From affcc2cc3655531351048a4ad8ac67e22d1e398d Mon Sep 17 00:00:00 2001
From: V02460 <V02460@gmail.com>
Date: Thu, 7 Nov 2019 10:43:51 +0100
Subject: [PATCH] Fix LruCache callback deduplication (#6213)

---
 changelog.d/6213.bugfix            |  1 +
 synapse/util/caches/descriptors.py | 48 +++++++++++++++++++++++-------
 2 files changed, 38 insertions(+), 11 deletions(-)
 create mode 100644 changelog.d/6213.bugfix

diff --git a/changelog.d/6213.bugfix b/changelog.d/6213.bugfix
new file mode 100644
index 000000000..072264fba
--- /dev/null
+++ b/changelog.d/6213.bugfix
@@ -0,0 +1 @@
+Fix LruCache callback deduplication.
diff --git a/synapse/util/caches/descriptors.py b/synapse/util/caches/descriptors.py
index 0e8da27f5..84f5ae22c 100644
--- a/synapse/util/caches/descriptors.py
+++ b/synapse/util/caches/descriptors.py
@@ -17,8 +17,8 @@ import functools
 import inspect
 import logging
 import threading
-from collections import namedtuple
-from typing import Any, cast
+from typing import Any, Tuple, Union, cast
+from weakref import WeakValueDictionary
 
 from six import itervalues
 
@@ -38,6 +38,8 @@ from . import register_cache
 
 logger = logging.getLogger(__name__)
 
+CacheKey = Union[Tuple, Any]
+
 
 class _CachedFunction(Protocol):
     invalidate = None  # type: Any
@@ -430,7 +432,7 @@ class CacheDescriptor(_CacheDescriptorBase):
             # Add our own `cache_context` to argument list if the wrapped function
             # has asked for one
             if self.add_cache_context:
-                kwargs["cache_context"] = _CacheContext(cache, cache_key)
+                kwargs["cache_context"] = _CacheContext.get_instance(cache, cache_key)
 
             try:
                 cached_result_d = cache.get(cache_key, callback=invalidate_callback)
@@ -624,14 +626,38 @@ class CacheListDescriptor(_CacheDescriptorBase):
         return wrapped
 
 
-class _CacheContext(namedtuple("_CacheContext", ("cache", "key"))):
-    # We rely on _CacheContext implementing __eq__ and __hash__ sensibly,
-    # which namedtuple does for us (i.e. two _CacheContext are the same if
-    # their caches and keys match). This is important in particular to
-    # dedupe when we add callbacks to lru cache nodes, otherwise the number
-    # of callbacks would grow.
-    def invalidate(self):
-        self.cache.invalidate(self.key)
+class _CacheContext:
+    """Holds cache information from the cached function higher in the calling order.
+
+    Can be used to invalidate the higher level cache entry if something changes
+    on a lower level.
+    """
+
+    _cache_context_objects = (
+        WeakValueDictionary()
+    )  # type: WeakValueDictionary[Tuple[Cache, CacheKey], _CacheContext]
+
+    def __init__(self, cache, cache_key):  # type: (Cache, CacheKey) -> None
+        self._cache = cache
+        self._cache_key = cache_key
+
+    def invalidate(self):  # type: () -> None
+        """Invalidates the cache entry referred to by the context."""
+        self._cache.invalidate(self._cache_key)
+
+    @classmethod
+    def get_instance(cls, cache, cache_key):  # type: (Cache, CacheKey) -> _CacheContext
+        """Returns an instance constructed with the given arguments.
+
+        A new instance is only created if none already exists.
+        """
+
+        # We make sure there are no identical _CacheContext instances. This is
+        # important in particular to dedupe when we add callbacks to lru cache
+        # nodes, otherwise the number of callbacks would grow.
+        return cls._cache_context_objects.setdefault(
+            (cache, cache_key), cls(cache, cache_key)
+        )
 
 
 def cached(