CVE-2021-29581: TensorFlow-gpu - Denial of Service via Segmentation Faults

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 domains, including research, industry, and academia.

Vulnerability Detail

The vulnerability (CVE-2021-29581) exists in the tf.raw_ops.CTCBeamSearchDecoder function of TensorFlow-gpu. It arises due to a lack of validation, allowing an attacker to trigger a denial of service (DoS) attack by causing segmentation faults. The vulnerable implementation fails to detect cases when the input tensor is empty, leading to an attempt to read data from a null buffer. This can result in crashes and instability of the TensorFlow-gpu application. The issue has been identified and fixed in TensorFlow 2.5.0. However, earlier versions, including TensorFlow 2.4.2, TensorFlow 2.3.3, TensorFlow 2.2.3, and TensorFlow 2.1.4, are also affected.

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

Workaround

As a temporary workaround, users can avoid the vulnerability by implementing input validation in their own code before invoking the tf.raw_ops.CTCBeamSearchDecoder function. This can be done by checking if the input tensor is empty or null before passing it to the function. Here's an example of how to perform this validation in Python:

import tensorflow as tf

def decode_ctc_beam_search(input_tensor):
    if input_tensor is None or len(input_tensor) == 0:
        raise ValueError('Input tensor is empty or null')
    else:
        return tf.raw_ops.CTCBeamSearchDecoder(input_tensor)

By incorporating this validation step, the application can prevent the vulnerability from being exploited.

Conclusion

In conclusion, the vulnerability (CVE-2021-29581) in TensorFlow-gpu's tf.raw_ops.CTCBeamSearchDecoder function can be exploited to trigger denial of service attacks. It is crucial to update TensorFlow-gpu to version 2.5.0 or later to fix the vulnerability. In the meantime, implementing input validation in the application code can serve as a temporary workaround. For further assistance or support in vulnerability remediation, please contact DevNack.com.

See every vulnerability recorded for tensorflow-gpu →

FAQ