MISFIRE

Browse / AI Framework and Platform Flaws / CVE-2021-37654

Heap OOB and CHECK fail in `ResourceGather`

CVE-2021-37654 ยท 2021-08-25 high

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

How it works, in short

An excessive batch_dims value makes ResourceGather read past the tensor's heap buffer because its rank check is only a debug check. (Our summary, written by a model. The source text is below.)

Details

Impact

An attacker can trigger a crash via a CHECK-fail in debug builds of TensorFlow using tf.raw_ops.ResourceGather or a read from outside the bounds of heap allocated data in the same API in a release build:

import tensorflow as tf

tensor = tf.constant(value=[[1,2],[3,4],[5,6]],shape=(3,2),dtype=tf.uint32)
v = tf.Variable(tensor)
tf.raw_ops.ResourceGather(
  resource=v.handle,
  indices=[0],
  dtype=tf.uint32,
  batch_dims=10,
  validate_indices=False)

The implementation does not check that the batch_dims value that the user supplies is less than the rank of the input tensor.

Since the implementation uses several for loops over the dimensions of tensor, this results in reading data from outside the bounds of heap allocated buffer backing the tensor:

    // batch_dims_ = > params.dims() (10 > 2)
    for (int i = 0; i < batch_dims_; ++i) {
      result_shape.AddDim(params.dim_size(i));
    }
    for (int i = batch_dims_; i < indices.dims(); ++i) {
      result_shape.AddDim(indices.dim_size(i));
    }
    for (int i = batch_dims_ + 1; i < params.dims(); ++i) {
      result_shape.AddDim(params.dim_size(i));
    }

In debug mode, .dim_size(i) validates that the argument is less than .dims() using a DCHECK. But the DCHECK is a no-op in release builds.

Patches

We have patched the issue in GitHub commit bc9c546ce7015c57c2f15c168b3d9201de679a1d.

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.