IPv4 option schemas: a fifth bool-laundering site, a hardcoded Quick-Start length, and a dropped ts_data keyword

[JarryShaw]

Three defects in IPv4 option schemas, grouped because they share pcapkit/protocols/schema/internet/ipv4.py and want one pass. All found by a sweep of deferred work across this release's 68 merged PRs.

1. TSOption.post_process launders a bool into an address — the fifth instance

pcapkit/protocols/schema/internet/ipv4.py:278,293 call bare ipaddress.ip_address(ip). Measured on main at e7e9ba98f:

TSOption(type=TS, length=12, pointer=5, flags={'oflw':0,'flag':1}, ts_data=[True, 5]).pack()
-> 440c050100000001000000050000000000000000

True became IPv4Address('0.0.0.1') with no exception. Reachable through public IPv4.make(options=[...]), since ipv4.py:1229 accepts a caller-built schema and calls .pack().

This is the fifth site of a defect already fixed four times — #481, #500, #539, with #540 filed for four more. #540 lists this site only as an unconfirmed candidate; it is now confirmed. The same sweep established the other three #540 candidates — esp.py:575, hip.py:3072/:3077, ipv4.py:308 — do not reproduce, being already guarded or bytes-only, which closes #540's open question in both directions.

Fix: route through parse_ip_address from pcapkit/corekit/fields/ipaddress.py, as #539 did for its seven sites.

2. Quick-Start returns a hardcoded 5-octet length

pcapkit/protocols/schema/internet/ipv4.py:128:

return SchemaField(length=5, schema=schema)

The length is fixed at 5 regardless of which Quick-Start suboption the selector resolved, but the option is 5 or 8 octets. On reparse the reader consumes 5, resynchronises on the nonce's second byte, and dies with ProtocolError — silent corruption rather than a clean failure.

Logged during review of #430 and again in #434, and no issue was filed either time.

3. _make_opt_ts passes a keyword the schema does not have

pcapkit/protocols/internet/ipv4.py:1593 passes data=ts_list where the schema field is ts_data. The value is silently dropped, which makes the IPv4 Timestamp option unbuildable through make. Also flagged in #430 and re-flagged in #434 as "carried across unchanged", never filed.

Note the schema rejects unknown field names with UnknownFieldWarning rather than an error, which is why this is silent — worth considering whether that warning should be fatal in make.

Coverage

Each needs a test proving it fails without its fix. For 1, assert the rejection and that a real address still round-trips. For 2, a test that a 8-octet Quick-Start suboption survives a make/parse cycle byte-for-byte. For 3, a test that builds the Timestamp option through IPv4.make and checks the timestamps are present rather than dropped.