CVE-2023-25659: TensorFlow Out-of-Bounds Read in DynamicStitch

Background

TensorFlow is an open-source machine learning framework widely used for developing and training deep learning models. It provides a comprehensive ecosystem for building and deploying machine learning applications. TensorFlow is vulnerable to an Out-of-Bounds Read in the DynamicStitch function. This function can trigger a stack-based out-of-bounds read if the indices parameter does not match the shape of the data parameter. This vulnerability can potentially lead to information disclosure or denial of service.

Vulnerability Detail

The vulnerability arises when the indices parameter for the DynamicStitch function does not match the shape of the data parameter. This mismatch can cause a stack-based out-of-bounds read, potentially leading to information disclosure or denial of service. An attacker could exploit this vulnerability by providing malicious input to the affected function. The vulnerability has been assigned a CVSS score of 7.5, indicating a high severity level.

pip install --upgrade "tensorflow==2.21.0"

pip install --upgrade "tensorflow==2.11.1"

Workaround

If updating TensorFlow is not immediately feasible, a temporary workaround can be implemented by validating the indices and data parameters before passing them to the DynamicStitch function. Ensure that the shapes of both parameters match to avoid triggering the vulnerability. Here is an example of how to perform the validation:

import tensorflow as tf

def safe_dynamic_stitch(indices, data):
    if len(indices) == len(data):
        return tf.raw_ops.DynamicStitch(indices=indices, data=data)
    else:
        raise ValueError('Indices and data must have the same shape.')

# Usage
indices = [[0xdeadbeef], [405], [519], [758], [1015]]
data = [[110.27793884277344], [120.29475402832031], [157.2418212890625], [157.2626953125], [188.45382690429688]]
y = safe_dynamic_stitch(indices, data)

By implementing this workaround, the vulnerability can be mitigated until the TensorFlow library is updated.

Conclusion

The vulnerability in TensorFlow's DynamicStitch function can lead to an out-of-bounds read, potentially resulting in information disclosure or denial of service. To mitigate this vulnerability, it is recommended to update TensorFlow to version 2.11.1 or later. A temporary workaround is available by validating the parameters before using the affected function. For further assistance and support in vulnerability remediation, please contact DevNack.com.

See every vulnerability recorded for tensorflow →

FAQ