CVE-2021-29549: TensorFlow - Division by Zero Denial of Service

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 is widely used in various industries and research domains for tasks such as image recognition, natural language processing, and recommendation systems. However, a vulnerability has been discovered in TensorFlow that can lead to a division by zero error and denial of service.

Vulnerability Detail

The vulnerability exists in the tf.raw_ops.QuantizedBatchNormWithGlobalNormalization function of TensorFlow. The implementation of this function performs a modulo operation without validating that the divisor is not zero. By manipulating the input shapes, an attacker can trigger scenarios where the divisor becomes zero, causing a runtime division by zero error. This error can lead to a denial of service, as the affected operation will fail and potentially crash the application. The vulnerability has been assigned the identifier CVE-2021-29549.

pip install --upgrade "tensorflow==2.21.0"

Workaround

As a temporary workaround, you can modify the affected code to include a validation check for the divisor before performing the modulo operation. Here is an example of how you can add the validation check:

def quantized_batch_norm_with_global_normalization(divisor):
    if divisor == 0:
        # Handle the zero divisor case
        return
    # Perform the modulo operation
    result = tf.raw_ops.QuantizedBatchNormWithGlobalNormalization(divisor)
    return result

Conclusion

The division by zero denial of service vulnerability in TensorFlow can be mitigated by upgrading to the latest patched version. It is recommended to promptly update your TensorFlow installation to ensure the security and stability of your machine learning applications. For further assistance and support in vulnerability remediation, please contact DevNack.com.

See every vulnerability recorded for tensorflow →

FAQ