fix(hip): HOST_ID packs DI-Type/DI-Length as 4 octets where RFC 7401 has one 16-bit word

[JarryShaw]

• Searched for similar pull requests
• Followed the coding style (make pylint, make mypy, make isort) — ran isort/mypy; both already fail on schema/internet/hip.py at origin/main (measured), and this diff adds no imports
• make test passes, and a test case covers the change — ran tests/protocols/internet/test_hip_host_id_di_width_unit.py, tests/protocols/test_option_roundtrip_unit.py, tests/protocols/internet/test_hip_unit.py, the two sibling HIP width modules, and tests/project
• Added a changelog entry under docs/source/changelog/ and regenerated CHANGELOG.md, if the change is user-visible — N/A, added centrally after the wave

What is the purpose of your pull request?

• fix — corrects a defect
• feat — adds a feature
• perf — changes performance, not behaviour
• refactor — changes neither behaviour nor performance
• test — tests only
• docs — documentation only
• ci — workflows or build tooling
• chore — anything else

---

Description of your pull request and other information

Closes #1118.

RFC 7401 §5.2.9 packs DI-Type (4 bits) and DI Length (12 bits) into one 16-bit word between HI Length and Algorithm; the schema's namespace said so, the BitField's length did not. length=4 → length=2.

Probe, ECDSA/NIST P-256 with a 4-octet key — before / after:

made: 02c1000c0006 00000000 0007000100000000  18 octets, len=12   # two surplus octets
made: 02c1000c0006 0000     0007000100000000  16 octets, len=12
hand-built RFC record → algorithm 1 (the curve) / algorithm 7

Rebased onto c24df4fbe after #1128 landed. With #1128's HIP_TRANSFORM fix and this one, no HIP case fails at either setting of HIP_COPIES — measured 49 OK of 49 at one copy and at two — so EXPECTED_FAILURES loses its last HIP entry and the block goes with it; the HIP_COPIES note records that instead of calling HOST_ID a gap. make_samples.py now leaves out 3 cases, none of them HIP, and options-internet.pcap carries both 0x02c1 (705) and 0x0241 (577).

Fails without the fix — reverting length=2 → length=4 turns all three new tests red: ProtocolError: HIPv2: invalid format, '02c1000e00061002000000070001aabbccdd6162…' != '02c1000e0006100200070001aabbccdd6162…', and HIAlgorithm.NULL_ENCRYPT != HIAlgorithm.ECDSA. With the fix: 3 passed; roundtrip 6 passed, 360 subtests; test_hip_unit.py 34 passed, 168 subtests; tests/project 385 passed, 1 skipped, 1289 subtests.

[JarryShaw]

Cross-review verdict on 7693f2099: GOOD TO GO (ran on Sonnet; author Opus)

• RFC 7401 §5.2.9: DI-Type and DI Length share one 16-bit word, so length=2 is right.
  • 144 hand-built HOST_ID records (ECDSA and RSA, DI lengths 0–5) parse and re-pack correctly on this PR.
  • main misreads the algorithm in 66 cases.
• _make_param_host_id: len = 6 + hi_len + di_len matches the RFC. Padding to the 8-octet boundary matches an independent formula across 66 combinations.
• Regression tests: with the schema reverted, the 3 new tests fail.
• Existing tests: test_option_roundtrip_unit (360 subtests) and test_hip_unit (34) pass.
• BitField audit: none of the 106 schema BitFields overflows. Four reserved-bit fields checked against their RFCs are correct: MH 833, 1914, 2357 and ipv6_route 206.

Merge order with #1128: the two PRs conflict in options.py and test_option_roundtrip_unit.py, both textually and semantically. Whichever merges second has to rebase:

• delete both HIP EXPECTED_FAILURES entries;
• rewrite the HIP_COPIES note, which would otherwise call HOST_ID the remaining gap.

[github-actions]

Coverage: 88.75% (unit tier, Python 3.14, c25424888, Unit Tests run success)

Package	Statements	Missed	Branches	Partial	Cover
pcapkit (top level)	104	4	20	4	93.55%
pcapkit/const	18757	1035	2342	846	90.23%
pcapkit/corekit	1873	91	578	22	94.33%
pcapkit/dumpkit	136	0	40	0	100.00%
pcapkit/foundation	2425	145	842	34	92.56%
pcapkit/interface	112	7	40	5	92.11%
pcapkit/protocols	15659	188	3944	163	98.18%
pcapkit/toolkit	487	65	144	1	85.42%
pcapkit/utilities	429	4	122	4	98.55%
pcapkit/vendor	4409	2359	1006	158	42.84%

Per-file detail: the coverage-html artifact of this run.

[JarryShaw]

Cross-review verdict on c25424888: GOOD TO GO (ran on Sonnet; author Opus, round 2)

Delta review of the rebase after #1128.

• The fix is unchanged from round 1: di_data is length=2, and the new test module is byte-identical.
• EXPECTED_FAILURES: no HIP entries remain, checked by importing from the worktree. A comment credits test(generators): HIP_TRANSFORM is generated into a HIPv2 packet although it is HIPv1-only #1128 and fix(hip): HOST_ID packs DI-Type/DI-Length as 4 octets where RFC 7401 has one 16-bit word #1118.
• The HIP_COPIES note is now true:
  • HIP_VERSION still carries test(generators): HIP_TRANSFORM is generated into a HIPv2 packet although it is HIPv1-only #1128's 577: 1.
  • options.outcomes() over all 49 hip-parameter cases gives 49 OK.
• Tests: the new module (3), test_option_roundtrip_unit (360 subtests) and test_hip_unit (34) pass.

Unverified prose detail: the note's len=8 figure for the old HOST_ID record. Round 1 measured len=12 for a 4-octet key. This does not affect the fix.