CVE-2021-29593: TensorFlow CPU - Division by Zero in BatchToSpaceNd Operator

Background

TensorFlow is an end-to-end open source platform for machine learning. It provides a comprehensive ecosystem of tools, libraries, and resources for developing and deploying machine learning models. The TensorFlow CPU package, available on PyPI, is a CPU-only version of TensorFlow that allows users to run TensorFlow on systems without GPU support.

Vulnerability Detail

The vulnerability (CVE-2021-29593) exists in the implementation of the BatchToSpaceNd TFLite operator in TensorFlow CPU. The issue arises when a crafted model is created with a dimension of the block input set to 0, resulting in a division by zero error. This vulnerability can be exploited by an attacker to cause a denial-of-service condition or potentially execute arbitrary code on the affected system. The vulnerability was identified in the source code of the batch_to_space_nd.cc file. The fix for this vulnerability has been included in TensorFlow 2.5.0, and patches have been backported to TensorFlow 2.4.2, TensorFlow 2.3.3, TensorFlow 2.2.3, and TensorFlow 2.1.4.

pip install --upgrade "tensorflow-cpu==2.21.0"

Workaround

As a temporary workaround, users can avoid the vulnerability by ensuring that the block input of the BatchToSpaceNd operator does not have any dimensions set to 0. This can be achieved by validating the input dimensions before passing them to the operator. Here is an example code snippet demonstrating the workaround:

import tensorflow as tf

# Validate block dimensions
block = [2, 0]
if 0 in block:
    raise ValueError('Invalid block dimensions')

# Perform BatchToSpaceNd operation
output = tf.nn.batch_to_space_nd(input, block, crops)

By validating the block dimensions, the vulnerability can be mitigated.

Conclusion

The vulnerability (CVE-2021-29593) in TensorFlow CPU's BatchToSpaceNd operator can be exploited by an attacker to cause a division by zero error. This can lead to a denial-of-service condition or potential execution of arbitrary code. Users are advised to upgrade to TensorFlow CPU version 2.5.0 or apply the respective patches for their installed versions. A temporary workaround is available by validating the block dimensions to avoid setting any dimensions to 0. For further assistance on vulnerability remediation, please contact DevNack.com.

See every vulnerability recorded for tensorflow-cpu →

FAQ