CVE-2021-29588: TensorFlow - Division by Zero Error in TransposeConv 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 ecosystem of tools, libraries, and community resources that enable researchers and developers to build and deploy machine learning models. TensorFlow includes a TFLite operator called TransposeConv which is used for transposed convolution operations. This operator is optimized for performance. However, a vulnerability has been discovered in the optimized implementation of the TransposeConv operator, which can lead to a division by zero error.

Vulnerability Detail

The vulnerability in the TransposeConv TFLite operator arises due to a lack of validation of the stride_{h,w} arguments. An attacker can craft a malicious model with stride_{h,w} values set to 0, triggering a division by zero error during the execution of the operator. This vulnerability can potentially be exploited to cause a denial of service (DoS) condition or other unexpected behavior in TensorFlow applications that utilize the TransposeConv operator. The vulnerability affects versions of TensorFlow ranging from 0 to 2.4.2.

pip install --upgrade "tensorflow==2.21.0"

Workaround

As a temporary workaround, users can manually validate the stride_{h,w} arguments before calling the TransposeConv operator. By ensuring that the stride_{h,w} values are non-zero, the division by zero error can be avoided. Here is an example of how to validate the arguments in Python:

stride_h = ...  # Get the stride_h value
stride_w = ...  # Get the stride_w value

if stride_h == 0 or stride_w == 0:
    # Handle the error condition
    ...
else:
    # Call the TransposeConv operator with the validated arguments
    ...

Conclusion

The vulnerability in the optimized implementation of the TransposeConv TFLite operator in TensorFlow can be mitigated by updating to TensorFlow 2.5.0 or applying the relevant patch. Users can also manually validate the stride_{h,w} arguments as a temporary workaround. It is important to regularly update TensorFlow to benefit from security fixes and improvements. For further support on vulnerability remediation, please contact DevNack.com.

See every vulnerability recorded for tensorflow →

FAQ