CVE-2021-41199: Overflow/crash in `tf.image.resize` when size is large
Impact If tf.image.resize is called with a large input argument then the TensorFlow process will crash due to a CHECK-failure caused by an overflow.
python import tensorflow as tf import numpy as np
tf.keras.layers.UpSampling2D( size=1610637938, dataformat='channelsfirst', interpolation='bilinear')(np.ones((5,1,1,1)))
The number of elements in the output tensor is too much for the int64t type and the overflow is detected via a CHECK statement. This aborts the process.
Patches We have patched the issue in GitHub commit e5272d4204ff5b46136a1ef1204fc00597e21837 (merging #51497).
The fix will be included in TensorFlow 2.7.0. We will also cherrypick this commit on TensorFlow 2.6.1, TensorFlow 2.5.2, and TensorFlow 2.4.4, 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 externally via a GitHub issue.
Other sources
TensorFlow is an open source platform for machine learning. In affected versions if tf.image.resize is called with a large input argument then the TensorFlow process will crash due to a CHECK-failure caused by an overflow. The number of elements in the output tensor is too much for the int64t type and the overflow is detected via a CHECK statement. This aborts the process. The fix will be included in TensorFlow 2.7.0. We will also cherrypick this commit on TensorFlow 2.6.1, TensorFlow 2.5.2, and TensorFlow 2.4.4, as these are also affected and still in supported range.
Affected Software
Remediation
Event History
Frequently Asked Questions
What is the severity of CVE-2021-41199?
CVE-2021-41199 has a high severity rating due to the potential for crashes in the TensorFlow process on large input arguments.
How do I fix CVE-2021-41199?
To fix CVE-2021-41199, update TensorFlow to version 2.4.4 or higher, specifically to versions 2.5.2 or 2.6.1.
Which versions of TensorFlow are affected by CVE-2021-41199?
CVE-2021-41199 affects TensorFlow versions prior to 2.4.4 as well as 2.5.0 and 2.6.0.
What causes the crash in CVE-2021-41199?
The crash is caused by a CHECK-failure due to an overflow when calling tf.image.resize with a large input argument.
Is CVE-2021-41199 a remote attack vector?
CVE-2021-41199 does not provide a remote attack vector as it requires local execution of the TensorFlow code with specific inputs.