Ashin Walpola
0e9525162d
obu-firmware builds against vanetza-idf from microbu-esp32c5/external, but that tree was gitignored, so a clone of this repository could not build the firmware it ships. It is now committed here as ordinary files in its own folder, microbu-esp32c5/: the colleague's commit cf4b99f plus the V2X2MAP bridge's signature verification (--trust) used on the bench. Nothing is fetched from or pushed to the colleague's repository; this repository and its remotes carry everything. The folder's own .gitignore keeps build output, downloaded components and private key material out, as it did there; the committed file set is identical to that repository's tracked files. The ESP32-C5 is still flashed from obu-firmware/, which only takes vanetza-idf from microbu-esp32c5/, so the two stay separate folders. FLASHING.md says how to take a newer version of the colleague's tree (copy it over the folder, rebuild, test, commit).
2.6 KiB
security-host-03: AtsSecurity, GN-MGMT profile (sending side)
Campaign (re-run of security-host-01 with the 2026-09-14 sources: p2pcd accessor patch, ITS time helper in the adapter and pool tool): ports/esp_idf/tests/etsi_security_gn.cfg (8 cases), host SUT
vidf_sut --security-pool ./certificates (native Linux build,
VIDF_SECURITY=ON), pool generated by vidf_test_pool (hashes in
result.json, pool_sha256). The adapter enforces the framework's own
verification of every IUT transmission (enable_security_checks=1, see
test.cfg); a PASS therefore covers a signature the framework verified against
CERT_IUT_A_AT from the pool, not only the message structure.
| Case | Verdict | Cause |
|---|---|---|
| TC_SEC_ITSS_SND_GENMSG_01..04, 06..08 | pass | secured beacons (psid 141), TS 103 097 clause 7.1.3 profile |
| TC_SEC_ITSS_SND_GENMSG_05_BV | fail | ATS implementation defect, IUT-independent (below) |
TC_SEC_ITSS_SND_GENMSG_05_BV
The testcase (pinned ItsSecurity_TestCases.ttcn line 7313) compares the
signer certificate's validityPeriod.start (a Time32, seconds since
2004-01-01) against a range built from v_curTime in microseconds:
match(v_signerIdentifier.certificate[0].toBeSigned.validityPeriod.start_,
Time32:(v_curTime - c_timeLimit / 1000000 .. v_curTime + c_timeLimit / 1000000))
Only c_timeLimit is scaled to seconds; v_curTime (f_getCurrentTime() * 1000) is not. Values from the MTC log of this run:
| Quantity | Value | Unit |
|---|---|---|
v_curTime |
716425780437000 | µs |
| range checked | 716425780436700 .. 716425780437300 | (µs magnitude) |
validityPeriod.start_ |
716425702 | s |
generationTime |
716425785435539 | µs (= 716425785 s) |
duration |
24 | hours |
No Time32 can fall into a range of that magnitude, so the branch fails for
every IUT. The test purpose quoted in the testcase header (TS 103 096-2:
X_START_VALIDITY <= GEN_TIME and duration > GEN_TIME - X_START_VALIDITY)
holds for the observed values: 716425702 s <= 716425785 s and the difference
(83 s) is below 24 h. The current upstream head of the Security ATS
(forge.etsi.org/rep/ITS/ttcn/ats_sec_ts103096-3, branch master, line 6672)
carries the same expression. The verdict is retained as recorded; the testcase
was not modified.
Time base note
The framework's base_time derives ITS time from UTC without the five leap
seconds inserted since 2004; the SUT clock driven by the adapter is TAI-based
(IEEE 1609.2 Time64), hence the constant 5 s offset visible in the
generation time check lines of console.txt (well inside the 30 s window the
framework applies and the 5 min window of the testcases).