CVE-2021-29589: TensorFlow - Division by Zero Error in GatherNd TFLite 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 GatherNd TFLite operator is a reference implementation in TensorFlow that allows gathering elements from multiple tensors based on indices. This operator is used in various machine learning tasks and workflows.

Vulnerability Detail

The reference implementation of the GatherNd TFLite operator in TensorFlow is vulnerable to a division by zero error. This vulnerability arises when the params input tensor is empty, resulting in a zero value for at least one dimension of the params_shape.Dims(.). When this occurs, a division operation is performed, leading to a division by zero error. An attacker can exploit this vulnerability by crafting a malicious model that triggers the division by zero error. The impact of this vulnerability can range from denial of service to potential code execution.

pip install --upgrade "tensorflow==2.21.0"

Workaround

As a workaround, you can validate the input tensors before using the GatherNd TFLite operator to ensure that the params input tensor is not empty. This can be done by checking the shape of the params tensor and ensuring that it has at least one element in each dimension. Here is an example code snippet in Python:

import tensorflow as tf

params = ...  # your input tensor

if tf.size(params) == 0:
    # Handle empty tensor case
    ...
else:
    # Use GatherNd TFLite operator
    ...

Conclusion

In conclusion, the GatherNd TFLite operator in TensorFlow is vulnerable to a division by zero error when the params input tensor is empty. This vulnerability can have serious consequences, including denial of service and potential code execution. It is crucial to update your TensorFlow installation to a fixed version or apply the available backported fixes to mitigate this vulnerability. For further support and assistance in vulnerability remediation, please contact DevNack.com.

See every vulnerability recorded for tensorflow →

FAQ