CVE-2022-35969: TensorFlow GPU Vulnerable to CHECK Fail in Conv2DBackpropInput

Background

TensorFlow is an open-source machine learning framework widely used for building and training deep learning models. It provides a comprehensive ecosystem for developing and deploying machine learning applications. The framework includes various operations, including convolutional layers, which are essential for processing image data. However, a vulnerability has been identified in the Conv2DBackpropInput operation, which can lead to denial of service under specific conditions.

Vulnerability Detail

The vulnerability arises from the Conv2DBackpropInput operation requiring input_sizes to be a 4-dimensional tensor. If this requirement is not met, it triggers a CHECK failure, which can be exploited to cause a denial of service. This means that an attacker could potentially crash the application by providing malformed input sizes, leading to unexpected behavior and service interruptions. The vulnerability has been assigned a CVSS score of 5.9, indicating a moderate severity level.

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

Workaround

As a workaround, ensure that the input_sizes parameter passed to the Conv2DBackpropInput operation is always a 4-dimensional tensor. For example, you can modify your code as follows:

import tensorflow as tf

strides = [1, 1, 1, 1] padding = 'SAME' use_cudnn_on_gpu = True explicit_paddings = [] dilations = [1, 1, 1, 1] input_sizes = tf.constant([1, 64, 64, 2], shape=[4], dtype=tf.int32) # Ensure this is 4D filter = tf.constant(0.159749106, shape=[3, 3, 2, 2], dtype=tf.float32) out_backprop = tf.constant(0, shape=[], dtype=tf.float32) tf.raw_ops.Conv2DBackpropInput(input_sizes=input_sizes, filter=filter, out_backprop=out_backprop, strides=strides, padding=padding, use_cudnn_on_gpu=use_cudnn_on_gpu, explicit_paddings=explicit_paddings, data_format='NHWC', dilations=dilations)

Conclusion

The vulnerability CVE-2022-35969 in TensorFlow's Conv2DBackpropInput operation poses a risk of denial of service due to improper input size handling. It is crucial for users to update to the patched versions or implement workarounds to mitigate the risk. Regular monitoring for vulnerabilities using tools like Vulert can help maintain the security of your applications.

See every vulnerability recorded for tensorflow-gpu →

FAQ