3.11: _PyUnicode_Equal call sites lack error handling
Nadie ha tomado este issue todavía.
- Lenguaje dominante
- Python
- Estrellas
- 77.2k
- Forks
- 36k
- Merge medio
- 1 d 9 h
- PR fusionados (30 d)
- 558
Descripción
(Found while looking at https://github.com/python/cpython/issues/98783)
In https://github.com/python/cpython/commit/81c72044a181dbbfbf689d7a977d0d99090f26a8, several uses of _PyUnicode_EqualToASCIIId, which cannot fail, were replaced with _PyUnicode_Equal, which can fail: it calls PyUnicode_READY() in Python 3.11, and returns -1 on failure.
In Python 3.12, this is not a problem: _PyUnicode_Equal cannot fail, since it does not need to call PyUnicode_READY() since the wstr APIs were removed in https://github.com/python/cpython/commit/f9c9354a7a173eaca2aa19e667b5cf12167b7fed. But it is theoretically a problem in 3.11.
In 3.11, git grep "_PyUnicode_Equal(" turns up the following:
Include/cpython/unicodeobject.h:PyAPI_FUNC(int) _PyUnicode_Equal(PyObject *, PyObject *);
Modules/_pickle.c: use_newobj_ex = _PyUnicode_Equal(name, &_Py_ID(__newobj_ex__));
Modules/_pickle.c: use_newobj = _PyUnicode_Equal(name, &_Py_ID(__newobj__));
Objects/longobject.c: else if (_PyUnicode_Equal(byteorder, &_Py_ID(little)))
Objects/longobject.c: else if (_PyUnicode_Equal(byteorder, &_Py_ID(big)))
Objects/longobject.c: else if (_PyUnicode_Equal(byteorder, &_Py_ID(little)))
Objects/longobject.c: else if (_PyUnicode_Equal(byteorder, &_Py_ID(big)))
Objects/typeobject.c: if (mod != NULL && !_PyUnicode_Equal(mod, &_Py_ID(builtins)))
Objects/typeobject.c: if (_PyUnicode_Equal(name, &_Py_ID(__dict__))) {
Objects/typeobject.c: if (_PyUnicode_Equal(name, &_Py_ID(__weakref__))) {
Objects/typeobject.c: if ((ctx->add_dict && _PyUnicode_Equal(slot, &_Py_ID(__dict__))) ||
Objects/typeobject.c: (ctx->add_weak && _PyUnicode_Equal(slot, &_Py_ID(__weakref__))))
Objects/typeobject.c: if (!_PyUnicode_Equal(slot, &_Py_ID(__qualname__)) &&
Objects/typeobject.c: !_PyUnicode_Equal(slot, &_Py_ID(__classcell__)))
Objects/typeobject.c: if (mod != NULL && !_PyUnicode_Equal(mod, &_Py_ID(builtins)))
Objects/typeobject.c: _PyUnicode_Equal(name, &_Py_ID(__class__)))
Objects/typeobject.c: if (_PyUnicode_Equal(name, &_Py_ID(__class__))) {
Objects/unicodeobject.c:_PyUnicode_Equal(PyObject *str1, PyObject *str2)
Python/ceval.c: int res = _PyUnicode_Equal(left, right);
Python/errors.c: if (!_PyUnicode_Equal(modulename, &_Py_ID(builtins)) &&
Python/errors.c: !_PyUnicode_Equal(modulename, &_Py_ID(__main__))) {
Python/pythonrun.c: if (!_PyUnicode_Equal(modulename, &_Py_ID(builtins)) &&
Python/pythonrun.c: !_PyUnicode_Equal(modulename, &_Py_ID(__main__)))
Broken down:
- _pickle.c
- Could the result of
_PyObject_LookupAttr(callable, &_Py_ID(__name__), &name)be unready?
- Could the result of
- longobject.c
- Safe because argument clinic calls
PyUnicode_READY()
- Safe because argument clinic calls
- typeobject.c
ctx->slotscould be an arbitrary tuple of potentially-unready strings, so there should be some call toPyUnicode_READY()at or beforetype_new_visit_slots.
- ceval.c
- Already has the error checking (though it could be removed in 3.12!)
- errors.c and pythonrun.c:
- Arbitrary result of
PyObject_GetAttr(exc_type, &_Py_ID(__module__));might not be_READY()?
- Arbitrary result of
The good news is that this would be hard to run into in practice: it requires both using the old deprecated wstr APIs and running into a memory error during PyUnicode_READY().
cc @ericsnowcurrently @methane
Guía de contribución
Primeros pasos
- Lee el issue completo y luego la guía de contribución del proyecto.
- Comenta en el issue que vas a ocuparte — evita que dos personas hagan lo mismo.
- Haz un fork del repositorio y trabaja en una rama.
- Abre un pull request que haga referencia al número del issue.
Línea de trabajo
Comienza con git grep para _PyUnicode_Equal en la rama Python 3.11 y, a continuación, lee los sitios de llamada enumerados en Modules/_pickle.c, Objects/longobject.c, Objects/typeobject.c, Python/ceval.c, Python/errors.c y Python/pythonrun.c. Rastrea si cada entrada puede ser un objeto Unicode no preparado y cómo se propagan los fallos de PyUnicode_READY(). La tarea está terminada cuando los sitios de llamada afectados tienen un manejo seguro de errores o una garantía de preparación verificada.
Escrito por el modelo de indexación a partir del texto del issue.
Evaluación
- Stack tecnológico
- c, python
- Área
- backend
- Tipo de issue
- Error
- Dificultad
- 4/5
- Tiempo estimado
- 3-5 días
- Estado de actividad
- Estancado
- Claridad
- Bastante claro
- Aptitud para principiantes
- 35/100