CVE-2022-35991: 
Python vulnerability analysis and mitigation

A vulnerability was discovered in TensorFlow's TensorListScatter and TensorListScatterV2 functions that could lead to a denial of service condition. The vulnerability, tracked as CVE-2022-35991, was identified in TensorFlow versions prior to 2.10.0 and affects the pip packages tensorflow, tensorflow-cpu, and tensorflow-gpu. The issue was discovered by 刘力源 from the Information System & Security and Countermeasures Experiments Center at Beijing Institute of Technology (GitHub Advisory).

The vulnerability occurs when TensorListScatter and TensorListScatterV2 functions receive an element_shape parameter with a rank greater than one, resulting in a CHECK fail condition. This can be triggered by passing a tensor with shape (2, 2, 2) as the element_shape parameter to these functions. The issue was addressed by adding validation to ensure element_shape must be at most rank 1 (GitHub Commit).

When successfully exploited, this vulnerability can trigger a denial of service condition in applications using the affected TensorFlow functions. The severity of this vulnerability has been rated as Low (GitHub Advisory).

The vulnerability has been patched in TensorFlow versions 2.7.4, 2.8.3, 2.9.2, and 2.10.0. Users are advised to upgrade to these patched versions. The fix implements input validation to ensure element_shape parameter is at most rank 1 (GitHub Advisory).