CVE-2022-35974: Segfault in `QuantizeDownAndShrinkRange` in TensorFlow
TensorFlow is an open source platform for machine learning. If QuantizeDownAndShrinkRange is given nonscalar inputs for inputmin or inputmax, it results in a segfault that can be used to trigger a denial of service attack. We have patched the issue in GitHub commit 73ad1815ebcfeb7c051f9c2f7ab5024380ca8613. The fix will be included in TensorFlow 2.10.0. We will also cherrypick this commit on TensorFlow 2.9.1, TensorFlow 2.8.1, and TensorFlow 2.7.2, as these are also affected and still in supported range. There are no known workarounds for this issue.
Affected Software
Remediation
Event History
Frequently Asked Questions
What is the severity of CVE-2022-35974?
CVE-2022-35974 has been rated as a denial of service vulnerability due to resulting segmentation faults.
How do I fix CVE-2022-35974?
To fix CVE-2022-35974, update TensorFlow to a version later than 2.10-rc3, which includes the patch.
What types of inputs are affected by CVE-2022-35974?
CVE-2022-35974 is triggered when nonscalar inputs are provided for `input_min` or `input_max` in the `QuantizeDownAndShrinkRange` function.
What can be the consequence of exploiting CVE-2022-35974?
Exploiting CVE-2022-35974 can lead to application crashes and denial of service.
Which TensorFlow versions are vulnerable to CVE-2022-35974?
The vulnerable TensorFlow versions include 2.7.0 through 2.7.2, 2.8.0 through 2.8.1, and 2.9.0 through 2.9.1, as well as pre-releases of version 2.10.