Learn about CVE-2021-29604, a TensorFlow vulnerability in TFLite hashtable lookup leading to division by zero error. Find impact, affected versions, and mitigation steps.
TensorFlow is an open-source platform for machine learning. A vulnerability in the TFLite implementation of hashtable lookup can lead to a division by zero error. Attackers can exploit this by crafting a model where
values
's first dimension is 0. The fix will be available in TensorFlow 2.5.0, with cherry-picks in versions 2.1.4, 2.2.3, 2.3.3, and 2.4.2.
Understanding CVE-2021-29604
This CVE identifies a division by zero vulnerability in TFLite's implementation of hashtable lookup.
What is CVE-2021-29604?
CVE-2021-29604 is a vulnerability in TensorFlow's TFLite hashtable lookup, allowing attackers to trigger a division by zero error.
The Impact of CVE-2021-29604
The vulnerability has a CVSS base score of 2.5 (Low severity), with an attack complexity of HIGH and attack vector of LOCAL. Although availability impact is LOW, attackers with low privileges can exploit it.
Technical Details of CVE-2021-29604
The vulnerability arises in TFLite's hashtable lookup implementation, enabling attackers to create malicious models causing a division by zero error.
Vulnerability Description
The vulnerability in TensorFlow's TFLite allows attackers to exploit a division by zero error using crafted models.
Affected Systems and Versions
Versions of TensorFlow prior to 2.1.4 and between 2.2.0 to 2.4.2 are affected by this vulnerability.
Exploitation Mechanism
Attackers can exploit the vulnerability by creating models that set the first dimension of
values
to 0.
Mitigation and Prevention
To address CVE-2021-29604, it is crucial to take immediate steps, adopt secure practices, and apply necessary patches and updates.
Immediate Steps to Take
Update TensorFlow to version 2.5.0 or apply the cherry-picked commits in versions 2.1.4, 2.2.3, 2.3.3, and 2.4.2.
Long-Term Security Practices
Ensure ongoing monitoring, threat intelligence, and adhering to security best practices to reduce the risk of similar vulnerabilities.
Patching and Updates
Regularly update TensorFlow to the latest versions and promptly apply security patches to mitigate risks.