CVE-2021-29612: TensorFlow-GPU - Heap Buffer Overflow in BandedTriangularSolve

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 the power of GPUs for accelerated computation. It is widely used in various industries and research domains for training and inference of deep learning models.

Vulnerability Detail

A heap buffer overflow vulnerability exists in the Eigen implementation of tf.raw_ops.BandedTriangularSolve in TensorFlow-GPU. The vulnerability arises due to the lack of proper input validation in the implementation. The affected code fails to validate that the two input tensors are not empty, leading to a heap buffer overflow when processing the input. An attacker can exploit this vulnerability by providing specially crafted input tensors, potentially resulting in arbitrary code execution or a denial of service condition. The vulnerability affects versions of TensorFlow-GPU ranging from 0 to 2.1.4, 2.2.0 to 2.2.3, 2.3.0 to 2.3.3, and 2.4.0 to 2.4.2.

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

Workaround

As a workaround, users can temporarily avoid the vulnerability by ensuring that the input tensors provided to tf.raw_ops.BandedTriangularSolve are not empty. This can be achieved by adding appropriate checks in the calling code to validate the input tensors before passing them to the vulnerable function. Here is an example of how to perform the input validation:

import tensorflow as tf

def banded_triangular_solve_wrapper(input_tensor1, input_tensor2):
    if input_tensor1.size == 0 or input_tensor2.size == 0:
        raise ValueError('Input tensors cannot be empty')
    return tf.raw_ops.BandedTriangularSolve(input_tensor1, input_tensor2)

By incorporating this validation check, the vulnerability can be mitigated until a proper fix is applied.

Conclusion

The heap buffer overflow vulnerability in TensorFlow-GPU's tf.raw_ops.BandedTriangularSolve can be mitigated by updating to version 2.5.0 or later. In the meantime, users can apply a temporary workaround by validating the input tensors. For further assistance or support in vulnerability remediation, please contact DevNack.com.

See every vulnerability recorded for tensorflow-gpu →

FAQ