Faster Remote Room Joins: tell remote homeservers that we are unable to authorise them if they query a room which has partial state on our server. (#13823)

11 files changed, 58 insertions, 42 deletions

diff --git a/changelog.d/13823.misc b/changelog.d/13823.misc
new file mode 100644
index 0000000000..527d79f4b2
--- /dev/null
+++ b/changelog.d/13823.misc
@@ -0,0 +1 @@
+Faster Remote Room Joins: tell remote homeservers that we are unable to authorise them if they query a room which has partial state on our server.
\ No newline at end of file
diff --git a/synapse/api/errors.py b/synapse/api/errors.py
index e6dea89c6d..1c6b53aa24 100644
--- a/synapse/api/errors.py
+++ b/synapse/api/errors.py
@@ -100,6 +100,12 @@ class Codes(str, Enum):
 
     UNREDACTED_CONTENT_DELETED = "FI.MAU.MSC2815_UNREDACTED_CONTENT_DELETED"
 
+    # Returned for federation requests where we can't process a request as we
+    # can't ensure the sending server is in a room which is partial-stated on
+    # our side.
+    # Part of MSC3895.
+    UNABLE_DUE_TO_PARTIAL_STATE = "ORG.MATRIX.MSC3895_UNABLE_DUE_TO_PARTIAL_STATE"
+
 
 class CodeMessageException(RuntimeError):
     """An exception with integer code and message string attributes.
diff --git a/synapse/config/experimental.py b/synapse/config/experimental.py
index bf27f6c101..595eb007a5 100644
--- a/synapse/config/experimental.py
+++ b/synapse/config/experimental.py
@@ -63,7 +63,8 @@ class ExperimentalConfig(Config):
         # MSC3706 (server-side support for partial state in /send_join responses)
         self.msc3706_enabled: bool = experimental.get("msc3706_enabled", False)
 
-        # experimental support for faster joins over federation (msc2775, msc3706)
+        # experimental support for faster joins over federation
+        # (MSC2775, MSC3706, MSC3895)
         # requires a target server with msc3706_enabled enabled.
         self.faster_joins_enabled: bool = experimental.get("faster_joins", False)
 
diff --git a/synapse/federation/federation_server.py b/synapse/federation/federation_server.py
index 3bf84cf625..907940e19e 100644
--- a/synapse/federation/federation_server.py
+++ b/synapse/federation/federation_server.py
@@ -530,13 +530,10 @@ class FederationServer(FederationBase):
     async def on_room_state_request(
         self, origin: str, room_id: str, event_id: str
     ) -> Tuple[int, JsonDict]:
+        await self._event_auth_handler.assert_host_in_room(room_id, origin)
         origin_host, _ = parse_server_name(origin)
         await self.check_server_matches_acl(origin_host, room_id)
 
-        in_room = await self._event_auth_handler.check_host_in_room(room_id, origin)
-        if not in_room:
-            raise AuthError(403, "Host not in room.")
-
         # we grab the linearizer to protect ourselves from servers which hammer
         # us. In theory we might already have the response to this query
         # in the cache so we could return it without waiting for the linearizer
@@ -560,13 +557,10 @@ class FederationServer(FederationBase):
         if not event_id:
             raise NotImplementedError("Specify an event")
 
+        await self._event_auth_handler.assert_host_in_room(room_id, origin)
         origin_host, _ = parse_server_name(origin)
         await self.check_server_matches_acl(origin_host, room_id)
 
-        in_room = await self._event_auth_handler.check_host_in_room(room_id, origin)
-        if not in_room:
-            raise AuthError(403, "Host not in room.")
-
         resp = await self._state_ids_resp_cache.wrap(
             (room_id, event_id),
             self._on_state_ids_request_compute,
@@ -955,6 +949,7 @@ class FederationServer(FederationBase):
         self, origin: str, room_id: str, event_id: str
     ) -> Tuple[int, Dict[str, Any]]:
         async with self._server_linearizer.queue((origin, room_id)):
+            await self._event_auth_handler.assert_host_in_room(room_id, origin)
             origin_host, _ = parse_server_name(origin)
             await self.check_server_matches_acl(origin_host, room_id)
 
diff --git a/synapse/handlers/event_auth.py b/synapse/handlers/event_auth.py
index c3ddc5d182..8249ca1ed2 100644
--- a/synapse/handlers/event_auth.py
+++ b/synapse/handlers/event_auth.py
@@ -31,7 +31,6 @@ from synapse.events import EventBase
 from synapse.events.builder import EventBuilder
 from synapse.events.snapshot import EventContext
 from synapse.types import StateMap, get_domain_from_id
-from synapse.util.metrics import Measure
 
 if TYPE_CHECKING:
     from synapse.server import HomeServer
@@ -156,9 +155,33 @@ class EventAuthHandler:
             Codes.UNABLE_TO_GRANT_JOIN,
         )
 
-    async def check_host_in_room(self, room_id: str, host: str) -> bool:
-        with Measure(self._clock, "check_host_in_room"):
-            return await self._store.is_host_joined(room_id, host)
+    async def is_host_in_room(self, room_id: str, host: str) -> bool:
+        return await self._store.is_host_joined(room_id, host)
+
+    async def assert_host_in_room(
+        self, room_id: str, host: str, allow_partial_state_rooms: bool = False
+    ) -> None:
+        """
+        Asserts that the host is in the room, or raises an AuthError.
+
+        If the room is partial-stated, we raise an AuthError with the
+        UNABLE_DUE_TO_PARTIAL_STATE error code, unless `allow_partial_state_rooms` is true.
+
+        If allow_partial_state_rooms is True and the room is partial-stated,
+        this function may return an incorrect result as we are not able to fully
+        track server membership in a room without full state.
+        """
+        if not allow_partial_state_rooms and await self._store.is_partial_state_room(
+            room_id
+        ):
+            raise AuthError(
+                403,
+                "Unable to authorise you right now; room is partial-stated here.",
+                errcode=Codes.UNABLE_DUE_TO_PARTIAL_STATE,
+            )
+
+        if not await self.is_host_in_room(room_id, host):
+            raise AuthError(403, "Host not in room.")
 
     async def check_restricted_join_rules(
         self,
diff --git a/synapse/handlers/federation.py b/synapse/handlers/federation.py
index dd4b9f66d1..583d5ecd77 100644
--- a/synapse/handlers/federation.py
+++ b/synapse/handlers/federation.py
@@ -804,7 +804,7 @@ class FederationHandler:
             )
 
         # now check that we are *still* in the room
-        is_in_room = await self._event_auth_handler.check_host_in_room(
+        is_in_room = await self._event_auth_handler.is_host_in_room(
             room_id, self.server_name
         )
         if not is_in_room:
@@ -1150,9 +1150,7 @@ class FederationHandler:
     async def on_backfill_request(
         self, origin: str, room_id: str, pdu_list: List[str], limit: int
     ) -> List[EventBase]:
-        in_room = await self._event_auth_handler.check_host_in_room(room_id, origin)
-        if not in_room:
-            raise AuthError(403, "Host not in room.")
+        await self._event_auth_handler.assert_host_in_room(room_id, origin)
 
         # Synapse asks for 100 events per backfill request. Do not allow more.
         limit = min(limit, 100)
@@ -1198,21 +1196,17 @@ class FederationHandler:
             event_id, allow_none=True, allow_rejected=True
         )
 
-        if event:
-            in_room = await self._event_auth_handler.check_host_in_room(
-                event.room_id, origin
-            )
-            if not in_room:
-                raise AuthError(403, "Host not in room.")
-
-            events = await filter_events_for_server(
-                self._storage_controllers, origin, [event]
-            )
-            event = events[0]
-            return event
-        else:
+        if not event:
             return None
 
+        await self._event_auth_handler.assert_host_in_room(event.room_id, origin)
+
+        events = await filter_events_for_server(
+            self._storage_controllers, origin, [event]
+        )
+        event = events[0]
+        return event
+
     async def on_get_missing_events(
         self,
         origin: str,
@@ -1221,9 +1215,7 @@ class FederationHandler:
         latest_events: List[str],
         limit: int,
     ) -> List[EventBase]:
-        in_room = await self._event_auth_handler.check_host_in_room(room_id, origin)
-        if not in_room:
-            raise AuthError(403, "Host not in room.")
+        await self._event_auth_handler.assert_host_in_room(room_id, origin)
 
         # Only allow up to 20 events to be retrieved per request.
         limit = min(limit, 20)
@@ -1257,7 +1249,7 @@ class FederationHandler:
             "state_key": target_user_id,
         }
 
-        if await self._event_auth_handler.check_host_in_room(room_id, self.hs.hostname):
+        if await self._event_auth_handler.is_host_in_room(room_id, self.hs.hostname):
             room_version_obj = await self.store.get_room_version(room_id)
             builder = self.event_builder_factory.for_room_version(
                 room_version_obj, event_dict
diff --git a/synapse/handlers/federation_event.py b/synapse/handlers/federation_event.py
index efcdb84057..2d7cde7506 100644
--- a/synapse/handlers/federation_event.py
+++ b/synapse/handlers/federation_event.py
@@ -238,7 +238,7 @@ class FederationEventHandler:
         #
         # Note that if we were never in the room then we would have already
         # dropped the event, since we wouldn't know the room version.
-        is_in_room = await self._event_auth_handler.check_host_in_room(
+        is_in_room = await self._event_auth_handler.is_host_in_room(
             room_id, self._server_name
         )
         if not is_in_room:
diff --git a/synapse/handlers/receipts.py b/synapse/handlers/receipts.py
index d2bdb9c8be..afaf3261df 100644
--- a/synapse/handlers/receipts.py
+++ b/synapse/handlers/receipts.py
@@ -70,7 +70,7 @@ class ReceiptsHandler:
             # If we're not in the room just ditch the event entirely. This is
             # probably an old server that has come back and thinks we're still in
             # the room (or we've been rejoined to the room by a state reset).
-            is_in_room = await self.event_auth_handler.check_host_in_room(
+            is_in_room = await self.event_auth_handler.is_host_in_room(
                 room_id, self.server_name
             )
             if not is_in_room:
diff --git a/synapse/handlers/room_summary.py b/synapse/handlers/room_summary.py
index ebd445adca..8d08625237 100644
--- a/synapse/handlers/room_summary.py
+++ b/synapse/handlers/room_summary.py
@@ -609,7 +609,7 @@ class RoomSummaryHandler:
         # If this is a request over federation, check if the host is in the room or
         # has a user who could join the room.
         elif origin:
-            if await self._event_auth_handler.check_host_in_room(
+            if await self._event_auth_handler.is_host_in_room(
                 room_id, origin
             ) or await self._store.is_host_invited(room_id, origin):
                 return True
@@ -624,9 +624,7 @@ class RoomSummaryHandler:
                     await self._event_auth_handler.get_rooms_that_allow_join(state_ids)
                 )
                 for space_id in allowed_rooms:
-                    if await self._event_auth_handler.check_host_in_room(
-                        space_id, origin
-                    ):
+                    if await self._event_auth_handler.is_host_in_room(space_id, origin):
                         return True
 
         logger.info(
diff --git a/synapse/handlers/typing.py b/synapse/handlers/typing.py
index a4cd8b8f0c..0d8466af11 100644
--- a/synapse/handlers/typing.py
+++ b/synapse/handlers/typing.py
@@ -340,7 +340,7 @@ class TypingWriterHandler(FollowerTypingHandler):
         # If we're not in the room just ditch the event entirely. This is
         # probably an old server that has come back and thinks we're still in
         # the room (or we've been rejoined to the room by a state reset).
-        is_in_room = await self.event_auth_handler.check_host_in_room(
+        is_in_room = await self.event_auth_handler.is_host_in_room(
             room_id, self.server_name
         )
         if not is_in_room:
diff --git a/tests/handlers/test_typing.py b/tests/handlers/test_typing.py
index 8adba29d7f..1a247f12e8 100644
--- a/tests/handlers/test_typing.py
+++ b/tests/handlers/test_typing.py
@@ -129,7 +129,7 @@ class TypingNotificationsTestCase(unittest.HomeserverTestCase):
         async def check_host_in_room(room_id: str, server_name: str) -> bool:
             return room_id == ROOM_ID
 
-        hs.get_event_auth_handler().check_host_in_room = check_host_in_room
+        hs.get_event_auth_handler().is_host_in_room = check_host_in_room
 
         async def get_current_hosts_in_room(room_id: str):
             return {member.domain for member in self.room_members}