Skip to content

`CHECK` failure in `TensorListReserve` via missing validation

Low
pak-laura published GHSA-v5xg-3q2c-c2r4 Sep 15, 2022

Package

pip tensorflow, tensorflow-cpu, tensorflow-gpu (pip)

Affected versions

< 2.10.0

Patched versions

2.7.4, 2.8.3, 2.9.2, 2.10.0

Description

Impact

In core/kernels/list_kernels.cc's TensorListReserve, num_elements is assumed to be a tensor of size 1. When a num_elements of more than 1 element is provided, then tf.raw_ops.TensorListReserve fails the CHECK_EQ in CheckIsAlignedAndSingleElement.

import tensorflow as tf

tf.raw_ops.TensorListReserve(element_shape=(1,1), num_elements=tf.constant([1,1], dtype=tf.int32), element_dtype=tf.int8)

Patches

We have patched the issue in GitHub commit b5f6fbfba76576202b72119897561e3bd4f179c7.

The fix will be included in TensorFlow 2.10.0. We will also cherrypick this commit on TensorFlow 2.9.1, TensorFlow 2.8.1, and TensorFlow 2.7.2, 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 Kang Hong Jin from Singapore Management University.

Severity

Low

CVE ID

CVE-2022-35960

Weaknesses

No CWEs