CVE-2021-29553: TensorFlow-GPU - Out-of-bounds Read in tf.raw_ops.QuantizeAndDequantizeV3

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-GPU is a variant of TensorFlow that leverages GPU acceleration for faster computation. It is widely used in various industries and research domains for deep learning tasks.

Vulnerability Detail

The vulnerability (CVE-2021-29553) exists in the tf.raw_ops.QuantizeAndDequantizeV3 function of TensorFlow-GPU. An attacker can exploit this vulnerability to read data outside the bounds of a heap-allocated buffer. The issue arises due to the lack of validation for the user-supplied axis attribute before using it to index the array backing the input argument. By providing a malicious axis value, an attacker can access sensitive information or trigger a crash, potentially leading to a denial-of-service (DoS) scenario. The vulnerability affects versions of TensorFlow-GPU from 0 to 2.1.4, 2.2.0 to 2.2.3, 2.3.0 to 2.3.3, and 2.4.0 to 2.4.2.

pip install --upgrade "tensorflow-gpu==2.12.0"

Workaround

As a temporary workaround, users can avoid using the tf.raw_ops.QuantizeAndDequantizeV3 function or ensure that the axis attribute is properly validated before passing it to the function. Here's an example of how to validate the axis attribute:

import tensorflow as tf

def safe_quantize_and_dequantize_v3(input, axis):
    if axis < 0 or axis >= len(input.shape):
        raise ValueError('Invalid axis value')
    return tf.raw_ops.QuantizeAndDequantizeV3(input=input, axis=axis)

Conclusion

The vulnerability (CVE-2021-29553) in TensorFlow-GPU's tf.raw_ops.QuantizeAndDequantizeV3 function allows an attacker to read data outside the bounds of a heap-allocated buffer. It is crucial to update TensorFlow-GPU to version 2.5.0 or later to mitigate this vulnerability. Users can also implement a temporary workaround by validating the axis attribute before passing it to the vulnerable function. For further support on vulnerability remediation, please contact DevNack.com.

See every vulnerability recorded for tensorflow-gpu →

FAQ