Skip to content

Latest commit

 

History

History
27 lines (19 loc) · 1.29 KB

tfsa-2022-128.md

File metadata and controls

27 lines (19 loc) · 1.29 KB

TFSA-2022-128: CHECK fail in TensorListFromTensor

CVE Number

CVE-2022-35992

Impact

When TensorListFromTensor receives an element_shape of a rank greater than one, it gives a CHECK fail that can trigger a denial of service attack.

import tensorflow as tf
arg_0=tf.random.uniform(shape=(6, 6, 2), dtype=tf.bfloat16, maxval=None)
arg_1=tf.random.uniform(shape=(6, 9, 1, 3), dtype=tf.int64, maxval=65536)
arg_2=''
tf.raw_ops.TensorListFromTensor(tensor=arg_0, element_shape=arg_1, name=arg_2)

Patches

We have patched the issue in GitHub commit 3db59a042a38f4338aa207922fa2f476e000a6ee.

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.