fix: address final-review items (status endpoint, docs, tests)

Follow-up to the holistic review of the PIN-unification branch:

- /system/status now reads the real arm state from the arm_state_log
  table via get_current_arm_state, instead of returning a hardcoded
  'DISARMED' stub. Without this, polling after the new async 202
  arm/disarm flow was a UX dead-end — clients never saw the state
  change they just requested. DB read failures degrade gracefully.

- Operator guide: correct the claim that 'vigilar config set-pin'
  populates recovery_passphrase_hash. It doesn't. recovery_passphrase
  _hash has no CLI helper today; it must be set manually.

- Tests: add a fail-closed regression for verify_pin on malformed
  stored hashes, and a companion test confirming the deprecation
  warning stays silent on a fully migrated config.

All address specific review comments on the branch; no scope creep.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This commit was merged in pull request #7.
This commit is contained in:
adlee-was-taken
2026-04-05 12:26:30 -04:00
parent eb281ad058
commit 5745388880
5 changed files with 93 additions and 5 deletions

View File

@@ -37,3 +37,14 @@ def test_verify_pin_handles_unicode():
stored = hash_pin("p@ss!")
assert verify_pin("p@ss!", stored) is True
assert verify_pin("p@ss?", stored) is False
def test_verify_pin_rejects_malformed_hash():
"""verify_pin must return False (not raise) on malformed stored hashes.
Fail-closed is load-bearing: a misconfigured or partially-migrated
[security] pin_hash must lock out transitions, not grant access."""
assert verify_pin("1234", "sha256:deadbeef") is False
assert verify_pin("1234", "garbage") is False
assert verify_pin("1234", "pbkdf2_sha256$only$two$extra") is False
# Wrong algo prefix
assert verify_pin("1234", "argon2id$salt$dk") is False