Avoid throwing exception in ClosingTHPObjectPtr (pytorch#109758)

Previously, if ClosingTHPObjectPtr was destructed because we
were unwinding the stack from an exception, we would attempt to call
close() which just isn't going to work.  Two fixes:

1. Detect if we're unwinding due to a Python error, and don't try
   to do more Python stuff if so.

2. If close() fails somehow, write an unraisable exception, don't
   try to throw because that will terminate if you're in an
   exception.

Signed-off-by: Edward Z. Yang <[email protected]>

Pull Request resolved: pytorch#109758
Approved by: https://github.com/jansel

torch/csrc/dynamo/python_compiled_autograd.cpp

@@ -290,8 +290,15 @@ static PyObject* call_end_capture(PyObject* self, const variable_list& inputs) {
 struct ClosingTHPObjectPtr : public THPObjectPtr {
   ClosingTHPObjectPtr(PyObject* o) : THPObjectPtr(o) {}
   ~ClosingTHPObjectPtr() {
+    if (PyErr_Occurred()) {
+      // do nothing, do not attempt to close
+      return;
+    }
     static PyObject* method_name = PyUnicode_InternFromString("close");
-    check(PyObject_CallMethodNoArgs(get(), method_name));
+    if (PyObject_CallMethodNoArgs(get(), method_name) == nullptr) {
+      PyErr_WriteUnraisable(get());
+      PyErr_Clear();
+    }
   }
 };