CVE-2021-29611: Incomplete validation in TensorFlow SparseReshape

Background

TensorFlow is an open-source machine learning framework widely used for developing and training deep learning models. It provides a comprehensive set of tools and libraries for building and deploying machine learning applications. TensorFlow includes a module called SparseReshape, which is responsible for reshaping sparse tensors. Sparse tensors are a specialized data structure used in certain machine learning tasks to efficiently represent sparse data. However, a vulnerability has been discovered in the implementation of SparseReshape in TensorFlow versions 2.3.0 to 2.3.3 and 2.4.0 to 2.4.2.

Vulnerability Detail

The vulnerability, identified as CVE-2021-29611, stems from incomplete validation in the SparseReshape module. This flaw allows an attacker to trigger a denial of service (DoS) condition by exploiting a CHECK-failure. By providing invalid input arguments to the SparseReshape function, an attacker can cause the TensorFlow application to crash. The issue lies in the lack of validation for the input arguments, which should specify a valid sparse tensor. This vulnerability has a CVSS score of 3.6, indicating a low severity level.

pip install --upgrade "tensorflow==2.21.0"

Workaround

As a temporary workaround, users can avoid using the SparseReshape function in their TensorFlow applications. By removing or replacing the usage of SparseReshape, the vulnerability can be mitigated. Here is an example of how to modify the code to avoid using SparseReshape:

import tensorflow as tf

input_indices = tf.constant(41, shape=[1, 1], dtype=tf.int64)
input_shape = tf.zeros([11], dtype=tf.int64)
new_shape = tf.zeros([1], dtype=tf.int64)

# Instead of using SparseReshape
# tf.raw_ops.SparseReshape(input_indices=input_indices,
#     input_shape=input_shape,
#     new_shape=new_shape)

# Use an alternative method
reshaped_tensor = tf.sparse.reshape(tf.sparse.SparseTensor(indices=input_indices, values=[], dense_shape=input_shape), new_shape)

Conclusion

In conclusion, the vulnerability CVE-2021-29611 in TensorFlow's SparseReshape module allows for a denial of service (DoS) attack by exploiting a CHECK-failure. Users are advised to update their TensorFlow installation to one of the patched versions or apply the provided workaround. For further assistance or support in vulnerability remediation, please contact DevNack.com.

See every vulnerability recorded for tensorflow →

FAQ