CVE-2021-29531: Vulnerability in tf.raw_ops.EncodePng in TensorFlow

Background

TensorFlow is an open-source machine learning framework developed by Google. It provides a comprehensive ecosystem of tools, libraries, and resources for building and deploying machine learning models. The tf.raw_ops.EncodePng function in TensorFlow is responsible for encoding images in PNG format. This vulnerability affects versions 0 to 2.1.4, 2.2.0 to 2.2.3, 2.3.0 to 2.3.3, and 2.4.0 to 2.4.2 of the TensorFlow package.

Vulnerability Detail

The vulnerability, identified as CVE-2021-29531, allows an attacker to trigger a CHECK fail in PNG encoding by providing an empty input tensor as the pixel data. This can be achieved by passing an empty matrix to the tf.raw_ops.EncodePng function. The implementation of the function only checks for the total number of pixels in the image, without validating the content of the tensor. As a result, if the tensor is empty, the associated buffer becomes NULL, triggering a CHECK_NOTNULL assertion in the png::WriteImageToBuffer function. This assertion causes the program to abort, leading to a denial of service (DoS) attack. The vulnerability has a CVSS score of 2.5, indicating a low severity level.

pip install --upgrade "tensorflow==2.21.0"

Workaround

As a temporary workaround, users can ensure that the input tensor passed to the tf.raw_ops.EncodePng function is not empty. This can be done by checking the dimensions of the tensor and providing valid pixel data. Here is an example in Python:

import tensorflow as tf

image = tf.zeros([1, 1, 3])
image = tf.cast(image, dtype=tf.uint8)
tf.raw_ops.EncodePng(image=image)

In this example, a non-empty tensor with dimensions [1, 1, 3] is created and passed to the EncodePng function.

Conclusion

The vulnerability in tf.raw_ops.EncodePng in TensorFlow allows an attacker to trigger a CHECK fail in PNG encoding, resulting in a denial of service (DoS) attack. It has been assigned CVE-2021-29531 and has a CVSS score of 2.5. Users are advised to update their TensorFlow installations to the patched versions once they become available. In the meantime, a temporary workaround is provided to mitigate the vulnerability. For further support, please contact the TensorFlow team at DevNack.com.

See every vulnerability recorded for tensorflow →

FAQ