CVE-2021-29602: TensorFlow-GPU DepthwiseConv Operator Division by Zero Vulnerability

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. TensorFlow-GPU is a variant of TensorFlow that leverages the power of GPUs for accelerated computation. The DepthwiseConv operator is used in TensorFlow-GPU to perform depthwise separable convolution, a common operation in convolutional neural networks. This operator is vulnerable to a division by zero error when the fourth dimension of the input tensor is set to zero.

Vulnerability Detail

The vulnerability in the DepthwiseConv operator arises from a lack of proper input validation. An attacker can exploit this vulnerability by crafting a malicious model that sets the fourth dimension of the input tensor to zero. When the operator attempts to perform the convolution operation, a division by zero error occurs, leading to a program crash or potential code execution. This vulnerability can be leveraged to disrupt the execution of TensorFlow-GPU models or potentially execute arbitrary code in the context of the affected application.

pip install --upgrade "tensorflow-gpu==2.12.0"

Workaround

As a temporary workaround, users can ensure that the fourth dimension of the input tensor is always set to a non-zero value. This can be achieved by adding a validation step in the model creation process to check the dimensions of the input tensor and raise an error if the fourth dimension is zero. Here is an example of how to implement this workaround in Python:

import tensorflow as tf

def create_model(input_shape):
    if input_shape[3] == 0:
        raise ValueError('Invalid input shape: fourth dimension cannot be zero')
    # Rest of the model creation code

Conclusion

The TensorFlow-GPU DepthwiseConv operator division by zero vulnerability can be exploited by crafting a malicious model with a zero-valued fourth dimension of the input tensor. This can lead to a division by zero error and potential program crashes or code execution. It is crucial for users of TensorFlow-GPU to update to the fixed versions mentioned above or apply the provided workaround to mitigate the vulnerability. For further assistance or support in vulnerability remediation, please contact DevNack.com.

See every vulnerability recorded for tensorflow-gpu →

FAQ