René's URL Explorer Experiment

{"@context":"https://schema.org","@type":"DiscussionForumPosting","headline":"Data race between `gc.get_stats()` and a concurrent collection under free-threading","articleBody":"# Bug report\n\n### Bug description:\n\n`gc.get_stats()` (`gc_get_stats_impl`) plain-copies the per-generation stats structs \n`gcstate-\u003egeneration_stats-\u003e{young,old}.items[index]` without any lock or atomics,\n\nhttps://github.com/python/cpython/blob/9e863fab283eddca9c2a8f9d1ee30f4dc243e314/Modules/gcmodule.c#L377-L379\n\nwhile a collection on another thread writes those same `items[index]` fields (`collections++`, `collected += m`, ...) at the end of `gc_collect_main`,\n\nhttps://github.com/python/cpython/blob/9e863fab283eddca9c2a8f9d1ee30f4dc243e314/Python/gc_free_threading.c#L2284-L2292\n\nso `gc.get_stats()` races the collector.\n\nReproducer:\n```python\nimport gc\nfrom threading import Thread\n\ndef collector():\n    for _ in range(50000):\n        a = {}; a['self'] = a\n        gc.collect()\n\ndef reader():\n    for _ in range(50000):\n        gc.get_stats()\n\nthreads  = [Thread(target=collector) for _ in range(1)]\nthreads += [Thread(target=reader)    for _ in range(8)]\nfor t in threads: t.start()\nfor t in threads: t.join()\n```\n\nTSAN Report:\n```\n==================\nWARNING: ThreadSanitizer: data race (pid=2301505)\n  Read of size 8 at 0x7fffb41a0180 by thread T2:\n    #0 gc_get_stats_impl /cpython/Modules/gcmodule.c:379:16 \n    #1 gc_get_stats /cpython/Modules/clinic/gcmodule.c.h:456:12 \n    #2 cfunction_vectorcall_NOARGS /cpython/Objects/methodobject.c:508:24 \n    #3 _PyObject_VectorcallTstate /cpython/./Include/internal/pycore_call.h:144:11 \n    #4 PyObject_Vectorcall /cpython/Objects/call.c:327:12 \n    #5 _Py_VectorCallInstrumentation_StackRefSteal /cpython/Python/ceval.c:768:11 \n    #6 _PyEval_EvalFrameDefault /cpython/Python/generated_cases.c.h:1846:35 \n    ...\n\n  Previous write of size 8 at 0x7fffb41a0180 by thread T1:\n    #0 gc_collect_main /cpython/Python/gc_free_threading.c:2288:23 \n    #1 _PyGC_Collect /cpython/Python/gc_free_threading.c:2568:12 \n    #2 gc_collect_impl /cpython/Modules/gcmodule.c:93:12 \n    #3 gc_collect /cpython/Modules/clinic/gcmodule.c.h:143:21 \n    #4 _Py_BuiltinCallFastWithKeywords_StackRef /cpython/Python/ceval.c:841:11 \n    #5 _PyEval_EvalFrameDefault /cpython/Python/generated_cases.c.h:2508:35 \n    ...\n\nSUMMARY: ThreadSanitizer: data race /cpython/Modules/gcmodule.c:379:16 in gc_get_stats_impl\n==================\n```\n\n### CPython versions tested on:\n\nCPython main branch\n\n### Operating systems tested on:\n\nLinux\n\n\u003c!-- gh-linked-prs --\u003e\n### Linked PRs\n* gh-151766\n\u003c!-- /gh-linked-prs --\u003e\n","author":{"url":"https://github.com/Naserume","@type":"Person","name":"Naserume"},"datePublished":"2026-06-18T12:06:31.000Z","interactionStatistic":{"@type":"InteractionCounter","interactionType":"https://schema.org/CommentAction","userInteractionCount":1},"url":"https://github.com/151646/cpython/issues/151646"}