CVE-2021-29541: TensorFlow GPU - Null Pointer Dereference in StringNGrams

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-2021-29541) exists in the tf.raw_ops.StringNGrams function of TensorFlow GPU. It arises due to insufficient validation of the data_splits argument in the implementation of the function. As a result, when the output is computed to have a size of 0 or negative, the ngrams_data pointer becomes null. Subsequent writes to the output tensor then trigger a null pointer dereference. This vulnerability can be exploited by an attacker to cause a denial-of-service (DoS) condition or potentially execute arbitrary code.

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

Workaround

As a temporary workaround, users can avoid passing inputs to the tf.raw_ops.StringNGrams function that would result in an output of 0 or negative size. By ensuring the input data meets the necessary requirements, the null pointer dereference can be prevented. Here is an example of how to validate the input data before invoking the function:

import tensorflow as tf

def safe_string_ngrams(data_splits):
    if len(data_splits) <= 0:
        raise ValueError('Invalid input: data_splits must have at least one element')
    else:
        return tf.raw_ops.StringNGrams(data_splits=data_splits)

Conclusion

In conclusion, the vulnerability (CVE-2021-29541) in TensorFlow GPU's tf.raw_ops.StringNGrams function can lead to a null pointer dereference, potentially resulting in a denial-of-service (DoS) condition or arbitrary code execution. Users should upgrade to the fixed versions or apply the provided workaround to mitigate the risk. For further support on vulnerability remediation, you can contact DevNack.com.

See every vulnerability recorded for tensorflow-gpu →

FAQ