CVE-2021-29565: 
Python vulnerability analysis and mitigation

TensorFlow, an end-to-end open source platform for machine learning, was found to contain a null pointer dereference vulnerability (CVE-2021-29565) in the implementation of tf.raw_ops.SparseFillEmptyRows. The vulnerability was discovered by Yakun Zhang and Ying Wang of Baidu X-Team and disclosed on May 14, 2021. The issue affects multiple versions of TensorFlow including versions up to 2.1.4, 2.2.0-2.2.3, 2.3.0-2.3.3, and 2.4.0-2.4.2 (GitHub Advisory).

The vulnerability stems from missing validation in the implementation of tf.raw_ops.SparseFillEmptyRows. When the dense_shape tensor is empty, dense_shape_t.vec<>() causes a null pointer dereference in the operation's implementation. The issue was originally marked under a TODO comment in the code. The vulnerability has been assigned a CVSS v3.1 base score of 5.5 (MEDIUM) by NIST with a vector of CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H, while GitHub assessed it as LOW with a score of 2.5 (NVD).

The vulnerability allows an attacker to trigger a null pointer dereference in the TensorFlow library, which could lead to application crashes. The impact is primarily limited to availability issues, with no direct impact on confidentiality or integrity of the system (GitHub Advisory).

The vulnerability can be exploited by providing an empty dense_shape tensor to the tf.raw_ops.SparseFillEmptyRows operation. A proof-of-concept exploit involves using specific TensorFlow commands with empty tensors (GitHub Advisory).

The vulnerability has been patched in TensorFlow 2.5.0. The fix has also been backported to TensorFlow versions 2.4.2, 2.3.3, 2.2.3, and 2.1.4. Users are advised to upgrade to these patched versions. The fix includes adding validation to check if the dense shape is empty (GitHub Advisory).