CVE-2022-36000: TensorFlow GPU Null Dereference Vulnerability in Eig Function

Background

TensorFlow is an open-source machine learning framework that provides a comprehensive ecosystem for building and deploying machine learning models. It supports various tasks, including deep learning and numerical computation. TensorFlow GPU is a variant optimized for performance on NVIDIA GPUs. The framework is widely used in both research and production environments. However, it has been found to have a vulnerability that can lead to a denial of service when incorrect input is provided to the Eig function.

Vulnerability Detail

The vulnerability arises from the ability to feed an incorrect 'Tout' input to the Eig function in TensorFlow, which can lead to a null dereference. This results in a CHECK failure, potentially causing a denial of service attack. The issue can be exploited by providing a complex data type that is not compatible with the expected input, leading to application crashes and service interruptions.

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

Workaround

As a temporary workaround, avoid using the Eig function with incorrect 'Tout' inputs. Ensure that the input data types are compatible with the expected types to prevent triggering the vulnerability. Here is an example of how to correctly use the Eig function:

import tensorflow as tf
import numpy as np
arg_0 = tf.constant(value=np.random.random(size=(2, 2)), shape=(2, 2), dtype=tf.float32)
arg_1 = tf.float32  # Correct data type
arg_2 = True
arg_3 = ''
tf.raw_ops.Eig(input=arg_0, Tout=arg_1, compute_v=arg_2, name=arg_3)

Conclusion

The CVE-2022-36000 vulnerability in TensorFlow GPU poses a risk of denial of service due to incorrect input handling in the Eig function. Users are advised to update their packages to the latest patched versions to mitigate this risk. Additionally, using Vulert can help monitor for such vulnerabilities in real-time.

See every vulnerability recorded for tensorflow-gpu →

FAQ