CVE-2021-37657: Reference binding to nullptr in `MatrixDiagV*` ops in TensorFlow
Impact An attacker can cause undefined behavior via binding a reference to null pointer in all operations of type tf.rawops.MatrixDiagV:
python import tensorflow as tf
tf.rawops.MatrixDiagV3( diagonal=[1,0], k=[], numrows=[1,2,3], numcols=[4,5], paddingvalue=[], align='RIGHTRIGHT')
The implementation has incomplete validation that the value of k is a valid tensor. We have check that this value is either a scalar or a vector, but there is no check for the number of elements. If this is an empty tensor, then code that accesses the first element of the tensor is wrong:
cc auto& diagindex = context->input(1); ... lowerdiagindex = diagindex.flat<int32>()(0);
Patches We have patched the issue in GitHub commit f2a673bd34f0d64b8e40a551ac78989d16daad09.
The fix will be included in TensorFlow 2.6.0. We will also cherrypick this commit on TensorFlow 2.5.1, TensorFlow 2.4.3, and TensorFlow 2.3.4, 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 members of the Aivul Team from Qihoo 360.
Other sources
TensorFlow is an end-to-end open source platform for machine learning. In affected versions an attacker can cause undefined behavior via binding a reference to null pointer in all operations of type tf.rawops.MatrixDiagV. The implementation has incomplete validation that the value of k is a valid tensor. We have check that this value is either a scalar or a vector, but there is no check for the number of elements. If this is an empty tensor, then code that accesses the first element of the tensor is wrong. We have patched the issue in GitHub commit f2a673bd34f0d64b8e40a551ac78989d16daad09. The fix will be included in TensorFlow 2.6.0. We will also cherrypick this commit on TensorFlow 2.5.1, TensorFlow 2.4.3, and TensorFlow 2.3.4, as these are also affected and still in supported range.
Affected Software
Remediation
Recommended actions to resolve this vulnerability, in priority order.
- Upgrade
Upgrade
pip/tensorflow-gputo a version that resolves this vulnerability.Fixed in 2.5.1 - Upgrade
Upgrade
pip/tensorflow-gputo a version that resolves this vulnerability.Fixed in 2.4.3 - Upgrade
Upgrade
pip/tensorflow-gputo a version that resolves this vulnerability.Fixed in 2.3.4 - Upgrade
Upgrade
pip/tensorflow-cputo a version that resolves this vulnerability.Fixed in 2.5.1 - Upgrade
Upgrade
pip/tensorflow-cputo a version that resolves this vulnerability.Fixed in 2.4.3 - Upgrade
Upgrade
pip/tensorflow-cputo a version that resolves this vulnerability.Fixed in 2.3.4 - Upgrade
Upgrade
pip/tensorflowto a version that resolves this vulnerability.Fixed in 2.5.1 - Upgrade
Upgrade
pip/tensorflowto a version that resolves this vulnerability.Fixed in 2.4.3 - Upgrade
Upgrade
pip/tensorflowto a version that resolves this vulnerability.Fixed in 2.3.4 - Upgrade
Upgrade
TensorFlowto a version that resolves this vulnerability.Fixed in 2.6.0Patch f2a673bd34f0d64b8e40a551ac78989d16daad09 - Upgrade
Upgrade
TensorFlowto a version that resolves this vulnerability.Fixed in 2.5.1Patch f2a673bd34f0d64b8e40a551ac78989d16daad09 - Upgrade
Upgrade
TensorFlowto a version that resolves this vulnerability.Fixed in 2.4.3Patch f2a673bd34f0d64b8e40a551ac78989d16daad09 - Upgrade
Upgrade
TensorFlowto a version that resolves this vulnerability.Fixed in 2.3.4Patch f2a673bd34f0d64b8e40a551ac78989d16daad09
Event History
Frequently Asked Questions
What is the severity of CVE-2021-37657?
CVE-2021-37657 is classified as a high severity vulnerability due to potential undefined behavior that can be exploited by attackers.
How do I fix CVE-2021-37657?
To fix CVE-2021-37657, upgrade to TensorFlow version 2.5.1 or later, or to version 2.4.3 if using 2.4.x.
Which TensorFlow versions are affected by CVE-2021-37657?
CVE-2021-37657 affects TensorFlow versions 2.3.0 to 2.3.4, 2.4.0 to 2.4.3, and 2.5.0 up to 2.6.0-rc2.
What kind of attack can exploit CVE-2021-37657?
An attacker can exploit CVE-2021-37657 by invoking operations of type tf.raw_ops.MatrixDiagV* with a null pointer, leading to undefined behavior.
Is CVE-2021-37657 specific to any TensorFlow packages?
CVE-2021-37657 affects both the tensorflow-gpu and tensorflow-cpu packages across the impacted versions.