CVE-2021-29587: TensorFlow CPU - SpaceToDepth TFLite Operator Division by Zero

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 includes the SpaceToDepth TFLite operator, which is used for rearranging data in a tensor. This operator is vulnerable to a division by zero issue in the Prepare step, where it fails to check for a zero value before performing division. This vulnerability can be exploited by an attacker by crafting a malicious model that sets the block_size parameter to zero. The issue has been identified in TensorFlow versions 2.1.4 to 2.4.3, as well as in the tensorflow-cpu package on PyPI.

Vulnerability Detail

The vulnerability in the SpaceToDepth TFLite operator allows an attacker to trigger a division by zero error by providing a crafted model with a block_size parameter set to zero. This can lead to a crash or undefined behavior in the TensorFlow runtime. The impact of this vulnerability depends on the context in which TensorFlow is used. In some scenarios, it may allow an attacker to cause a denial of service (DoS) by crashing the application or potentially execute arbitrary code.

pip install --upgrade "tensorflow-cpu==2.21.0"

Workaround

As a temporary workaround, users can manually check the value of the block_size parameter before passing it to the SpaceToDepth TFLite operator. This can be done by adding a conditional check in the code that sets the block_size value, ensuring it is not zero. Here is an example of how to perform the check in Python:

block_size = ...  # Set the block size

if block_size != 0:
    # Call the SpaceToDepth TFLite operator
    ...
else:
    # Handle the case when block_size is zero
    ...

Conclusion

The division by zero vulnerability in the SpaceToDepth TFLite operator of TensorFlow can be exploited by an attacker to cause a crash or potentially execute arbitrary code. Users are advised to update their TensorFlow installations to the latest patched version or upgrade to tensorflow-cpu package version 2.4.3 or later. For further assistance or support in vulnerability remediation, please contact DevNack.com.

See every vulnerability recorded for tensorflow-cpu →

FAQ