CVE-2021-29513

Type confusion during tensor casts lead to dereferencing null pointers

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 as well as Ye Zhang and Yakun Zhang of Baidu X-Team.

Calling TF operations with tensors of non-numeric types when the operations expect numeric tensors result in null pointer dereferences.

There are multiple ways to reproduce this, listing a few examples here:

import tensorflow as tf
import numpy as np
data = tf.random.truncated_normal(shape=1,mean=np.float32(20.8739),stddev=779.973,dtype=20,seed=64)

import tensorflow as tf
import numpy as np
data =
tf.random.stateless_truncated_normal(shape=1,seed=[63,70],mean=np.float32(20.8739),stddev=779.973,dtype=20)

import tensorflow as tf
import numpy as np
data = tf.one_hot(indices=[62,50],depth=136,on_value=np.int32(237),off_value=158,axis=856,dtype=20)

import tensorflow as tf
import numpy as np
data = tf.range(start=np.int32(214),limit=660,delta=129,dtype=20)

import tensorflow as tf
import numpy as np
data = tf.raw_ops.ResourceCountUpTo(resource=np.int32(30), limit=872, T=3)

import tensorflow as tf
import numpy as np

writer_array = np.array([1,2],dtype=np.int32)
writer_tensor = tf.convert_to_tensor(writer_array,dtype=tf.resource)

All these examples and similar ones have the same behavior: the conversion from Python array to C++ array is vulnerable to a type confusion:

  int pyarray_type = PyArray_TYPE(array);
  PyArray_Descr* descr = PyArray_DESCR(array);
  switch (pyarray_type) {
    ...
    case NPY_VOID:
      // Quantized types are currently represented as custom struct types.
      // PyArray_TYPE returns NPY_VOID for structs, and we should look into
      // descr to derive the actual type.
      // Direct feeds of certain types of ResourceHandles are represented as a
      // custom struct type.
      return PyArrayDescr_to_TF_DataType(descr, out_tf_datatype);
    ...
  }

For the tensor types involved in the above example, the pyarray_type is NPY_VOID but the descr field is such that descr->field = NULL. Then PyArrayDescr_to_TF_DataType will trigger a null dereference:

Status PyArrayDescr_to_TF_DataType(PyArray_Descr* descr,
                                   TF_DataType* out_tf_datatype) {
  PyObject* key;
  PyObject* value;
  Py_ssize_t pos = 0;
  if (PyDict_Next(descr->fields, &pos, &key, &value)) {
    ...
  }
}

This is because the Python's PyDict_Next implementation would dereference the first argument.

The application dereferences a null pointer, causing a crash. Typical impact: denial of service via crash.

CVE-2021-29513 has a CVSS score of 2.5 (Low). The vector is requires local access, low privileges required, and no user interaction. A CVSS score reflects the worst-case severity of the vulnerability, not your specific exposure. Whether this affects your application depends on whether the vulnerable code is present and reachable in your environment. A fixed version is available (2.1.4, 2.2.3, 2.3.3, 2.4.2); upgrading removes the vulnerable code path.

Kodem intelligence

Already deployed Kodem?

We have patched the issue in GitHub commit 030af767d357d1b4088c4a25c72cb3906abac489.

The fix will be included in TensorFlow 2.5.0. We will also cherrypick this commit on TensorFlow 2.4.2, TensorFlow 2.3.3, TensorFlow 2.2.3 and TensorFlow 2.1.4, as these are also affected and still in supported range.

1. What is CVE-2021-29513? CVE-2021-29513 is a low-severity null pointer dereference vulnerability in tensorflow (pip), affecting versions < 2.1.4. It is fixed in 2.1.4, 2.2.3, 2.3.3, 2.4.2. The application dereferences a null pointer, causing a crash.
2. How severe is CVE-2021-29513? CVE-2021-29513 has a CVSS score of 2.5 (Low). This score reflects the worst-case severity of the vulnerability, not your specific exposure. Whether it represents real risk in your environment depends on whether the vulnerable code is present and reachable.
3. Which packages are affected by CVE-2021-29513?
   • tensorflow (pip) (versions < 2.1.4)
   • tensorflow-cpu (pip) (versions < 2.1.4)
   • tensorflow-gpu (pip) (versions < 2.1.4)
4. Is there a fix for CVE-2021-29513? Yes. CVE-2021-29513 is fixed in 2.1.4, 2.2.3, 2.3.3, 2.4.2. Upgrade to this version or later.
5. Is CVE-2021-29513 exploitable, and should I be worried? Whether CVE-2021-29513 is exploitable in your environment depends on whether the vulnerable code is present and reachable. A CVSS score is a worst-case rating; it does not account for your specific deployment, configuration, or usage patterns. Kodem, an Intelligent Application Security platform, uses runtime intelligence to show which vulnerabilities actually execute in production, so you can focus on the ones that represent real risk. Get a demo
6. What actually determines whether CVE-2021-29513 is exploitable, and how bad it is? Exploitability and impact are not fixed properties of a CVE. They depend on runtime truth: whether the vulnerable code is present, reachable, and actually executes in your application. A high CVSS score on a dependency that never runs is not the same as real risk. Kodem, an Intelligent Application Security platform, uses runtime intelligence to reveal which vulnerabilities actually execute in production, so teams prioritize the ones that genuinely matter.
7. How do I fix CVE-2021-29513?
   • Upgrade tensorflow to 2.1.4 or later
   • Upgrade tensorflow to 2.2.3 or later
   • Upgrade tensorflow to 2.3.3 or later
   • Upgrade tensorflow to 2.4.2 or later
   • Upgrade tensorflow-cpu to 2.1.4 or later
   • Upgrade tensorflow-cpu to 2.2.3 or later
   • Upgrade tensorflow-cpu to 2.3.3 or later
   • Upgrade tensorflow-cpu to 2.4.2 or later
   • Upgrade tensorflow-gpu to 2.1.4 or later
   • Upgrade tensorflow-gpu to 2.2.3 or later
   • Upgrade tensorflow-gpu to 2.3.3 or later
   • Upgrade tensorflow-gpu to 2.4.2 or later

Other vulnerabilities in tensorflow