[prev in list] [next in list] [prev in thread] [next in thread] 

List:       pykde
Subject:    Re: Python 3.11 and QObject.destroyed signals: _PyThreadState_PopFrame: Assertion `tstate->datastack
From:       Florian Bruhin <me () the-compiler ! org>
Date:       2022-07-13 10:39:33
Message-ID: 20220713103933.2ehm2ngvrunmuqad () aragog
[Download RAW message or body]


On Fri, Jul 08, 2022 at 05:30:27PM +0200, Florian Bruhin wrote:
> when testing my application with the current Python 3.11 beta, I've
> noticed a bug which seems to be triggered somehow by something PyQt
> does: https://github.com/python/cpython/issues/93252
> 
> This is the most trivial reproducer I could find:
> 
> from PyQt5.QtCore import QObject
> 
> def run():
> obj = QObject()
> for _ in range(202):
> obj.destroyed.connect(lambda: None)
> 
> run()
> 
> Which crashes CPython with:
> 
> python: Python/pystate.c:2201: _PyThreadState_PopFrame: Assertion \
> `tstate->datastack_top >= base' failed.

FWIW, we ended up finding a much simpler reproducer, not involving PyQt:

    def f():
        pass

    for _ in range(203):
        try:
            f(None)
        except:
            pass

Thus this happened when a stack frame can't be built up - Python never
cleaned it properly in that case. It's been fixed in 3.11.0b4:
https://github.com/python/cpython/pull/94708/commits/be41a146c6595200107cb3ece97809cb506dffba


I don't quite understand why/how the connection to QObject.destroyed
exhibits the same behavior, but in any case, the patch fixes it.

Florian

-- 
            me@the-compiler.org | https://www.qutebrowser.org 
       https://bruhin.software/ | https://github.com/sponsors/The-Compiler/
       GPG: 916E B0C8 FD55 A072 | https://the-compiler.org/pubkey.asc
             I love long mails! | https://email.is-not-s.ms/


["signature.asc" (application/pgp-signature)]

[prev in list] [next in list] [prev in thread] [next in thread]