)]}'
{
  "commit": "f22870148c1aec3fb22af656f4f9caf29dc31ead",
  "tree": "ef7c077a648ddb2ccb65c802a9d994ce56fba0ff",
  "parents": [
    "3bcbc984f89116432f1cad029ff444a25ef30051"
  ],
  "author": {
    "name": "graham sanderson",
    "email": "graham.sanderson@raspberrypi.com",
    "time": "Tue Aug 11 20:09:25 2026 -0500"
  },
  "committer": {
    "name": "graham sanderson",
    "email": "graham.sanderson@raspberrypi.com",
    "time": "Tue Aug 11 20:09:25 2026 -0500"
  },
  "message": "sync_interop_test: add D4.1, the alarm pool\u0027s always-armed invariant\n\nChecks that a pool leaves its hardware alarm armed after emptying, which is what lets\nta_wakes_up_on_or_before() answer from the compare value alone.\n\nTested as the invariant, not the symptom: the symptom needs a deadline over 71 minutes out,\nthe invariant is one register bit and takes two milliseconds. Uses the suite\u0027s private pool\nso the default pool\u0027s own traffic cannot keep an alarm armed by accident and hide a\nfailure - and verified to fail without the fix, reporting armed\u003d0x00.\n\nFirst case in a D4 group for alarm pool state rather than the wait primitives, which\nSYNC_TIME_TEST_GAPS.md has listed as a gap since the suite was written.\n\nCo-Authored-By: Claude Opus 5 \u003cnoreply@anthropic.com\u003e\n",
  "tree_diff": [
    {
      "type": "modify",
      "old_id": "57c315d2fa645b0b5421925c8d16be26c32c9bae",
      "old_mode": 33188,
      "old_path": "test/sync_interop_test/sync_interop_test.c",
      "new_id": "9e7d54f6612c2ef59ef3352617e3345d5394d547",
      "new_mode": 33188,
      "new_path": "test/sync_interop_test/sync_interop_test.c"
    }
  ]
}