Headline
GHSA-xxcj-rhqg-m46g: Segfault via invalid attributes in `pywrap_tfe_src.cc`
Impact
If a list of quantized tensors is assigned to an attribute, the pywrap code fails to parse the tensor and returns a nullptr, which is not caught. An example can be seen in tf.compat.v1.extract_volume_patches by passing in quantized tensors as input ksizes.
import numpy as np
import tensorflow as tf
a_input = np.array([1, -1], dtype= np.int32)
a_ksizes = a_strides = tf.constant(dtype=tf.dtypes.qint16, value=[[1, 4], [5, 2]])
tf.compat.v1.extract_volume_patches(input=a_input,ksizes=a_ksizes,strides=a_strides,padding='VALID')
Patches
We have patched the issue in GitHub commit e9e95553e5411834d215e6770c81a83a3d0866ce.
The fix will be included in TensorFlow 2.11. We will also cherrypick this commit on TensorFlow 2.10.1, 2.9.3, and TensorFlow 2.8.4, 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 Pattarakrit Rattankul.
Impact
If a list of quantized tensors is assigned to an attribute, the pywrap code fails to parse the tensor and returns a nullptr, which is not caught. An example can be seen in tf.compat.v1.extract_volume_patches by passing in quantized tensors as input ksizes.
import numpy as np import tensorflow as tf
a_input = np.array([1, -1], dtype= np.int32) a_ksizes = a_strides = tf.constant(dtype=tf.dtypes.qint16, value=[[1, 4], [5, 2]])
tf.compat.v1.extract_volume_patches(input=a_input,ksizes=a_ksizes,strides=a_strides,padding=’VALID’)
Patches
We have patched the issue in GitHub commit e9e95553e5411834d215e6770c81a83a3d0866ce.
The fix will be included in TensorFlow 2.11. We will also cherrypick this commit on TensorFlow 2.10.1, 2.9.3, and TensorFlow 2.8.4, 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 Pattarakrit Rattankul.
References
- GHSA-xxcj-rhqg-m46g
- https://nvd.nist.gov/vuln/detail/CVE-2022-41889
- tensorflow/tensorflow@e9e9555
- https://github.com/tensorflow/tensorflow/blob/master/tensorflow/core/kernels/image/generate_box_proposals_op.cu.cc
Related news
TensorFlow is an open source platform for machine learning. If a list of quantized tensors is assigned to an attribute, the pywrap code fails to parse the tensor and returns a `nullptr`, which is not caught. An example can be seen in `tf.compat.v1.extract_volume_patches` by passing in quantized tensors as input `ksizes`. We have patched the issue in GitHub commit e9e95553e5411834d215e6770c81a83a3d0866ce. The fix will be included in TensorFlow 2.11. We will also cherrypick this commit on TensorFlow 2.10.1, 2.9.3, and TensorFlow 2.8.4, as these are also affected and still in supported range.