CVE-2021-37674: Incomplete validation in MaxPoolGrad of TensorFlow

Background

TensorFlow is an open-source machine learning framework developed by Google. It provides a comprehensive ecosystem of tools, libraries, and community resources for building and deploying machine learning models. TensorFlow includes various operations and functions for performing mathematical computations and building neural networks. One of these operations is MaxPoolGrad, which is used for computing the gradients of the max pooling operation. However, a vulnerability has been discovered in the implementation of MaxPoolGrad that can lead to a denial of service (DoS) attack.

Vulnerability Detail

The vulnerability (CVE-2021-37674) in MaxPoolGrad arises from incomplete validation of the orig_input and orig_output tensors. An attacker can exploit this vulnerability by providing malicious input to the MaxPoolGrad operation, triggering a segmentation fault and causing a DoS condition. The issue was initially addressed in a previous fix (CVE-2021-29579), but the fix was incomplete. The impact of this vulnerability is considered moderate, with a CVSS score of 5.5. It is important to note that this vulnerability affects multiple versions of TensorFlow.

pip install --upgrade "tensorflow==2.21.0"

Workaround

As a temporary workaround, users can avoid using the MaxPoolGrad operation with untrusted or malicious input. By carefully validating and sanitizing the input tensors before passing them to MaxPoolGrad, the risk of triggering the vulnerability can be mitigated. Here is an example of how to validate the input tensors:

import tensorflow as tf

orig_input = tf.constant([], shape=[3, 0, 0, 2], dtype=tf.float32)
orig_output = tf.constant([], shape=[3, 0, 0, 2], dtype=tf.float32)
grad = tf.constant([], shape=[3, 0, 0, 2], dtype=tf.float32)

if orig_input.shape[1] > 0 and orig_output.shape[1] > 0:
    tf.raw_ops.MaxPoolGrad(
        orig_input=orig_input,
        orig_output=orig_output,
        grad=grad,
        ksize=[1, 16, 16, 1],
        strides=[1, 16, 18, 1],
        padding="EXPLICIT",
        explicit_paddings=[0, 0, 14, 3, 15, 5, 0, 0]
    )

Conclusion

The vulnerability in MaxPoolGrad of TensorFlow (CVE-2021-37674) can lead to a denial of service (DoS) condition. It is important for users to update their TensorFlow installations to the patched versions and follow the recommended security practices. For further assistance or support in vulnerability remediation, please contact DevNack.com.

See every vulnerability recorded for tensorflow →

FAQ