CVE-2021-29589: TensorFlow - Division by Zero Error in GatherNd TFLite Operator

Sponsored Habilume Habit tracker with a companion that grows with you Complete habits, earn Lume Points, and watch your Lume grow as you stay consistent. App Store Google Play

Background

TensorFlow is an end-to-end open source platform for machine learning. It provides a comprehensive set of tools, libraries, and resources for building and deploying machine learning models. The vulnerability affects the reference implementation of the GatherNd TFLite operator in TensorFlow. This operator is used for gathering elements from a tensor based on indices specified in another tensor. The vulnerability arises when the params input tensor is empty, resulting in a division by zero error. This vulnerability has been assigned the identifier CVE-2021-29589.

Vulnerability Detail

The vulnerability in the GatherNd TFLite operator allows an attacker to craft a malicious model that triggers a division by zero error. This can be achieved by providing an empty tensor as the params input, causing at least one dimension of params_shape.Dims(.) to be zero. The division by zero error can lead to various consequences, including crashes, denial of service, or potential information disclosure. The vulnerability has been classified as high severity with a CVSSv3 score of 8.8 (CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H).

Workaround

As a temporary workaround, users can check the dimensions of the params input tensor before passing it to the GatherNd TFLite operator. If any dimension is zero, the operator should not be invoked to avoid triggering the division by zero error. Here is an example code snippet demonstrating the workaround:

import tensorflow as tf

params = ...  # Get the `params` tensor
params_shape = tf.shape(params)
if 0 in params_shape:
    # Handle the empty tensor case
    ...
else:
    # Invoke the `GatherNd` TFLite operator
    ...

Conclusion

In conclusion, the vulnerability in the GatherNd TFLite operator of TensorFlow can be exploited by providing an empty tensor as the params input, resulting in a division by zero error. This vulnerability has been assigned the identifier CVE-2021-29589 and has a high severity rating. Users are advised to update their TensorFlow installations to the latest patched version and apply the provided workaround if necessary. For further support on vulnerability remediation, please contact DevNack.com.

See every vulnerability recorded for tensorflow →

FAQ