Issue #7406: Fix some occurrences of potential signed overflow in int
arithmetic.
diff --git a/Objects/intobject.c b/Objects/intobject.c
index dce569a..43dedf2 100644
--- a/Objects/intobject.c
+++ b/Objects/intobject.c
@@ -461,7 +461,8 @@
register long a, b, x;
CONVERT_TO_LONG(v, a);
CONVERT_TO_LONG(w, b);
- x = a + b;
+ /* casts in the line below avoid undefined behaviour on overflow */
+ x = (long)((unsigned long)a + b);
if ((x^a) >= 0 || (x^b) >= 0)
return PyInt_FromLong(x);
return PyLong_Type.tp_as_number->nb_add((PyObject *)v, (PyObject *)w);
@@ -473,7 +474,8 @@
register long a, b, x;
CONVERT_TO_LONG(v, a);
CONVERT_TO_LONG(w, b);
- x = a - b;
+ /* casts in the line below avoid undefined behaviour on overflow */
+ x = (long)((unsigned long)a - b);
if ((x^a) >= 0 || (x^~b) >= 0)
return PyInt_FromLong(x);
return PyLong_Type.tp_as_number->nb_subtract((PyObject *)v,
@@ -516,7 +518,8 @@
CONVERT_TO_LONG(v, a);
CONVERT_TO_LONG(w, b);
- longprod = a * b;
+ /* casts in the next line avoid undefined behaviour on overflow */
+ longprod = (long)((unsigned long)a * b);
doubleprod = (double)a * (double)b;
doubled_longprod = (double)longprod;