Dataset-driven goldens: any tested blob must be a known linux-firmware revision (sha-keyed) matching recorded metadata; works for any snapshot; CI + fetch-fw target

tests/test_extract.py

@@ -14,27 +14,29 @@ from pathlib import Path
 REPO = Path(__file__).resolve().parent.parent
 FW = Path(os.environ.get('MTK_FW_DIR', '/lib/firmware/mediatek'))
 
-# filename -> (chip_id, eco, n_region, hidden_trailer_string_or_None)
-# Golden values observed from linux-firmware HEAD (2026-06-19 tree:
-# mt7981/mt7986 = 20260515 builds; mt7915/mt7916 = 2022/2024 builds,
-# unchanged since). When a value here fails against a newer snapshot, the
-# blob changed — update the golden *and* record the diff (docs/reports/).
-GOLDEN_WM = {
-    'mt7981_wm.bin': (0x14, 0, 11,
-                      't-neptune-main-mt7915-1953-MT7981_MP2111_IMP-20260515121129'),
-    'mt7986_wm.bin': (0x0f, 0, 11,
-                      't-neptune-main-mt7915-1953-MT7986_MP2111_IMP-20260515120228'),
-    'mt7916_wm.bin': (0x13, 0, 9,
-                      't-neptune-main-mt7915-1953-MT7916_MP2111_IMP-20240823170147'),
-    'mt7915_wm.bin': (0x0b, 1, 7,
-                      't-neptune-mp-mt7915-2045-MT7915_MP_7_4_2045-20220929103802'),
-}
-GOLDEN_WA = {
-    'mt7981_wa.bin': (0x00, 0, 3, None),
-    'mt7986_wa.bin': (0x00, 0, 3, None),
-    'mt7916_wa.bin': (0x00, 0, 3, None),
-    'mt7915_wa.bin': (0x00, 1, 3, None),
-}
+# Golden source of truth: dataset/history.jsonl (one record per
+# linux-firmware revision, sha256-keyed). Any tested blob must be a known
+# revision and its parse must match the recorded metadata. A new upstream
+# build therefore fails here until the dataset is refreshed (by design):
+#
+#     python3 tools/fw_history.py firmware/linux-firmware -o dataset/history.jsonl
+import json
+
+HISTORY = json.loads(
+    '[' + ','.join(l for l in
+                   (REPO / 'dataset' / 'history.jsonl').read_text().splitlines()
+                   if l) + ']')
+BY_SHA = {r['sha256']: r for r in HISTORY}
+
+
+def _expected(path: Path, sha: str):
+    rec = BY_SHA.get(sha)
+    if rec is None:
+        raise AssertionError(
+            f'{path.name}: sha {sha} is not a known linux-firmware revision '
+            f'in dataset/history.jsonl — refresh the dataset (see comment '
+            f'above) or verify the blob provenance')
+    return rec
 
 
 def parse(path: Path, out: Path) -> dict:
@@ -67,40 +69,43 @@ class LayoutClosure(unittest.TestCase):
 
 
 class GoldenRAM(unittest.TestCase):
+    """Parsed metadata must match the dataset record for the blob's sha."""
 
-    def run_one(self, name, golden, outdir):
-        p = FW / name
-        if not p.exists():
-            self.skipTest(f'{name} not present')
-        info = parse(p, outdir)
-        chip, eco, n, hidden = golden
-        self.assertEqual(info['chip_id'], chip, name)
-        self.assertEqual(info['eco'], eco, name)
-        self.assertEqual(info['n_region'], n, name)
-        ht = info.get('hidden_trailer', {})
-        self.assertEqual(ht.get('string'), hidden, name)
-
-    def test_wm(self):
-        for name, golden in GOLDEN_WM.items():
-            with self.subTest(blob=name):
-                self.run_one(name, golden, REPO / 'extracted' / 'test_wm')
-
-    def test_wa(self):
-        for name, golden in GOLDEN_WA.items():
-            with self.subTest(blob=name):
-                self.run_one(name, golden, REPO / 'extracted' / 'test_wa')
+    def test_ram(self):
+        import hashlib
+        import sys as _sys
+        _sys.path.insert(0, str(REPO / 'tools'))
+        blobs = sorted(FW.glob('mt79*_wm.bin')) + sorted(FW.glob('mt79*_wa.bin'))
+        if not blobs:
+            self.skipTest('no mediatek blobs on this host')
+        for p in blobs:
+            with self.subTest(blob=p.name):
+                d = p.read_bytes()
+                rec = _expected(p, hashlib.sha256(d).hexdigest()[:16])
+                info = parse(p, REPO / 'extracted' / 'test_ram')
+                self.assertEqual(info['chip_id'], rec.get('chip_id'), p.name)
+                self.assertEqual(info['eco'], rec.get('eco'), p.name)
+                self.assertEqual(info['n_region'], rec.get('n_region'), p.name)
+                self.assertEqual(info.get('hidden_trailer', {}).get('string'),
+                                 rec.get('hidden'), p.name)
+                self.assertEqual(info['build_date'], rec.get('trailer_date'),
+                                 p.name)
 
 
 class GoldenPatch(unittest.TestCase):
 
     def test_patch(self):
-        golden = {'mt7981_rom_patch.bin': 1, 'mt7986_rom_patch.bin': 1,
-                  'mt7916_rom_patch.bin': 1, 'mt7915_rom_patch.bin': 2}
-        for p in sorted(FW.glob('mt79*_rom_patch.bin')):
+        import hashlib
+        blobs = sorted(FW.glob('mt79*_rom_patch.bin'))
+        if not blobs:
+            self.skipTest('no mediatek blobs on this host')
+        for p in blobs:
             with self.subTest(blob=p.name):
+                d = p.read_bytes()
+                rec = _expected(p, hashlib.sha256(d).hexdigest()[:16])
                 info = parse(p, REPO / 'extracted' / 'test_patch')
-                self.assertEqual(info['platform'], 'ALPS')
-                self.assertEqual(info['n_region'], golden[p.name])
+                self.assertEqual(info['platform'], rec.get('platform'), p.name)
+                self.assertEqual(info['n_region'], rec.get('n_section'), p.name)
                 for sec in info['sections']:
                     self.assertEqual(sec['type'], '0x30002')
                     self.assertEqual(sec['enc_type'], 0)  # plaintext (F4)