CVE-2020-15207

Out-of-bounds access in TensorFlow Lite (before versions 1.15.4, 2.0.3, 2.1.2, 2.2.1, and 2.3.1)

• CVSS 9.0
• CWE-787
• Memory Corruption
• Remote

In tensorflow-lite before versions 1.15.4, 2.0.3, 2.1.2, 2.2.1 and 2.3.1, to mimic Python's indexing with negative values, TFLite uses `ResolveAxis` to convert negative values to positive indices. However, the only check that the converted index is now valid is only present in debug builds. If the `DCHECK` does not trigger, then code execution moves ahead with a negative index. This, in turn, results in accessing data out of bounds which results in segfaults and/or data corruption. The issue is patched in commit 2d88f470dea2671b430884260f3626b1fe99830a, and is released in TensorFlow versions 1.15.4, 2.0.3, 2.1.2, 2.2.1, or 2.3.1.

CVE

CVE-2020-15207

CWE

CWE-787

CVSS base score

9.0

Published

2020-09-25

OWASP

A08 Software and Data Integrity Failures

Orthogonal defect classification

Checking

Code defect classification

Missing Check

Category

Memory Corruption

Subcategory

Out-of-Bound Accesses

Accessibility scope

Remote

Impact

Arbitrary Code Execution

Affected component

TensorFlow

Fixed by upgrading

Yes

Solution

Update to TensorFlow version 1.15.4, 2.0.3, 2.1.2, 2.2.1, or 2.3.1 or higher

import tensorflow as tf

def demonstrate_cve(tensor_shape):
    tensor = tf.zeros(tensor_shape)
    indices = tf.constant([[-1]])
    try:
        result = tf.gather_nd(tensor, indices)
        print("Result:", result)
    except Exception as e:
        print("Exception:", e)

tensor_shape = [3, 4, 5]
demonstrate_cve(tensor_shape)

import tensorflow as tf

def demonstrate_cve(tensor_shape):
    tensor = tf.zeros(tensor_shape)

    def check_axis(axis, tensor_shape):
        if axis < 0:
            axis += len(tensor_shape)
        if axis < 0 or axis >= len(tensor_shape):
            raise ValueError(f"Axis {axis} is out of bounds for tensor with shape {tensor_shape}")
        return axis

    indices = tf.constant([[-1]])
    resolved_axis = check_axis(indices[0][0], tensor_shape)
    indices = tf.constant([[resolved_axis]])

    try:
        result = tf.gather_nd(tensor, indices)
        print("Result:", result)
    except Exception as e:
        print("Exception:", e)

tensor_shape = [3, 4, 5]
demonstrate_cve(tensor_shape)

@misc{vaitp:cve202015207,
  title        = {{Out-of-bounds access in TensorFlow Lite (before versions 1.15.4, 2.0.3, 2.1.2, 2.2.1, and 2.3.1)}},
  author       = {Bogaerts, Fr\'ed\'eric and Ivaki, Naghmeh and Fonseca, Jos\'e},
  year         = {2020},
  note         = {VAITP Python Vulnerability Dataset, entry CVE-2020-15207},
  howpublished = {\url{https://netpack.pt/vaitp/vulnerability/CVE-2020-15207/}}
}