Fix perf regression in PR #3530

synapse/util/caches/stream_change_cache.py

@@ -74,12 +74,14 @@ def get_entities_changed(self, entities, stream_pos):
         assert type(stream_pos) is int
 
         if stream_pos >= self._earliest_known_stream_pos:
-            result = {
+            changed_entities = {
                 self._cache[k] for k in self._cache.islice(
                     start=self._cache.bisect_right(stream_pos),
                 )
             }
 
+            result = changed_entities.intersection(entities)
+
             self.metrics.inc_hits()
         else:
             result = set(entities)

tests/util/test_stream_change_cache.py

@@ -178,6 +178,22 @@ def test_get_entities_changed(self):
             ),
         )
 
+        # Query a subset of the entries mid-way through the stream. We should
+        # only get back the subset.
+        self.assertEqual(
+            cache.get_entities_changed(
+                [
+                    "bar@baz.net",
+                ],
+                stream_pos=2,
+            ),
+            set(
+                [
+                    "bar@baz.net",
+                ]
+            ),
+        )
+
     def test_max_pos(self):
         """
         StreamChangeCache.get_max_pos_of_last_change will return the most