CVE-2021-29580: TensorFlow - FractionalMaxPoolGrad Denial of Service Vulnerability

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 developing and deploying machine learning models. TensorFlow is widely used in various industries and research domains.

Vulnerability Detail

The vulnerability (CVE-2021-29580) in TensorFlow is related to the implementation of tf.raw_ops.FractionalMaxPoolGrad. When one of the input tensors is empty, it triggers an undefined behavior. Additionally, the code is susceptible to a denial of service attack due to a CHECK condition that becomes false and aborts the process. The vulnerability arises from the lack of validation for empty input and output tensors, as well as the assumption that they are of the same rank. This vulnerability can lead to unexpected program behavior and potential denial of service attacks.

pip install --upgrade "tensorflow==2.21.0"

Workaround

As a temporary workaround, users can manually validate the input and output tensors before invoking tf.raw_ops.FractionalMaxPoolGrad. This can be done by checking if the tensors are empty and ensuring they have the same rank. Here is an example of how to perform the validation in Python:

import tensorflow as tf

def fractional_max_pool_grad(input, grad, pool_ratio, pseudo_random, overlapping):
    if input.size == 0 or grad.size == 0 or input.ndim != grad.ndim:
        raise ValueError('Invalid input or grad tensors')
    
    # Rest of the code
    

Conclusion

The vulnerability in TensorFlow's tf.raw_ops.FractionalMaxPoolGrad can lead to undefined behavior and denial of service attacks. It is crucial for users to update their TensorFlow installations to version 2.5.0 once it is released. In the meantime, the provided workaround can be used to mitigate the vulnerability. For further support and assistance in vulnerability remediation, please contact DevNack.com.

See every vulnerability recorded for tensorflow →

FAQ