Cache literal sync filter validation (#17186)

The sliding sync proxy (amongst other things) use literal json blobs as
filters, and repeatedly validating them takes a bunch of CPU.

2 files changed, 14 insertions, 1 deletions

diff --git a/changelog.d/17186.misc b/changelog.d/17186.misc
new file mode 100644
index 0000000000..358209d5a0
--- /dev/null
+++ b/changelog.d/17186.misc
@@ -0,0 +1 @@
+Cache literal sync filter validation for performance.
diff --git a/synapse/rest/client/sync.py b/synapse/rest/client/sync.py
index 2b103ca6a8..d19aaf0e22 100644
--- a/synapse/rest/client/sync.py
+++ b/synapse/rest/client/sync.py
@@ -47,6 +47,7 @@ from synapse.http.site import SynapseRequest
 from synapse.logging.opentracing import trace_with_opname
 from synapse.types import JsonDict, Requester, StreamToken
 from synapse.util import json_decoder
+from synapse.util.caches.lrucache import LruCache
 
 from ._base import client_patterns, set_timeline_upper_limit
 
@@ -110,6 +111,11 @@ class SyncRestServlet(RestServlet):
         self._msc2654_enabled = hs.config.experimental.msc2654_enabled
         self._msc3773_enabled = hs.config.experimental.msc3773_enabled
 
+        self._json_filter_cache: LruCache[str, bool] = LruCache(
+            max_size=1000,
+            cache_name="sync_valid_filter",
+        )
+
     async def on_GET(self, request: SynapseRequest) -> Tuple[int, JsonDict]:
         # This will always be set by the time Twisted calls us.
         assert request.args is not None
@@ -177,7 +183,13 @@ class SyncRestServlet(RestServlet):
                 filter_object = json_decoder.decode(filter_id)
             except Exception:
                 raise SynapseError(400, "Invalid filter JSON", errcode=Codes.NOT_JSON)
-            self.filtering.check_valid_filter(filter_object)
+
+            # We cache the validation, as this can get quite expensive if people use
+            # a literal json blob as a query param.
+            if not self._json_filter_cache.get(filter_id):
+                self.filtering.check_valid_filter(filter_object)
+                self._json_filter_cache[filter_id] = True
+
             set_timeline_upper_limit(
                 filter_object, self.hs.config.server.filter_timeline_limit
             )