Fix for SF bug 528132 (Armin Rigo): classmethod().__get__() segfault

The proper fix is not quite what was submitted; it's really better to
take the class of the object passed rather than calling PyMethod_New
with NULL pointer args, because that can then cause other core dumps
later.

I also added a testcase for the fix to classmethods() in test_descr.py.

I've already applied this to the 2.2 branch.
diff --git a/Lib/test/test_descr.py b/Lib/test/test_descr.py
index 1188e1d..ce28cce 100644
--- a/Lib/test/test_descr.py
+++ b/Lib/test/test_descr.py
@@ -1208,6 +1208,11 @@
     vereq(d.goo(1), (D, 1))
     vereq(d.foo(1), (d, 1))
     vereq(D.foo(d, 1), (d, 1))
+    # Test for a specific crash (SF bug 528132)
+    def f(cls, arg): return (cls, arg)
+    ff = classmethod(f)
+    vereq(ff.__get__(0, int)(42), (int, 42))
+    vereq(ff.__get__(0)(42), (int, 42))
 
 def staticmethods():
     if verbose: print "Testing static methods..."
diff --git a/Objects/funcobject.c b/Objects/funcobject.c
index f24cbcf..426b8f4 100644
--- a/Objects/funcobject.c
+++ b/Objects/funcobject.c
@@ -474,6 +474,8 @@
 				"uninitialized classmethod object");
 		return NULL;
 	}
+	if (type == NULL)
+		type = (PyObject *)(obj->ob_type);
  	return PyMethod_New(cm->cm_callable,
 			    type, (PyObject *)(type->ob_type));
 }