gh-123089: Make weakref.WeakSet safe against concurrent mutations while it is being iterated (#123279)

* Make `weakref.WeakSet` safe against concurrent mutations while it is being iterated.

`_IterationGuard` is no longer used for `WeakSet`, it now relies on copying the underlying set which is an atomic operation while iterating so that it can be modified by other threads.
This commit is contained in:
Kumar Aditya 2024-08-27 18:34:03 +05:30 committed by GitHub
parent 6754566a51
commit 03f5abf15a
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 11 additions and 43 deletions

View File

@ -36,41 +36,26 @@ class _IterationGuard:
class WeakSet:
def __init__(self, data=None):
self.data = set()
def _remove(item, selfref=ref(self)):
self = selfref()
if self is not None:
if self._iterating:
self._pending_removals.append(item)
else:
self.data.discard(item)
self.data.discard(item)
self._remove = _remove
# A list of keys to be removed
self._pending_removals = []
self._iterating = set()
if data is not None:
self.update(data)
def _commit_removals(self):
pop = self._pending_removals.pop
discard = self.data.discard
while True:
try:
item = pop()
except IndexError:
return
discard(item)
def __iter__(self):
with _IterationGuard(self):
for itemref in self.data:
item = itemref()
if item is not None:
# Caveat: the iterator will keep a strong reference to
# `item` until it is resumed or closed.
yield item
for itemref in self.data.copy():
item = itemref()
if item is not None:
# Caveat: the iterator will keep a strong reference to
# `item` until it is resumed or closed.
yield item
def __len__(self):
return len(self.data) - len(self._pending_removals)
return len(self.data)
def __contains__(self, item):
try:
@ -83,21 +68,15 @@ class WeakSet:
return self.__class__, (list(self),), self.__getstate__()
def add(self, item):
if self._pending_removals:
self._commit_removals()
self.data.add(ref(item, self._remove))
def clear(self):
if self._pending_removals:
self._commit_removals()
self.data.clear()
def copy(self):
return self.__class__(self)
def pop(self):
if self._pending_removals:
self._commit_removals()
while True:
try:
itemref = self.data.pop()
@ -108,18 +87,12 @@ class WeakSet:
return item
def remove(self, item):
if self._pending_removals:
self._commit_removals()
self.data.remove(ref(item))
def discard(self, item):
if self._pending_removals:
self._commit_removals()
self.data.discard(ref(item))
def update(self, other):
if self._pending_removals:
self._commit_removals()
for element in other:
self.add(element)
@ -136,8 +109,6 @@ class WeakSet:
def difference_update(self, other):
self.__isub__(other)
def __isub__(self, other):
if self._pending_removals:
self._commit_removals()
if self is other:
self.data.clear()
else:
@ -151,8 +122,6 @@ class WeakSet:
def intersection_update(self, other):
self.__iand__(other)
def __iand__(self, other):
if self._pending_removals:
self._commit_removals()
self.data.intersection_update(ref(item) for item in other)
return self
@ -184,8 +153,6 @@ class WeakSet:
def symmetric_difference_update(self, other):
self.__ixor__(other)
def __ixor__(self, other):
if self._pending_removals:
self._commit_removals()
if self is other:
self.data.clear()
else:

View File

@ -0,0 +1 @@
Make :class:`weakref.WeakSet` safe against concurrent mutations while it is being iterated. Patch by Kumar Aditya.