Skip to content

Free-threaded GC can abort during concurrent subinterpreter teardown #161751

Open

Description

@XuehaiPan

Crash report

OpTree CI hit this assertion on an unmodified Windows x64 free-threaded debug build of CPython 3.14.10 (b703b2c). An InterpreterPoolExecutor worker was destroying its interpreter. The native stack went from Py_EndInterpreter() through shutdown GC to mi_abandoned_visit_blocks() and validate_alive_bits().

CI backtrace: https://github.com/metaopt/optree/actions/runs/42811376193/job/129813367183#step:18:1

Python/gc_free_threading.c:1082: validate_alive_bits:
Assertion "!gc_is_alive(op)" failed: object should not be marked as alive yet

I suspect a race in the abandoned-segment scan introduced with the mimalloc 2.5.2 update. One interpreter can temporarily clear a segment’s bitmap bit, discover that the segment belongs to another interpreter, and restore the bit. The owning interpreter’s GC reads the bitmap before acquiring abandoned_visit_lock. If it reads during that interval, it can skip the segment. The same scan is present on main at 9d4a35c.

Allocator code: https://github.com/python/cpython/blob/v3.14.10/Objects/mimalloc/arena-abandon.c#L209-L269


Standard Library Reproducer

Starting from v3.14.10, add usleep(100); inside the foreign-segment branch of mi_arena_segment_clear_abandoned_at(), immediately before it restores the bitmap bit.

   if (segment->subproc != subproc) {
+    usleep(100);
     // it is from another sub-process, re-mark it and continue searching

Build with ./configure --with-pydebug --disable-gil and make. Save this as repro.py and run ./python.exe -I -X dev repro.py on macOS.

from concurrent import interpreters
from concurrent.futures import ThreadPoolExecutor
from threading import Barrier

workers = 8
barrier = Barrier(workers)


def run(_worker_id):
    for _ in range(10):
        interp = interpreters.create()
        barrier.wait()
        interp.exec('import gc\ngc.collect()')
        interp.close()


with ThreadPoolExecutor(workers) as pool:
    list(pool.map(run, range(workers)))

With the timing change, this reduced case aborted in validate_gc_objects() on all three runs. A longer standard-library stress test also reproduced the exact validate_alive_bits() assertion from Windows. On stock macOS debug builds, that longer test completed 340 rounds each on 3.14.9 and 3.14.10.

As a control, I kept the delay and moved lock acquisition ahead of the first bitmap read for full scans (previous->visit_all). The longer workload then completed a 10-round run. The precise thread interleaving in the original Windows crash is unknown.

CPython versions tested on:

3.14

Operating systems tested on:

macOS

Output from running 'python -VV' on the command line:

No response

Linked PRs

Activity

  1. added
    type-crashA hard crash of the interpreter, possibly with a core dump
    on Nov 10, 2026
  2. sergey-miryanov commented on Nov 10, 2026

    @sergey-miryanov
    Contributor
  3. self-assigned this
    on Nov 14, 2026
  4. added a commit that references this issue on Nov 14, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

Labels

interpreter-core(Objects, Python, Grammar, and Parser dirs)topic-free-threadingtype-crashA hard crash of the interpreter, possibly with a core dump

Projects

No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions