Overflow when computing `__sizeof__` of objects with underlying code objects

[federicovalenso]

Bug report

Bug description:

>>> def F(a,b):
...     x=a*b
...     frame = inspect.currentframe()
...     print(frame.__sizeof__())
...     return x
... 
>>> F(1,2)
184
>>> F.__code__ = F.__code__.replace(co_stacksize=1000)
>>> F(1,2)
8152
>>> F.__code__ = F.__code__.replace(co_stacksize=2147483646)
>>> F(1,2)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
SystemError: ../Modules/gcmodule.c:2313: bad argument to internal function

It happens due to overflow in frame_sizeof

CPython versions tested on:

3.11

Operating systems tested on:

Linux

Linked PRs

• gh-126119: fix some crashes in code objects if co_stacksize is absurdly large #126122

[rruuaanng]

If there are no comments, I might be able to fix it :)
Do you want to fix it? @federicovalenso

Edit
But in fact this cannot be fundamentally solved, maybe a warning or comment can be added to it.

[picnixz]

Thanks but I was already working on the patch (for future references: when I self-assign an issue, that means that I'm working on it unless stated otherwise).

[picnixz]

By the way: I labelled as a "type-crash" because I got a crash on my side in DEBUG mode:

python: Python/pystate.c:2907: _PyThreadState_PushFrame: Assertion `size < INT_MAX/sizeof(PyObject *)' failed.
Aborted (core dumped)

The reason why the release mode was raising a SystemError is likely due to the fact that this assertion was by-passed.

[federicovalenso]

@picnixz , I found some similar problems:

• frame.c
• ceval.c - 3.11, but not in main
• ceval.c - 3.11, but not in main

Should I create another issues? Or will these be duplicates in general?