CVE-2021-29549: Denial of Service in TensorFlow

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 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 denial of service (DoS) attack.

Vulnerability Detail

The vulnerability, identified as CVE-2021-29549, exists in the tf.raw_ops.QuantizedBatchNormWithGlobalNormalization function of TensorFlow. An attacker can exploit this vulnerability by triggering a runtime division by zero error. The issue arises due to the lack of validation for a divisor being zero in the implementation of the function. By manipulating input shapes, an attacker can cause the computation of vector_num_elements to be zero, leading to a division by zero error. This vulnerability can result in a DoS condition, rendering the affected TensorFlow instance unresponsive.

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

Workaround

As a temporary workaround, users can implement input validation to ensure that the divisor in the tf.raw_ops.QuantizedBatchNormWithGlobalNormalization function is not zero. By checking the divisor before performing the modulo operation, the vulnerability can be mitigated. Here is an example of how the input validation can be implemented in Python:

import tensorflow as tf

def safe_quantized_batch_norm_with_global_normalization(input, scale, offset, mean, variance, variance_epsilon, name=None):
    if tf.reduce_any(tf.equal(scale, 0.0)):
        raise ValueError('Scale divisor cannot be zero.')
    return tf.raw_ops.QuantizedBatchNormWithGlobalNormalization(input, scale, offset, mean, variance, variance_epsilon, name=name)

By using this modified function in place of the vulnerable one, the risk of a DoS attack can be mitigated.

Conclusion

In conclusion, a vulnerability has been identified in TensorFlow that can lead to a denial of service (DoS) attack. It is crucial to update TensorFlow to one of the fixed versions mentioned to mitigate the vulnerability. Additionally, implementing input validation as a temporary workaround can provide an extra layer of protection. For further support and assistance in vulnerability remediation, please contact DevNack.com.

See every vulnerability recorded for tensorflow-cpu →

FAQ