[JarryShaw]

• Searched for similar pull requests
• Followed the coding style — ran pylint/mypy/isort on the two source files: no new findings (the 4 mypy, isort and pylint hits are on untouched lines and also on main); make targets not run
• make test passes, and a test case covers the change — per-module only, listed below
• Added a changelog entry — N/A, added centrally

What is the purpose of your pull request?

• fix — corrects a defect

---

Closes #1177. _decode_next_layer now records the chain in wire order as a hidden __exthdr__ key ((parser class, info) pairs, excluded like __next_type__); _make_data re-emits it ahead of the upper layer. The per-header keys (route, opts, ...) can't carry it, because a repeated header overwrites its key.

Probe: Routing (3b00000000000000): 48 → 40 before, 48 → 48 after; with 4 trailing octets 52 → 44 before, 52 → 52 after.
Capture sweep (1609 frames, non-exact): 183 → 51. IPv6>MH 99 → 0, IPv6_Route 3 → 0, IPv6_Frag 4 → 0, HOPOPT 14 → 1, IPv6_Opts 14 → 1, HIP 49 → 49; no frame regressed. The 51 now raise in the extension classes' own from_data (also on main standalone): HIP.alias reads _info before it is set (49), and _make_opt_mpl calls bit_length() on an IPv6Address in hopopt.py/ipv6_opts.py (2). Out of scope here.

Tests: new test_ipv6_from_data_ext_unit.py (HOPOPT, Routing, Dest Opts, MH, a 4-header chain with a repeated Dest Opts): 7 failed without the fix, 3 passed / 6 subtests with it. test_ipv6_unit, test_ipv6_extension_unit/_runtime, test_ipv6_ext_unit, test_mh_unit, test_hip_unit, test_option_roundtrip_unit all pass; tests/project 389 passed, 1 skipped; run_unittest_leg.py protocols/internet 475 tests, 0 failures.

[JarryShaw]

Cross-review verdict on 55c8ed67e: GOOD TO GO (ran on Sonnet; author Opus). Labelled review: good-to-go.

• Sweep, per frame against main, 1602 frames: 1422 were exact on both; 130 went from wrong bytes to exact (MH 99, HOPOPT 12, IPv6_Opts 12, Frag 4, Route 3); 50 went from wrong bytes to AttributeError; 0 regressed from exact.
• The 50 raising frames: a loud failure now replaces a silent wrong answer. from_data is a user-facing rebuild API that extract, the engines and dumpkit never call. The causes are two pre-existing defects, now filed as fix(hip): from_data raises AttributeError, alias reads _info before it is set #1184 (HIP, hip.py:475) and fix(ipv6-opts): _make_opt_mpl calls bit_length() on an IPv6Address seed #1185 (MPL seed, hopopt.py:1878 / ipv6_opts.py:1867). I checked both sites on main myself.
• The __exthdr__ key stays hidden: it is absent from to_dict, iteration and repr, and from extract json/tree/plist output. It survives pickle and deepcopy.
• Repeated headers: two Destination Options headers rebuild 64/64.
• Tests: the new test fails without the fix (7 failures). The protocols/internet ordering leg ran 475 tests, 0F/0E.
• Nit: test_chain_is_hidden_from_info_views passes on main too. It guards against a leak but does not prove the fix.

[github-actions]

Coverage: 88.89% (unit tier, Python 3.14, 55c8ed67e, Unit Tests run success)

Package	Statements	Missed	Branches	Partial	Cover
pcapkit (top level)	104	4	20	4	93.55%
pcapkit/const	18757	1034	2342	845	90.24%
pcapkit/corekit	1878	91	580	22	94.34%
pcapkit/dumpkit	142	0	42	0	100.00%
pcapkit/foundation	2442	102	852	39	94.38%
pcapkit/interface	112	7	40	5	92.11%
pcapkit/protocols	15715	188	3976	163	98.19%
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]

Merge note: merge this together with #1188. On its own, this PR makes IPv6.from_data raise AttributeError on all 49 IPv6>HIP frames of options-internet.pcap (re-measured at 55c8ed67e). With #1188 merged in, all of them rebuild byte-exactly.