[3.10] gh-145986: Avoid unbound C recursion in `conv_content_model` in `pyexpat.c` (CVE 2026-4224) (GH-145987) (#146002)
* [3.10] gh-145986: Avoid unbound C recursion in `conv_content_model` in `pyexpat.c` (CVE 2026-4224) (GH-145987)
Fix C stack overflow (CVE-2026-4224) when an Expat parser
with a registered `ElementDeclHandler` parses inline DTD
containing deeply nested content model.
---------
(cherry picked from commit eb0e8be3a7e11b87d198a2c3af1ed0eccf532768)
(cherry picked from commit e5caf45faac74b0ed869e3336420cffd3510ce6e)
Co-authored-by: Stan Ulbrych <89152624+StanFromIreland@users.noreply.github.com>
Co-authored-by: Bénédikt Tran <10796600+picnixz@users.noreply.github.com>
* Update Misc/NEWS.d/next/Security/2026-03-14-17-31-39.gh-issue-145986.ifSSr8.rst
---------
Co-authored-by: Bénédikt Tran <10796600+picnixz@users.noreply.github.com>
diff --git a/Lib/test/test_pyexpat.py b/Lib/test/test_pyexpat.py
index 54be696..1024e20 100644
--- a/Lib/test/test_pyexpat.py
+++ b/Lib/test/test_pyexpat.py
@@ -16,7 +16,7 @@
from xml.parsers import expat
from xml.parsers.expat import errors
-from test.support import import_helper
+from test.support import import_helper, infinite_recursion
from test.support import sortdict
@@ -648,6 +648,24 @@ def test_change_size_2(self):
parser.Parse(xml2, True)
self.assertEqual(self.n, 4)
+class ElementDeclHandlerTest(unittest.TestCase):
+ def test_deeply_nested_content_model(self):
+ # This should raise a RecursionError and not crash.
+ # See https://github.com/python/cpython/issues/145986.
+ N = 500_000
+ data = (
+ b'<!DOCTYPE root [\n<!ELEMENT root '
+ + b'(a, ' * N + b'a' + b')' * N
+ + b'>\n]>\n<root/>\n'
+ )
+
+ parser = expat.ParserCreate()
+ parser.ElementDeclHandler = lambda _1, _2: None
+ with infinite_recursion():
+ with self.assertRaises(RecursionError):
+ parser.Parse(data)
+
+
class MalformedInputTest(unittest.TestCase):
def test1(self):
xml = b"\0\r\n"
diff --git a/Misc/NEWS.d/next/Security/2026-03-14-17-31-39.gh-issue-145986.ifSSr8.rst b/Misc/NEWS.d/next/Security/2026-03-14-17-31-39.gh-issue-145986.ifSSr8.rst
new file mode 100644
index 0000000..cb9dbad
--- /dev/null
+++ b/Misc/NEWS.d/next/Security/2026-03-14-17-31-39.gh-issue-145986.ifSSr8.rst
@@ -0,0 +1,4 @@
+:mod:`xml.parsers.expat`: Fixed a crash caused by unbounded C recursion when
+converting deeply nested XML content models with
+:meth:`~xml.parsers.expat.xmlparser.ElementDeclHandler`.
+This addresses `CVE-2026-4224 <https://www.cve.org/CVERecord?id=CVE-2026-4224>`_.
diff --git a/Modules/pyexpat.c b/Modules/pyexpat.c
index 8838c32..a679ed0 100644
--- a/Modules/pyexpat.c
+++ b/Modules/pyexpat.c
@@ -574,6 +574,10 @@ static PyObject *
conv_content_model(XML_Content * const model,
PyObject *(*conv_string)(const XML_Char *))
{
+ if (Py_EnterRecursiveCall(" in conv_content_model")) {
+ return NULL;
+ }
+
PyObject *result = NULL;
PyObject *children = PyTuple_New(model->numchildren);
int i;
@@ -585,7 +589,7 @@ conv_content_model(XML_Content * const model,
conv_string);
if (child == NULL) {
Py_XDECREF(children);
- return NULL;
+ goto done;
}
PyTuple_SET_ITEM(children, i, child);
}
@@ -593,6 +597,8 @@ conv_content_model(XML_Content * const model,
model->type, model->quant,
conv_string,model->name, children);
}
+done:
+ Py_LeaveRecursiveCall();
return result;
}