[JarryShaw]

• Searched for similar pull requests
• Followed the coding style (make pylint, make mypy, make isort) — pylint/mypy/isort run with the Makefile flags on the three changed source files; no new messages
• make test passes, and a test case covers the change — the modules listed below only
• Added a changelog entry — N/A, added centrally

What is the purpose of your pull request?

• fix — corrects a defect

---

Description

Closes #1207
Closes #1208

#1207: the flags BitField gains reserved_1 (bits 2-3), reserved_2 (bit 5) and reserved_3 (bits 8-11); Flags.reserved holds them in on-wire positions (mask 0x34F0), and make(reserved=...) writes them back. 200200010002ff03002145000000 was rebuilt as 000200…; it now rebuilds byte-exactly, as do masks 0x1000, 0x0400, 0x00f0 and 0x34f0, also inside UDP/1701.

#1208: RFC 2661 §3.1 has S mark Ns and Nr together, so a lone value cannot be represented. I chose ProtocolError over defaulting the other to 0: a default would write an Nr/Ns the caller never asked for, and the library's other builders refuse to invent wire values. L2TPv2(ns=7) used to return 000200000000; it now raises ProtocolError: L2TPv2: ns and nr must be given together.

Breaking: data.link.l2tp.Flags has a new required field reserved, and the schema flags dict three new keys. make() now raises where it used to drop a lone ns/nr, and also raises for reserved bits outside 0x34F0. tests/protocols/link/test_link_unit.py hand-built fixtures gained the new keys (they did not pin wrong behaviour, but they omitted keys the real schema always produces).

Tests: new tests/protocols/link/test_l2tpv2_roundtrip_unit.py: 6 passed / 12 subtests with the fix; reverting the fix gives 12 failed, 3 passed (the 3 being a positive control and the outer tests whose subtests failed). Also passed: test_link_unit (24), test_l2tpv2_length_roundtrip_unit (4), test_l2tpv2_type_unit (8), test_l2tp_version_unit (7), test_make_data_keys_unit (2), test_construction_keyword_check_unit (21), test_option_roundtrip_unit (6, no L2TP entry in EXPECTED_FAILURES); run_unittest_leg.py protocols --exclude protocols/internet: 856 tests, 0 failures; tests/project: 389 passed, 1 skipped.

[github-actions]

Coverage: 88.91% (unit tier, Python 3.14, 1c11bb347, 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	15781	190	4040	165	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 1c11bb347: GOOD TO GO (Sonnet reviewed; the author ran on Opus).

• RFC 2661 §3.1 layout: checked against the RFC text. The reserved bits are 2–3, 5 and 8–11, MSB-first, so the mask is 0x34F0. T, L, S, O and P sit at bits 0, 1, 4, 6 and 7.
• Reserved bits survive a rebuild: 200200010002ff03002145000000, and masks 0x1000, 0x0400, 0x00f0 and 0x34f0, all rebuild byte-exactly, both bare and inside UDP port 1701. I re-ran the first case myself.
• ns/nr handling: §3.1 says Ns and Nr are present together when S is set. L2TPv2(ns=7) now raises ProtocolError, which I re-ran, and ns=7, nr=0 round-trips. No other caller passes only one of the two.
• Existing behaviour: no fixture contains L2TP. In 7 synthetic UDP/1701 frames, the parsed info is identical to main apart from the new reserved key.
• Tests: the edited test_link_unit.py hunks only add keys. The new tests fail on main (12 of 15). run_unittest_leg.py protocols --exclude protocols/internet: 856 tests, 0 failures.
• Nit: the module docstring table in l2tpv2.py still lists the reserved bits as unnamed.