CVE-2021-29578: Heap buffer overflow in `FractionalAvgPoolGrad`
Impact The implementation of tf.rawops.FractionalAvgPoolGrad is vulnerable to a heap buffer overflow: python import tensorflow as tf
originputtensorshape = tf.constant([1, 3, 2, 3], shape=[4], dtype=tf.int64) outbackprop = tf.constant([2], shape=[1, 1, 1, 1], dtype=tf.int64) rowpoolingsequence = tf.constant([1], shape=[1], dtype=tf.int64) colpoolingsequence = tf.constant([1], shape=[1], dtype=tf.int64)
tf.rawops.FractionalAvgPoolGrad( originputtensorshape=originputtensorshape, outbackprop=outbackprop, rowpoolingsequence=rowpoolingsequence, colpoolingsequence=colpoolingsequence, overlapping=False)
The implementation fails to validate that the pooling sequence arguments have enough elements as required by the outbackprop tensor shape.
Patches We have patched the issue in GitHub commit 12c727cee857fa19be717f336943d95fca4ffe4f.
The fix will be included in TensorFlow 2.5.0. We will also cherrypick this commit on TensorFlow 2.4.2, TensorFlow 2.3.3, TensorFlow 2.2.3 and TensorFlow 2.1.4, as these are also affected and still in supported range.
For more information Please consult our security guide for more information regarding the security model and how to contact us with issues and questions.
Attribution This vulnerability has been reported by Ying Wang and Yakun Zhang of Baidu X-Team.
Other sources
TensorFlow is an end-to-end open source platform for machine learning. The implementation of tf.rawops.FractionalAvgPoolGrad is vulnerable to a heap buffer overflow. The implementation(https://github.com/tensorflow/tensorflow/blob/dcba796a28364d6d7f003f6fe733d82726dda713/tensorflow/core/kernels/fractionalavgpoolop.cc#L216) fails to validate that the pooling sequence arguments have enough elements as required by the outbackprop tensor shape. The fix will be included in TensorFlow 2.5.0. We will also cherrypick this commit on TensorFlow 2.4.2, TensorFlow 2.3.3, TensorFlow 2.2.3 and TensorFlow 2.1.4, as these are also affected and still in supported range.
Affected Software
Remediation
Event History
Frequently Asked Questions
What is the severity of CVE-2021-29578?
CVE-2021-29578 has a high severity due to the potential for a heap buffer overflow, which can lead to arbitrary code execution.
How do I fix CVE-2021-29578?
To fix CVE-2021-29578, upgrade TensorFlow to version 2.4.2 or later.
What versions of TensorFlow are affected by CVE-2021-29578?
CVE-2021-29578 affects TensorFlow versions prior to 2.4.2, specifically versions 2.1.0 through 2.4.1.
Is CVE-2021-29578 a critical vulnerability for TensorFlow?
Yes, CVE-2021-29578 is considered critical due to the risk it poses to application security.
What should I do if I cannot upgrade TensorFlow to fix CVE-2021-29578?
If upgrading is not possible, review your application for vulnerabilities and implement additional security measures to mitigate risks.