Policy Version skew during a rollout, or an implementation bug that matched the
wrong rule. The PV/PI record flags (<xref target="record"/>) let an EP distinguish an
expected context difference (version skew or policy layering) from a true bug:
when neither flag is set, a Filter mismatch is an unexpected divergence to
investigate. Reporting it exposes drift or a bug that would otherwise go
unnoticed.</t>
      </section>
      <section anchor="case-5-policer-color-and-severity">
        <name>Case 5: Policer Color and Severity</name>
        <t>Action is POLICE; the Policer State is GREEN (within the configured rate) or RED
(exceeding it -- "out of profile"). Severity counts the RED hops along the path;
a GREEN hop does not add to it.</t>
        <artwork><![CDATA[
  EP1 RED   (+1)  carried severity 1   ICV = f52d43f8...
  EP2 GREEN (+0)  carried severity 1   ICV = f9bd54a3...
  EP3 RED   (+1)  carried severity 2   ICV = 9e19a585... (carried)
  Step 1 Integrity -> PASS
  path severity (final carried value) = 2  (two out-of-profile hops)
]]></artwork>
        <t>An egress <bcp14>MAY</bcp14> use the Severity count or a RED marking to treat the packet more
strictly (see <xref target="inline"/>).</t>
      </section>
      <section anchor="scaling-to-more-eps">
        <name>Scaling to More EPs</name>
        <t>Path length changes only the record array, never the ICV: the ICV is always 16
octets, and each EP adds one 8-octet record, so the option grows linearly. For
five EPs (all ACCEPT/0x1001) the size is:</t>
        <artwork><![CDATA[
  option content =  8  (Initial Header)
                 + 32  (Integrity Header: 16-octet ICV + 16 octets of
                        Method ID, Nonce Length, Reserved, and Nonce)
                 + 5 x 8  (per-node records)
                 = 80 octets

  full HBH header = 80 + 4 (framing) + 4 (PadN) = 88 octets
  carried ICV (EP5) = efea617e...
]]></artwork>
      </section>
    </section>
    <section anchor="inline">
      <name>Inline Use and Threshold-Triggered Export</name>
      <t>Inline, an EP or the egress <bcp14>MAY</bcp14> use the Policer color (RED) or the Severity count
to treat a packet more strictly -- deprioritizing it or policing it tighter when
giving out resources. A node that acts on carried Policer State or Severity <bcp14>MUST</bcp14>
first validate the upstream chain (that is, act as a validator for those
records); otherwise it is relying on unauthenticated data an on-path attacker
could have altered. Carried severity is thus trustworthy only at a validating
node, which in the common case is the egress.</t>
      <t>When Severity crosses a configured threshold, the packet's metadata <bcp14>MAY</bcp14> be
exported to a collector. Because each record is authenticated and ordered, the
collector reads the exact path and the RED-marking EPs directly from the packet,
instead of reconstructing the path from independent per-device records. By
correlating Filter IDs across exported packets, the collector can see where
related flows entered on other EPs and build an attack graph.</t>
      <t>Flows do not merge in band: two flows are two packets, each carrying its own
chain. A convergence point <bcp14>MAY</bcp14> independently mark RED when the aggregate of
several separately-policed flows exceeds its own limit. In the reference run, two
flows both cross EP3 (RED):</t>
      <artwork><![CDATA[
  Flow A: EP1(RED) -> EP3(RED)   carried ICV = b76d76c1...
  Flow B: EP2(RED) -> EP3(RED)   carried ICV = 0f4058f2...
]]></artwork>
      <t>The correlation itself is conventional collector and telemetry work (for example
<xref target="RFC8955"/> and <xref target="RFC8811"/>); what is new is the authenticated, deterministic,
per-hop enforcement evidence it consumes -- something flow-record correlation and
probabilistic packet marking cannot provide. Thresholds <bcp14>SHOULD</bcp14> account for path
length, since Severity is bounded by the number of hops.</t>
    </section>
    <section anchor="option-sizing">
      <name>Option Sizing</name>
      <t>The option content is Initial Header (8) + Integrity Header (32) + N records
(8 each). The full Hop-by-Hop Options header adds a 2-octet extension header plus
2-octet option type/length and PadN to a multiple of 8 octets.</t>
      <table>
        <thead>
          <tr>
            <th align="center">Records (N)</th>
            <th align="center">Option content</th>
            <th align="center">Full HBH header</th>
          </tr>
        </thead>
        <tbody>
          <tr>
            <td align="center">3</td>
            <td align="center">64 octets</td>
            <td align="center">72 octets</td>
          </tr>
          <tr>
            <td align="center">5</td>
            <td align="center">80 octets</td>
            <td align="center">88 octets</td>
          </tr>
          <tr>
            <td align="center">8</td>
            <td align="center">104 octets</td>
            <td align="center">112 octets</td>
          </tr>
        </tbody>
      </table>
      <t>The per-hop record is deliberately compact (8 octets) to keep the option small
and bounded by path length, targeting the common firewall and DDoS use cases
while limiting growth of the packet. The ingress <bcp14>MUST</bcp14> size the option --
choosing the number of
pre-allocated records, carried as Capacity (<xref target="inithdr"/>) -- so that the resulting
outer packet, including encapsulation overhead, stays within the PMTU of the
domain and does not require outer fragmentation. An HBH header is copied into
every fragment (<xref target="RFC8200"/>) -- unlike a Destination Options header after a
Routing header, which would appear only in the first fragment -- so fragmenting
would give each fragment its own copy of the mutable option that transit EPs
update independently, producing divergent ICV chains this specification does not
define how to merge. Outer fragmentation therefore <bcp14>MUST NOT</bcp14> be used with this
carriage; the ingress <bcp14>MUST</bcp14> handle an oversized packet using the applicable tunnel
PMTU procedures rather than adding the option to an oversized outer packet. If
an EP still has no free slot (NextIndex equals Capacity), it sets the Overflow (O)
indicator in the Initial Header Flags (<xref target="inithdr"/>) to mark the metadata
incomplete.</t>
    </section>
    <section anchor="security-considerations">
      <name>Security Considerations</name>
      <t>The ICV provides origin authentication and integrity for the per-node records
and the immutable header fields, inheriting the properties and key-management
assumptions of <xref target="I-D.ietf-ippm-ioam-data-integrity"/>. Points specific to this
document:</t>
      <ul spacing="normal">
        <li>
          <t>Mutable header bits (Flags, NextIndex, Reserved) are not authenticated. An on-path
attacker can flip them undetected. No security decision may depend on them;
security-critical status <bcp14>MUST</bcp14> be in immutable per-node fields.</t>
        </li>
        <li>
          <t>The ICV authenticates the records present, not their completeness, so a
validator <bcp14>MUST</bcp14> verify that every expected EP is represented (the completeness
check of <xref target="checking"/>); a skipped or stripped record is thereby detected even
though a chain over a subset is itself valid. Deleting a record while leaving
the ICV is caught as a Step 1 mismatch, but removing the entire option (records
and ICV together) is not detectable here and is left to the encapsulating
protocol, as in <xref target="I-D.ietf-ippm-ioam-data-integrity"/>.</t>
        </li>
        <li>
          <t>Distributed (inline) validation requires EPs to hold upstream keys, widening
key exposure. Because the construction is symmetric, any key holder (including
the validator) can in principle forge records, so this mechanism proves
enforcement within a trust domain, not to an external third party.</t>
        </li>
        <li>
          <t>Replay across packets is bounded by the nonce counter; reuse of a
(Key, Nonce) pair <bcp14>MUST</bcp14> be avoided, per
<xref target="I-D.ietf-ippm-ioam-data-integrity"/>.</t>
        </li>
        <li>
          <t>HBH processing <bcp14>MUST</bcp14> be explicitly enabled only on participating nodes within
the domain.</t>
        </li>
      </ul>
      <t>This mechanism is intended for a single limited domain <xref target="RFC8799"/>.</t>
    </section>
    <section anchor="iana-considerations">
      <name>IANA Considerations</name>
      <t>This document requests an IPv6 Option Type from the Destination Options and
Hop-by-Hop Options registry for firewall metadata (value TBD), for use only in
the Hop-by-Hop Options header. The action bits are 00 (skip an unrecognized
option) and the change-en-route bit is 1 (bit pattern 001xxxxx). It also requests
registries for Action values and
Policer State values, and (for the Optional Data Present variant of
<xref target="record-ext"/>) a registry for Optional Data Present bits. Namespace IDs <bcp14>MAY</bcp14>
reuse the IOAM Namespace-ID registry of <xref target="RFC9197"/>.</t>
    </section>
  </middle>
  <back>
    <references anchor="sec-combined-references">
      <name>References</name>
      <references anchor="sec-normative-references">
        <name>Normative References</name>
        <reference anchor="RFC2119">
          <front>
            <title>Key words for use in RFCs to Indicate Requirement Levels</title>
            <author fullname="S. Bradner" initials="S." surname="Bradner"/>
            <date month="March" year="1997"/>
            <abstract>
              <t>In many standards track documents several words are used to signify the requirements in the specification. These words are often capitalized. This document defines these words as they should be interpreted in IETF documents. This document specifies an Internet Best Current Practices for the Internet Community, and requests discussion and suggestions for improvements.</t>
            </abstract>
          </front>
          <seriesInfo name="BCP" value="14"/>
          <seriesInfo name="RFC" value="2119"/>
          <seriesInfo name="DOI" value="10.17487/RFC2119"/>
        </reference>
        <reference anchor="RFC8174">
          <front>
            <title>Ambiguity of Uppercase vs Lowercase in RFC 2119 Key Words</title>
            <author fullname="B. Leiba" initials="B." surname="Leiba"/>
            <date month="May" year="2017"/>
            <abstract>
              <t>RFC 2119 specifies common key words that may be used in protocol specifications. This document aims to reduce the ambiguity by clarifying that only UPPERCASE usage of the key words have the defined special meanings.</t>
            </abstract>
          </front>
          <seriesInfo name="BCP" value="14"/>
          <seriesInfo name="RFC" value="8174"/>
          <seriesInfo name="DOI" value="10.17487/RFC8174"/>
        </reference>
        <reference anchor="RFC8200">
          <front>
            <title>Internet Protocol, Version 6 (IPv6) Specification</title>
            <author fullname="S. Deering" initials="S." surname="Deering"/>
            <author fullname="R. Hinden" initials="R." surname="Hinden"/>
            <date month="July" year="2017"/>
            <abstract>
              <t>This document specifies version 6 of the Internet Protocol (IPv6). It obsoletes RFC 2460.</t>
            </abstract>
          </front>
          <seriesInfo name="STD" value="86"/>
          <seriesInfo name="RFC" value="8200"/>
          <seriesInfo name="DOI" value="10.17487/RFC8200"/>
        </reference>
      </references>
      <references anchor="sec-informative-references">
        <name>Informative References</name>
        <reference anchor="RFC9197">
          <front>
            <title>Data Fields for In Situ Operations, Administration, and Maintenance (IOAM)</title>
            <author fullname="F. Brockners" initials="F." role="editor" surname="Brockners"/>
            <author fullname="S. Bhandari" initials="S." role="editor" surname="Bhandari"/>
            <author fullname="T. Mizrahi" initials="T." role="editor" surname="Mizrahi"/>
            <date month="May" year="2022"/>
            <abstract>
              <t>In situ Operations, Administration, and Maintenance (IOAM) collects operational and telemetry information in the packet while the packet traverses a path between two points in the network. This document discusses the data fields and associated data types for IOAM. IOAM-Data-Fields can be encapsulated into a variety of protocols, such as Network Service Header (NSH), Segment Routing, Generic Network Virtualization Encapsulation (Geneve), or IPv6. IOAM can be used to complement OAM mechanisms based on, e.g., ICMP or other types of probe packets.</t>
            </abstract>
          </front>
          <seriesInfo name="RFC" value="9197"/>
          <seriesInfo name="DOI" value="10.17487/RFC9197"/>
        </reference>
        <reference anchor="RFC9486">
          <front>
            <title>IPv6 Options for In Situ Operations, Administration, and Maintenance (IOAM)</title>
            <author fullname="S. Bhandari" initials="S." role="editor" surname="Bhandari"/>
            <author fullname="F. Brockners" initials="F." role="editor" surname="Brockners"/>
            <date month="September" year="2023"/>
            <abstract>
              <t>In situ Operations, Administration, and Maintenance (IOAM) records operational and telemetry information in the packet while the packet traverses a path between two points in the network. This document outlines how IOAM Data-Fields are encapsulated in IPv6.</t>
            </abstract>
          </front>
          <seriesInfo name="RFC" value="9486"/>
          <seriesInfo name="DOI" value="10.17487/RFC9486"/>
        </reference>
        <reference anchor="RFC9673">
          <front>
            <title>IPv6 Hop-by-Hop Options Processing Procedures</title>
            <author fullname="R. Hinden" initials="R." surname="Hinden"/>
            <author fullname="G. Fairhurst" initials="G." surname="Fairhurst"/>
            <date month="October" year="2024"/>
            <abstract>
              <t>This document specifies procedures for processing IPv6 Hop-by-Hop options in IPv6 routers and hosts. It modifies the procedures specified in the IPv6 Protocol Specification (RFC 8200) to make processing of the IPv6 Hop-by-Hop Options header practical with the goal of making IPv6 Hop-by-Hop options useful to deploy and use at IPv6 routers and hosts. This document updates RFC 8200.</t>
            </abstract>
          </front>
          <seriesInfo name="RFC" value="9673"/>
          <seriesInfo name="DOI" value="10.17487/RFC9673"/>
        </reference>
        <reference anchor="RFC8754">
          <front>
            <title>IPv6 Segment Routing Header (SRH)</title>
            <author fullname="C. Filsfils" initials="C." role="editor" surname="Filsfils"/>
            <author fullname="D. Dukes" initials="D." role="editor" surname="Dukes"/>
            <author fullname="S. Previdi" initials="S." surname="Previdi"/>
            <author fullname="J. Leddy" initials="J." surname="Leddy"/>
            <author fullname="S. Matsushima" initials="S." surname="Matsushima"/>
            <author fullname="D. Voyer" initials="D." surname="Voyer"/>
            <date month="March" year="2020"/>
            <abstract>
              <t>Segment Routing can be applied to the IPv6 data plane using a new type of Routing Extension Header called the Segment Routing Header (SRH). This document describes the SRH and how it is used by nodes that are Segment Routing (SR) capable.</t>
            </abstract>
          </front>
          <seriesInfo name="RFC" value="8754"/>
          <seriesInfo name="DOI" value="10.17487/RFC8754"/>
        </reference>
        <reference anchor="RFC8986">
          <front>
            <title>Segment Routing over IPv6 (SRv6) Network Programming</title>
            <author fullname="C. Filsfils" initials="C." role="editor" surname="Filsfils"/>
            <author fullname="P. Camarillo" initials="P." role="editor" surname="Camarillo"/>
            <author fullname="J. Leddy" initials="J." surname="Leddy"/>
            <author fullname="D. Voyer" initials="D." surname="Voyer"/>
            <author fullname="S. Matsushima" initials="S." surname="Matsushima"/>
            <author fullname="Z. Li" initials="Z." surname="Li"/>
            <date month="February" year="2021"/>
            <abstract>
              <t>The Segment Routing over IPv6 (SRv6) Network Programming framework enables a network operator or an application to specify a packet processing program by encoding a sequence of instructions in the IPv6 packet header.</t>
              <t>Each instruction is implemented on one or several nodes in the network and identified by an SRv6 Segment Identifier in the packet.</t>
              <t>This document defines the SRv6 Network Programming concept and specifies the base set of SRv6 behaviors that enables the creation of interoperable overlays with underlay optimization.</t>
            </abstract>
          </front>
          <seriesInfo name="RFC" value="8986"/>
          <seriesInfo name="DOI" value="10.17487/RFC8986"/>
        </reference>
        <reference anchor="RFC8799">
          <front>
            <title>Limited Domains and Internet Protocols</title>
            <author fullname="B. Carpenter" initials="B." surname="Carpenter"/>
            <author fullname="B. Liu" initials="B." surname="Liu"/>
            <date month="July" year="2020"/>
            <abstract>
              <t>There is a noticeable trend towards network behaviors and semantics that are specific to a particular set of requirements applied within a limited region of the Internet. Policies, default parameters, the options supported, the style of network management, and security requirements may vary between such limited regions. This document reviews examples of such limited domains (also known as controlled environments), notes emerging solutions, and includes a related taxonomy. It then briefly discusses the standardization of protocols for limited domains. Finally, it shows the need for a precise definition of "limited domain membership" and for mechanisms to allow nodes to join a domain securely and to find other members, including boundary nodes.</t>
              <t>This document is the product of the research of the authors. It has been produced through discussions and consultation within the IETF but is not the product of IETF consensus.</t>
            </abstract>
          </front>
          <seriesInfo name="RFC" value="8799"/>
          <seriesInfo name="DOI" value="10.17487/RFC8799"/>
        </reference>
        <reference anchor="RFC2697">
          <front>
            <title>A Single Rate Three Color Marker</title>
            <author fullname="J. Heinanen" initials="J." surname="Heinanen"/>
            <author fullname="R. Guerin" initials="R." surname="Guerin"/>
            <date month="September" year="1999"/>
            <abstract>
              <t>This document defines a Single Rate Three Color Marker (srTCM), which can be used as component in a Diffserv traffic conditioner. This memo provides information for the Internet community.</t>
            </abstract>
          </front>
          <seriesInfo name="RFC" value="2697"/>
          <seriesInfo name="DOI" value="10.17487/RFC2697"/>
        </reference>
        <reference anchor="RFC8955">
          <front>
            <title>Dissemination of Flow Specification Rules</title>
            <author fullname="C. Loibl" initials="C." surname="Loibl"/>
            <author fullname="S. Hares" initials="S." surname="Hares"/>
            <author fullname="R. Raszuk" initials="R." surname="Raszuk"/>
            <author fullname="D. McPherson" initials="D." surname="McPherson"/>
            <author fullname="M. Bacher" initials="M." surname="Bacher"/>
            <date month="December" year="2020"/>
            <abstract>
              <t>This document defines a Border Gateway Protocol Network Layer Reachability Information (BGP NLRI) encoding format that can be used to distribute (intra-domain and inter-domain) traffic Flow Specifications for IPv4 unicast and IPv4 BGP/MPLS VPN services. This allows the routing system to propagate information regarding more specific components of the traffic aggregate defined by an IP destination prefix.</t>
              <t>It also specifies BGP Extended Community encoding formats, which can be used to propagate Traffic Filtering Actions along with the Flow Specification NLRI. Those Traffic Filtering Actions encode actions a routing system can take if the packet matches the Flow Specification.</t>
              <t>This document obsoletes both RFC 5575 and RFC 7674.</t>
            </abstract>
          </front>
          <seriesInfo name="RFC" value="8955"/>
          <seriesInfo name="DOI" value="10.17487/RFC8955"/>
        </reference>
        <reference anchor="RFC8811">
          <front>
            <title>DDoS Open Threat Signaling (DOTS) Architecture</title>
            <author fullname="A. Mortensen" initials="A." role="editor" surname="Mortensen"/>
            <author fullname="T. Reddy.K" initials="T." role="editor" surname="Reddy.K"/>
            <author fullname="F. Andreasen" initials="F." surname="Andreasen"/>
            <author fullname="N. Teague" initials="N." surname="Teague"/>
            <author fullname="R. Compton" initials="R." surname="Compton"/>
            <date month="August" year="2020"/>
            <abstract>
              <t>This document describes an architecture for establishing and maintaining Distributed Denial-of-Service (DDoS) Open Threat Signaling (DOTS) within and between domains. The document does not specify protocols or protocol extensions, instead focusing on defining architectural relationships, components, and concepts used in a DOTS deployment.</t>
            </abstract>
          </front>
          <seriesInfo name="RFC" value="8811"/>
          <seriesInfo name="DOI" value="10.17487/RFC8811"/>
        </reference>
        <reference anchor="I-D.ietf-ippm-ioam-data-integrity">
          <front>
            <title>Integrity Protection of In Situ Operations, Administration, and Maintenance (IOAM) Data Fields</title>
            <author fullname="Frank Brockners" initials="F." surname="Brockners">
              <organization>Cisco Systems, Inc.</organization>
            </author>
            <author fullname="Shwetha Bhandari" initials="S." surname="Bhandari">
              <organization>Databricks</organization>
            </author>
            <author fullname="Tal Mizrahi" initials="T." surname="Mizrahi">
              <organization>Huawei</organization>
            </author>
            <author fullname="Justin Iurman" initials="J." surname="Iurman">
              <organization>University of Liege</organization>
            </author>
            <date day="18" month="July" year="2026"/>
            <abstract>
              <t>   In Situ Operations, Administration, and Maintenance (IOAM) records
   operational (including telemetry) information in packets while they
   traverse a path in the network.  RFC 9197 specifies data fields for
   IOAM (a.k.a IOAM-Data-Fields) and associated data types.  This
   document specifies integrity protection of IOAM-Data-Fields for
   Intra-IOAM-Domain use cases.

              </t>
            </abstract>
          </front>
          <seriesInfo name="Internet-Draft" value="draft-ietf-ippm-ioam-data-integrity-20"/>
        </reference>
        <reference anchor="I-D.singh-macneil-ioam-firewall-metadata">
          <front>
            <title>IOAM Data Fields for Firewall and Security Metadata</title>
            <author fullname="PRASHANT SINGH" initials="S." surname="Prashant">
              <organization>Hewlett Packard Enterprise</organization>
            </author>
            <author fullname="Erin MacNeil" initials="E." surname="MacNeil">
              <organization>Hewlett Packard Enterprise</organization>
            </author>
            <date day="1" month="July" year="2026"/>
            <abstract>
              <t>   This document defines a new IOAM Namespace and associated Data Fields
   for carrying firewall and security-related metadata within the In-
   situ Operations, Administration, and Maintenance (IOAM) framework
   (RFC 9197).  The mechanism allows each security Enforcement Point on
   a path to attach a verifiable attestation of the action it took on a
   packet.  This lets downstream Enforcement Points and other nodes
   confirm that security policy was actually executed on the data path
   -- detecting policy bypasses and detecting divergence between
   Enforcement Points that are expected to agree -- while each
   Enforcement Point continues to apply its own policy independently
   rather than delegating enforcement.  This validation of execution
   complements the management plane's validation of intent.  The
   metadata also enables real-time signaling of policer states for DDoS
   mitigation.  To ensure the authenticity and integrity of this
   information, this document leverages the integrity protection
   mechanisms defined in the IOAM data integrity specification (draft-
   ietf-ippm-ioam-data-integrity).

              </t>
            </abstract>
          </front>
          <seriesInfo name="Internet-Draft" value="draft-singh-macneil-ioam-firewall-metadata-00"/>
        </reference>
      </references>
    </references>
    <?line 975?>

<section numbered="false" anchor="relationship-to-ioam">
      <name>Relationship to IOAM</name>
      <t>The integrity construction is taken directly, by reference, from IOAM data
integrity <xref target="I-D.ietf-ippm-ioam-data-integrity"/>: the chained AES-GMAC ICV,
Method ID 0, the 12-octet nonce (Key ID, Encapsulating Node ID, counter), the
AND-mask of immutable fields, and all key and nonce management (per-node keys,
Key ID rotation, the counter limit, and avoidance of key-nonce reuse) are those
of that document and are not restated or redefined here. The IPv6 carriage
uses the HBH model of <xref target="RFC9486"/> and the IOAM data fields of <xref target="RFC9197"/>; the
Namespace ID mirrors the IOAM Namespace-ID. This document defines no new
cryptography.</t>
      <t>This document is the native-IPv6 carriage of the same security-enforcement
metadata defined for IOAM in <xref target="I-D.singh-macneil-ioam-firewall-metadata"/>. The
two share field semantics (Policy ID, Policy Version, Filter ID, Action, Policer
State, Severity) and differ only in encoding: the companion document carries the
metadata as IOAM-Data-Fields within an Integrity-Protected Trace Option-Type,
while this document carries a compact, fixed-layout equivalent in a native IPv6
Hop-by-Hop option for deployments that do not run IOAM. On-wire field widths are
compacted here to limit packet growth (see Option Sizing); the semantics are
intended to stay aligned with the companion document.</t>
    </section>
    <section numbered="false" anchor="changes-from-00">
      <name>Changes from -00</name>
      <ul spacing="normal">
        <li>
          <t>Changed the carriage from the Destination Options header to the Hop-by-Hop
Options header, so that transit EPs process the option (<xref target="RFC8200"/>), with
consequential updates to the HBH processing discussion (per <xref target="RFC9673"/>),
the fragmentation and sizing text, and the IANA request.</t>
        </li>
      </ul>
    </section>
    <section numbered="false" anchor="acknowledgments">
      <name>Acknowledgments</name>
      <t>The authors thank Brian Carpenter for noting that Destination Options are not
processed in transit (<xref target="RFC8200"/>), which motivated the Hop-by-Hop carriage.</t>
      <t>The authors thank the authors and contributors of the IOAM data integrity work,
on which the ICV here is based.</t>
    </section>
  </back>
  <!-- ##markdown-source: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-->

</rfc>
