CVE-2021-29520: TensorFlow - Missing Validation in Conv3DBackprop Operations

Background

TensorFlow is an end-to-end open source platform for machine learning. It provides a comprehensive ecosystem of tools, libraries, and community resources that enable researchers and developers to build and deploy machine learning models. TensorFlow is widely used in various domains, including computer vision, natural language processing, and deep learning. This vulnerability affects the TensorFlow package, specifically the tf.raw_ops.Conv3DBackprop* operations.

Vulnerability Detail

The vulnerability arises from missing validation between arguments to tf.raw_ops.Conv3DBackprop* operations in TensorFlow. The implementation assumes that the input, filter_sizes, and out_backprop tensors have the same shape, as they are accessed in parallel. However, if these tensors have different shapes, it can lead to heap buffer overflows. An attacker could exploit this vulnerability to execute arbitrary code or cause a denial of service. The issue has been identified and fixed in TensorFlow 2.5.0, and the fix has also been backported to earlier versions within the supported range (2.4.2, 2.3.3, 2.2.3, and 2.1.4).

pip install --upgrade "tensorflow==2.21.0"

Workaround

If upgrading to a fixed version is not immediately feasible, a workaround can be implemented by manually validating the shapes of the input, filter_sizes, and out_backprop tensors before invoking the tf.raw_ops.Conv3DBackprop* operations. Here's an example of how the validation can be performed:

import tensorflow as tf

def conv3d_backprop_workaround(input, filter_sizes, out_backprop):
    if input.shape != filter_sizes.shape or input.shape != out_backprop.shape:
        raise ValueError('Input shapes do not match')
    # Perform the Conv3DBackprop operation
    result = tf.raw_ops.Conv3DBackprop(...)
    return result

Conclusion

The vulnerability in TensorFlow, identified as CVE-2021-29520, can lead to heap buffer overflows if the input, filter_sizes, and out_backprop tensors have different shapes. This can be exploited by an attacker to execute arbitrary code or cause a denial of service. It is crucial to upgrade to TensorFlow 2.5.0 or apply the provided patches to mitigate the vulnerability. For further support on vulnerability remediation, you can contact DevNack.com.

See every vulnerability recorded for tensorflow →

FAQ