[Python-checkins] bpo-1621: Avoid signed integer overflow in set_table_resize() (GH-9059) (GH-9198)

[Victor Stinner]

https://github.com/python/cpython/commit/6665802549006eb50c1a68c3489ee3aaf81d0c8e
commit: 6665802549006eb50c1a68c3489ee3aaf81d0c8e
branch: 3.7
author: Miss Islington (bot) <31488909+miss-islington at users.noreply.github.com>
committer: Victor Stinner <vstinner at redhat.com>
date: 2018-10-20T00:50:34+02:00
summary:

bpo-1621: Avoid signed integer overflow in set_table_resize() (GH-9059) (GH-9198)

Address a C undefined behavior signed integer overflow issue in set object table resizing.  Our -fwrapv compiler flag and practical reasons why sets are unlikely to get this large should mean this was never an issue but it was incorrect code that generates code analysis warnings.

Co-authored-by: Sergey Fedoseev <fedoseev.sergey at gmail.com>

files:
A Misc/NEWS.d/next/Core and Builtins/2018-09-11-15-19-37.bpo-1621.7o19yG.rst
M Objects/setobject.c

diff --git a/Misc/NEWS.d/next/Core and Builtins/2018-09-11-15-19-37.bpo-1621.7o19yG.rst b/Misc/NEWS.d/next/Core and Builtins/2018-09-11-15-19-37.bpo-1621.7o19yG.rst
new file mode 100644
index 000000000000..4047ff3bfe86
--- /dev/null
+++ b/Misc/NEWS.d/next/Core and Builtins/2018-09-11-15-19-37.bpo-1621.7o19yG.rst	
@@ -0,0 +1,2 @@
+Do not assume signed integer overflow behavior (C undefined behavior) when
+performing set hash table resizing.
diff --git a/Objects/setobject.c b/Objects/setobject.c
index ce35aa2a0cdb..357e48ca3ce0 100644
--- a/Objects/setobject.c
+++ b/Objects/setobject.c
@@ -302,7 +302,6 @@ actually be smaller than the old one.
 static int
 set_table_resize(PySetObject *so, Py_ssize_t minused)
 {
-    Py_ssize_t newsize;
     setentry *oldtable, *newtable, *entry;
     Py_ssize_t oldmask = so->mask;
     size_t newmask;
@@ -313,13 +312,9 @@ set_table_resize(PySetObject *so, Py_ssize_t minused)
 
     /* Find the smallest table size > minused. */
     /* XXX speed-up with intrinsics */
-    for (newsize = PySet_MINSIZE;
-         newsize <= minused && newsize > 0;
-         newsize <<= 1)
-        ;
-    if (newsize <= 0) {
-        PyErr_NoMemory();
-        return -1;
+    size_t newsize = PySet_MINSIZE;
+    while (newsize <= (size_t)minused) {
+        newsize <<= 1; // The largest possible value is PY_SSIZE_T_MAX + 1.
     }
 
     /* Get space for a new table. */