CVE-2022-21728

EIP tracks 1 public exploit for CVE-2022-21728. PoCs published by mwina.

AI-analyzed exploit summary This PoC demonstrates CVE-2022-21728, a vulnerability in TensorFlow's ReverseSequence operation that can lead to a denial-of-service (DoS) via invalid tensor dimensions. The code triggers the issue by passing invalid seq_dim and batch_dim values.

Tensorflow is an Open Source Machine Learning Framework. The implementation of shape inference for `ReverseSequence` does not fully validate the value of `batch_dim` and can result in a heap OOB read. There is a check to make sure the value of `batch_dim` does not go over the rank of the input, but there is no check for negative values. Negative dimensions are allowed in some cases to mimic Python's negative indexing (i.e., indexing from the end of the array), however if the value is too negative then the implementation of `Dim` would access elements before the start of an array. The fix will be included in TensorFlow 2.8.0. We will also cherrypick this commit on TensorFlow 2.7.1, TensorFlow 2.6.3, and TensorFlow 2.5.3, as these are also affected and still in supported range.