gh-156939: Detect buffer overflow in PyBytesWriter in debug mode

Lib/test/test_capi/test_bytes.py

@@ -1,5 +1,7 @@
+import textwrap
 import unittest
 from test.support import import_helper
+from test.support.script_helper import assert_python_failure
 
 _testlimitedcapi = import_helper.import_module('_testlimitedcapi')
 _testcapi = import_helper.import_module('_testcapi')
@@ -378,15 +380,6 @@ def test_format_i(self):
         writer.format_i(b'y=%i', 456)
         self.assertEqual(writer.finish(), self.result_type(b'x=123, y=456'))
 
-    def test_example_abc(self):
-        self.assertEqual(_testcapi.byteswriter_abc(), b'abc')
-
-    def test_example_resize(self):
-        self.assertEqual(_testcapi.byteswriter_resize(), b'Hello World')
-
-    def test_example_highlevel(self):
-        self.assertEqual(_testcapi.byteswriter_highlevel(), b'Hello World!')
-
 
 class BytesWriterTest(BaseWriterTest, unittest.TestCase):
     result_type = bytes
@@ -424,6 +417,34 @@ def test_singletons(self):
             writer.write_bytes(unused_text, len(unused_text))
             self.assertIs(writer.finish_with_size(1), singletons[ch])
 
+    def test_example_abc(self):
+        self.assertEqual(_testcapi.byteswriter_abc(), b'abc')
+
+    def test_example_resize(self):
+        self.assertEqual(_testcapi.byteswriter_resize(), b'Hello World')
+
+    def test_example_highlevel(self):
+        self.assertEqual(_testcapi.byteswriter_highlevel(), b'Hello World!')
+
+    def test_canary_byte(self):
+        small_buffer = _testcapi.PyBytesWriter_small_buffer
+        large_size = small_buffer * 10
+
+        # Test small buffer and large buffer
+        for size in (0, 3, large_size):
+            with self.subTest(size=size):
+                code = textwrap.dedent(f"""
+                    from test.support import SuppressCrashReport
+                    import _testcapi
+                    size = {size}
+                    data = b'x' * size
+                    with SuppressCrashReport():
+                        _testcapi.byteswriter_test_canary_byte(data)
+                """)
+                proc = assert_python_failure('-c', code)
+                self.assertIn(b'Buffer overflow detected in PyBytesWriter',
+                              proc.err)
+
 
 class ByteArrayWriterTest(BaseWriterTest, unittest.TestCase):
     result_type = bytearray

Misc/NEWS.d/next/C_API/2026-09-04-16-41-07.gh-issue-156939.bKaQuE.rst

@@ -0,0 +1,2 @@
+When Python is built in debug mode, :c:type:`PyBytesWriter` now detects
+buffer overflow.  Patch by Victor Stinner.

Modules/_testcapi/bytes.c

@@ -151,7 +151,7 @@ writer_write_bytes(PyObject *self_raw, PyObject *args)
         return NULL;
     }
 
-    char *bytes;
+    const char *bytes;
     Py_ssize_t unused_size, size;
     if (!PyArg_ParseTuple(args, "y#n", &bytes, &unused_size, &size)) {
         return NULL;
@@ -353,12 +353,45 @@ byteswriter_highlevel(PyObject *Py_UNUSED(module), PyObject *Py_UNUSED(args))
 }
 
 
+// Trigger a buffer overflow on purpose to test the canary byte feature
+// which detects buffer overflow
+static PyObject *
+byteswriter_test_canary_byte(PyObject *Py_UNUSED(module), PyObject *args)
+{
+    const char *str;
+    Py_ssize_t len;
+    if (!PyArg_ParseTuple(args, "s#", &str, &len)) {
+        return NULL;
+    }
+
+    PyBytesWriter *writer = PyBytesWriter_Create(0);
+    if (writer == NULL) {
+        goto error;
+    }
+    if (PyBytesWriter_Grow(writer, len) < 0) {
+        goto error;
+    }
+    char *data = PyBytesWriter_GetData(writer);
+    if (len) {
+        memcpy(data, str, len);
+    }
+    data[len] = '#';  // Overflow!
+
+    return PyBytesWriter_Finish(writer);
+
+error:
+    PyBytesWriter_Discard(writer);
+    return NULL;
+}
+
+
 static PyMethodDef test_methods[] = {
     {"bytes_resize", bytes_resize, METH_VARARGS},
     {"bytes_join", bytes_join, METH_VARARGS},
     {"byteswriter_abc", byteswriter_abc, METH_NOARGS},
     {"byteswriter_resize", byteswriter_resize, METH_NOARGS},
     {"byteswriter_highlevel", byteswriter_highlevel, METH_NOARGS},
+    {"byteswriter_test_canary_byte", byteswriter_test_canary_byte, METH_VARARGS},
     {NULL},
 };
 

Objects/bytesobject.c

@@ -2995,6 +2995,8 @@ _PyBytes_FromList(PyObject *x)
     }
     char *str = PyBytesWriter_GetData(writer);
     size = _PyBytesWriter_GetAllocated(writer);
+    str = _PyBytesWriter_ResizeAndUpdatePointer(writer, size, str);
+    assert(str != NULL);
 
     for (Py_ssize_t i = 0; i < PyList_GET_SIZE(x); i++) {
         PyObject *item = _PyList_GetItemRef((PyListObject *)x, i);
@@ -3018,6 +3020,8 @@ _PyBytes_FromList(PyObject *x)
                 goto error;
             }
             size = _PyBytesWriter_GetAllocated(writer);
+            str = _PyBytesWriter_ResizeAndUpdatePointer(writer, size, str);
+            assert(str != NULL);
         }
         *str++ = (char) value;
     }
@@ -3077,6 +3081,8 @@ _PyBytes_FromIterator(PyObject *it, PyObject *x)
     }
     char *str = PyBytesWriter_GetData(writer);
     size = _PyBytesWriter_GetAllocated(writer);
+    str = _PyBytesWriter_ResizeAndUpdatePointer(writer, size, str);
+    assert(str != NULL);
 
     /* Run the iterator to exhaustion */
     for (i = 0; ; i++) {
@@ -3111,6 +3117,8 @@ _PyBytes_FromIterator(PyObject *it, PyObject *x)
                 goto error;
             }
             size = _PyBytesWriter_GetAllocated(writer);
+            str = _PyBytesWriter_ResizeAndUpdatePointer(writer, size, str);
+            assert(str != NULL);
         }
         *str++ = (char) value;
     }
@@ -3589,6 +3597,10 @@ _PyBytes_RepeatBuffer(char* dest, Py_ssize_t len_dest,
 
 // --- PyBytesWriter API -----------------------------------------------------
 
+// Use a value different than NUL (0) to be able to detect overflow writing
+// one extra NUL byte which is a common error.
+#define PyBytesWriter_CANARY_BYTE PYMEM_DEADBYTE
+
 static inline char*
 byteswriter_data(PyBytesWriter *writer)
 {
@@ -3611,6 +3623,49 @@ byteswriter_allocated(PyBytesWriter *writer)
 }
 
 
+#ifdef Py_DEBUG
+static int
+byteswriter_has_canary_byte(PyBytesWriter *writer)
+{
+    if (writer->obj == NULL) {
+        return (writer->size < byteswriter_allocated(writer));
+    }
+    else {
+        // For bytes or bytearray, use the trailing NUL byte
+        // as the canary byte.
+        return 1;
+    }
+}
+
+
+static void
+byteswriter_check_canary_byte(PyBytesWriter *writer)
+{
+    assert(byteswriter_has_canary_byte(writer));
+
+    const unsigned char *data = (const unsigned char*)byteswriter_data(writer);
+    unsigned char canary = data[writer->size];
+    if (canary != PyBytesWriter_CANARY_BYTE) {
+        _Py_FatalErrorFormat(__func__,
+                             "Buffer overflow detected in PyBytesWriter %p: "
+                             "one byte written after the buffer "
+                             "(at position %zd)",
+                             writer, writer->size);
+    }
+}
+
+
+static void
+byteswriter_write_canary_byte(PyBytesWriter *writer)
+{
+    assert(byteswriter_has_canary_byte(writer));
+
+    unsigned char *data = (unsigned char*)byteswriter_data(writer);
+    data[writer->size] = PyBytesWriter_CANARY_BYTE;
+}
+#endif
+
+
 #ifdef MS_WINDOWS
    /* On Windows, overallocate by 50% is the best factor */
 #  define OVERALLOCATE_FACTOR 2
@@ -3625,6 +3680,17 @@ byteswriter_resize(PyBytesWriter *writer, Py_ssize_t size, int resize)
     assert(size >= 0);
 
     Py_ssize_t old_allocated = byteswriter_allocated(writer);
+#ifdef Py_DEBUG
+    if (writer->obj == NULL) {
+        // If the small_buffer is used, reserve one byte for the canary byte.
+        if (size <= (old_allocated - 1)) {
+            return 0;
+        }
+    }
+    else
+    // bytes and bytearray always allocates one extra byte for a trailing
+    // NUL byte: use this byte as the canary byte.
+#endif
     if (size <= old_allocated) {
         return 0;
     }
@@ -3713,9 +3779,12 @@ byteswriter_create(Py_ssize_t size, int use_bytearray)
         }
         writer->size = size;
     }
+
 #ifdef Py_DEBUG
     memset(byteswriter_data(writer), 0xff, byteswriter_allocated(writer));
+    byteswriter_write_canary_byte(writer);
 #endif
+
     return writer;
 }
 
@@ -3747,6 +3816,17 @@ PyBytesWriter_Discard(PyBytesWriter *writer)
 PyObject*
 PyBytesWriter_FinishWithSize(PyBytesWriter *writer, Py_ssize_t size)
 {
+#ifdef Py_DEBUG
+    byteswriter_check_canary_byte(writer);
+    if (writer->obj != NULL) {
+        // byteswriter_write_canary_byte() can override the trailing NUL byte.
+        // So reset the trailing NUL byte to NUL.
+        Py_ssize_t allocated = byteswriter_allocated(writer);
+        char *data = byteswriter_data(writer);
+        data[allocated] = '\0';
+    }
+#endif
+
     PyObject *result;
     if (size == 0) {
         result = bytes_get_empty();
@@ -3846,6 +3926,9 @@ PyBytesWriter_Resize(PyBytesWriter *writer, Py_ssize_t size)
         return -1;
     }
     writer->size = size;
+#ifdef Py_DEBUG
+    byteswriter_write_canary_byte(writer);
+#endif
     return 0;
 }
 
@@ -3879,6 +3962,9 @@ PyBytesWriter_Grow(PyBytesWriter *writer, Py_ssize_t size)
         return -1;
     }
     writer->size = size;
+#ifdef Py_DEBUG
+    byteswriter_write_canary_byte(writer);
+#endif
     return 0;
 }