The implementation of [tf.raw ops.StagePeek](https://github.com/tensorflow/tensorflow/blob/f3b9bf4c3c0597563b289c0512e98d4ce81f886e/tensorflow/core/kernels/stage op.cc#L261) does not fully validate the input arguments. This results in a CHECK-failure which can be used to trigger a denial of service attack:

import tensorflow as tf

index = tf.constant([], shape=[0], dtype=tf.int32)
tf.raw ops.StagePeek(index=index, dtypes=[tf.int32])

The code assumes index is a scalar but there is no validation for this before accessing its value:

std::size t index = ctx->input(0).scalar<int>()();

We have patched the issue in GitHub commit cebe3c45d76357d201c65bdbbf0dbe6e8a63bbdb.

The fix will be included in TensorFlow 2.9.0. We will also cherrypick this commit on TensorFlow 2.8.1, TensorFlow 2.7.2, and TensorFlow 2.6.4, 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 Neophytos Christou from Secure Systems Lab at Brown University.