CVE-2021-29595: TensorFlow DepthToSpace TFLite 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 community resources that enable researchers and developers to build and deploy machine learning models. TensorFlow is widely used in various domains, including computer vision, natural language processing, and reinforcement learning. The DepthToSpace TFLite operator is a component of TensorFlow that rearranges data from depth into blocks of spatial data. This operator is used in many machine learning models to perform tasks such as upsampling and feature extraction.

Vulnerability Detail

The implementation of the DepthToSpace TFLite operator in TensorFlow is vulnerable to a division by zero error. The specific vulnerability occurs in the depth_to_space.cc file (https://github.com/tensorflow/tensorflow/blob/0d45ea1ca641b21b73bcf9c00e0179cda284e7e7/tensorflow/lite/kernels/depth_to_space.cc#L63-L69) where a division operation is performed without checking if the divisor, params->block_size, is zero. An attacker can exploit this vulnerability by crafting a malicious model that sets params->block_size to zero. When the model is executed, a division by zero error will occur, leading to a crash or potentially allowing the attacker to execute arbitrary code.

pip install --upgrade "tensorflow==2.21.0"

Workaround

As a temporary workaround, users can manually modify the affected code in the depth_to_space.cc file to include a check for a zero divisor before performing the division operation. Here is an example of the modified code:

if (params->block_size != 0) {
  // Perform division operation
  ...
} else {
  // Handle the case when block_size is zero
  ...
}

Conclusion

The DepthToSpace TFLite operator in TensorFlow is vulnerable to a division by zero error, which can be exploited by an attacker to crash the application or execute arbitrary code. The issue has been addressed in TensorFlow 2.5.0 and will also be patched in earlier versions. Users are advised to update their TensorFlow installations to the latest patched version and regularly check for updates to ensure the security of their machine learning models. For further support on vulnerability remediation, please contact DevNack.com.

See every vulnerability recorded for tensorflow →

FAQ