CVE-2021-29567: Lack of validation in SparseDenseCwiseMul in TensorFlow

Background

TensorFlow is an open-source machine learning framework that provides a comprehensive ecosystem of tools, libraries, and community resources. It is widely used for developing and deploying machine learning models. The SparseDenseCwiseMul function in TensorFlow is responsible for performing element-wise multiplication between a sparse tensor and a dense tensor. This vulnerability affects versions of TensorFlow prior to 2.1.4, 2.2.3, 2.3.3, and 2.4.2.

Vulnerability Detail

The vulnerability arises from a lack of validation in the SparseDenseCwiseMul function. Attackers can exploit this vulnerability to trigger denial of service by causing CHECK-fails or accessing memory outside the bounds of heap allocated data. The implementation of the function only validates the rank of the input arguments, but does not enforce constraints between dimensions. This allows attackers to abuse the function and potentially cause program termination or write to memory outside the allocated tensor buffers.

pip install --upgrade "tensorflow==2.21.0"

Workaround

As a workaround, users can validate the input arguments manually before calling the SparseDenseCwiseMul function. This can be done by checking the dimensions of the input tensors and ensuring they meet the required constraints. Here is an example of how to perform the validation:

import tensorflow as tf

indices = tf.constant([], shape=[10, 0], dtype=tf.int64)
values = tf.constant([], shape=[0], dtype=tf.int64)
shape = tf.constant([0, 0], shape=[2], dtype=tf.int64)
dense = tf.constant([], shape=[0], dtype=tf.int64)

if indices.shape[1] != values.shape[0] or indices.shape[0] != shape[0] or shape[1] != dense.shape[0]:
    raise ValueError('Invalid dimensions for input tensors')

# Call the patched version of SparseDenseCwiseMul
# ...

Note: This workaround is not a substitute for applying the official patch.

Conclusion

The lack of validation in the SparseDenseCwiseMul function in TensorFlow can be exploited by attackers to trigger denial of service or access memory outside the allocated tensor buffers. It is crucial to update to the patched version of TensorFlow or apply the fix on the affected versions to mitigate this vulnerability. For further support on vulnerability remediation, please contact DevNack.com.

See every vulnerability recorded for tensorflow →

FAQ