CVE-2021-37687: Tensorflow Heap OOB Vulnerability

Q: What is CVE-2021-37687?

CVE-2021-37687 is a security vulnerability in Tensorflow involving Heap OOB in TFLite's `Gather*` implementations under CWE-125: Out-of-bounds Read.

Q: Which Tensorflow versions are affected?

Affected versions include releases prior to 2.3.4, 2.4.3, 2.5.1 across affected packages: tensorflow, tensorflow-cpu, tensorflow-gpu.

Q: What version fixes CVE-2021-37687?

Upgrading to 2.3.4, 2.4.3, 2.5.1 (or the latest patched release for your respective branch) resolves the vulnerability.

Q: Is tensorflow affected?

Yes, tensorflow is affected in releases prior to 2.3.4, 2.4.3, 2.5.1. Upgrade your dependency to 2.3.4, 2.4.3, 2.5.1 or later.

Q: Is tensorflow-cpu affected?

Yes, tensorflow-cpu is affected in releases prior to 2.3.4, 2.4.3, 2.5.1. Upgrade your dependency to 2.3.4, 2.4.3, 2.5.1 or later.

Q: What is the CVSS score of CVE-2021-37687?

CVE-2021-37687 has a CVSS v3.1 base score of 5.5 (MODERATE) with vector string CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:N/A:N.

Q: What CWE is associated with CVE-2021-37687?

CVE-2021-37687 is classified under CWE-125: Out-of-bounds Read (The product reads data past the end, or before the beginning, of the intended buffer.).

Q: How can I check PyPI dependencies for CVE-2021-37687?

Run the dependency tree command: 'pip show tensorflow'. This command lists direct and transitive dependencies in your build configuration so you can verify if a vulnerable version is resolved.

MODERATE

Advisory Identifier: GHSA-jwf9-w5xm-f437

CVSS 5.5

CWE-125

Published 2021-08-25

Advisory Summary

CVE-2021-37687 is a MODERATE (CVSS 5.5) heap oob vulnerability (CWE-125: Out-of-bounds Read) affecting Tensorflow libraries in the PyPI ecosystem. The issue involves heap oob in tflite's `gather*` implementations. Remediation is available by upgrading affected packages to version 2.3.4, 2.4.3, 2.5.1.

CVE-2021-37687 at a Glance

CVE Identifier CVE-2021-37687

GHSA Identifier GHSA-jwf9-w5xm-f437

Severity MODERATE

CVSS v3.1 Base Score 5.5

CWE Classification CWE-125: Out-of-bounds Read

Affected Product Tensorflow

Package Ecosystem PyPI

Published Date 2021-08-25

Last Updated 2024-11-13

Security Impact

CVSS: 5.5

The security impact metrics below are parsed directly from the CVSS v3.1 vector string (CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:N/A:N).

Attack Vector Local (AV:L)

Attack Complexity Low (AC:L)

Privileges Required Low (PR:L)

User Interaction None (UI:N)

Scope Unchanged (S:U)

Confidentiality High (C:H)

Integrity None (I:N)

Availability None (A:N)

How to Fix CVE-2021-37687

To remediate this vulnerability, upgrade all affected Tensorflow dependencies to a patched release. Both direct and transitive dependencies should be checked, as frameworks or security libraries may pull in vulnerable versions transitively.

tensorflow fixed: 2.3.4

pip install tensorflow>=2.3.4

tensorflow fixed: 2.4.3

pip install tensorflow>=2.4.3

tensorflow fixed: 2.5.1

pip install tensorflow>=2.5.1

tensorflow-cpu fixed: 2.3.4

pip install tensorflow-cpu>=2.3.4

tensorflow-cpu fixed: 2.4.3

pip install tensorflow-cpu>=2.4.3

tensorflow-cpu fixed: 2.5.1

pip install tensorflow-cpu>=2.5.1

tensorflow-gpu fixed: 2.3.4

pip install tensorflow-gpu>=2.3.4

tensorflow-gpu fixed: 2.4.3

pip install tensorflow-gpu>=2.4.3

tensorflow-gpu fixed: 2.5.1

pip install tensorflow-gpu>=2.5.1

Verify Direct and Transitive Dependencies

Run the following command in your project directory to verify that all resolved dependencies use the patched release:

pip show tensorflow

How to Check if You Are Affected

To determine whether your Python environment is affected by CVE-2021-37687, inspect your installed packages using pip:

pip show tensorflow

Analyze the terminal output to locate any version of Tensorflow in your dependency tree. If any resolved version matches the vulnerable ranges listed in the table below (for example, versions earlier than 2.3.4), your application is affected.

Note: The pip show tensorflow command checks currently installed packages in your active Python environment.

Technical Details

Vulnerability Mechanism & Technical Analysis

This vulnerability is classified under CWE-125: Out-of-bounds Read, where the product reads data past the end, or before the beginning, of the intended buffer. In Tensorflow, software state validation or input boundaries are circumvented during processing. Review the official source advisory statement below for specific affected components, classes, and reproduction details.

Source Advisory Statement

### Impact TFLite's [`GatherNd` implementation](https://github.com/tensorflow/tensorflow/blob/149562d49faa709ea80df1d99fc41d005b81082a/tensorflow/lite/kernels/gather_nd.cc#L124) does not support negative indices but there are no checks for this situation. Hence, an attacker can read arbitrary data from the heap by carefully crafting a model with negative values in `indices`. Similar issue exists in [`Gather` implementation](https://github.com/tensorflow/tensorflow/blob/149562d49faa709ea80df1d99fc41d005b81082a/tensorflow/lite/kernels/gather.cc). ```python import tensorflow as tf import numpy as np tf.compat.v1.disable_v2_behavior() params = tf.compat.v1.placeholder(name="params", dtype=tf.int64, shape=(1,)) indices = tf.compat.v1.placeholder(name="indices", dtype=tf.int64, shape=()) out = tf.gather(params, indices, name='out') with tf.compat.v1.Session() as sess: converter = tf.compat.v1.lite.TFLiteConverter.from_session(sess, [params, indices], [out]) tflite_model = converter.convert() interpreter = tf.lite.Interpreter(model_content=tflite_model) interpreter.allocate_tensors() input_details = interpreter.get_input_details() output_details = interpreter.get_output_details() params_data = np.reshape(np.array([1], dtype=np.int64), newshape=(1,)) indices_data = np.reshape(np.array(-10, dtype=np.int64), newshape=()) interpreter.set_tensor(input_details[0]['index'], params_data) interpreter.set_tensor(input_details[1]['index'], indices_data) interpreter.invoke() ``` ### Patches We have patched the issue in GitHub commits [bb6a0383ed553c286f87ca88c207f6774d5c4a8f](https://github.com/tensorflow/tensorflow/commit/bb6a0383ed553c286f87ca88c207f6774d5c4a8f) and [eb921122119a6b6e470ee98b89e65d721663179d](https://github.com/tensorflow/tensorflow/commit/eb921122119a6b6e470ee98b89e65d721663179d). 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](https://github.com/tensorflow/tensorflow/blob/master/SECURITY.md) for more information regarding the security model and how to contact us with issues and questions. ### Attribution This vulnerability has been reported by Yakun Zhang of Baidu Security.

CVE-2021-37687 Affected Tensorflow Versions

9 component(s)

Fixed Versions:

2.3.42.4.32.5.1

Product / Package	Ecosystem	Affected Versions	Fixed Version
tensorflow pypi/tensorflow	PyPI	< 2.3.4	2.3.4
tensorflow pypi/tensorflow	PyPI	≥ 2.4.0 < 2.4.3	2.4.3
tensorflow pypi/tensorflow	PyPI	≥ 2.5.0 < 2.5.1 All 1 versions 2.5.0	2.5.1
tensorflow-cpu pypi/tensorflow-cpu	PyPI	< 2.3.4	2.3.4
tensorflow-cpu pypi/tensorflow-cpu	PyPI	≥ 2.4.0 < 2.4.3	2.4.3
tensorflow-cpu pypi/tensorflow-cpu	PyPI	≥ 2.5.0 < 2.5.1 All 1 versions 2.5.0	2.5.1
tensorflow-gpu pypi/tensorflow-gpu	PyPI	< 2.3.4	2.3.4
tensorflow-gpu pypi/tensorflow-gpu	PyPI	≥ 2.4.0 < 2.4.3	2.4.3
tensorflow-gpu pypi/tensorflow-gpu	PyPI	≥ 2.5.0 < 2.5.1 All 1 versions 2.5.0	2.5.1

References

GitHub Advisory / Wiki ADVISORY

https://github.com/tensorflow/tensorflow/security/advisories/GHSA-jwf9-w5xm-f437

NVD — CVE-2021-37687 NVD

https://nvd.nist.gov/vuln/detail/CVE-2021-37687

Security Fix Commit FIX COMMIT

https://github.com/tensorflow/tensorflow/commit/bb6a0383ed553c286f87ca88c207f6774d5c4a8f

Security Fix Commit FIX COMMIT

https://github.com/tensorflow/tensorflow/commit/eb921122119a6b6e470ee98b89e65d721663179d

Tensorflow Source Repository REPOSITORY

https://github.com/pypa/advisory-database/tree/main/vulns/tensorflow-cpu/PYSEC-2021-600.yaml

Tensorflow Source Repository REPOSITORY

https://github.com/pypa/advisory-database/tree/main/vulns/tensorflow-gpu/PYSEC-2021-798.yaml

Tensorflow Source Repository REPOSITORY

https://github.com/pypa/advisory-database/tree/main/vulns/tensorflow/PYSEC-2021-309.yaml

Tensorflow Source Repository REPOSITORY

https://github.com/tensorflow/tensorflow

Tensorflow Source Repository REPOSITORY

https://github.com/tensorflow/tensorflow/blob/149562d49faa709ea80df1d99fc41d005b81082a/tensorflow/lite/kernels/gather.cc

Tensorflow Source Repository REPOSITORY

https://github.com/tensorflow/tensorflow/blob/149562d49faa709ea80df1d99fc41d005b81082a/tensorflow/lite/kernels/gather_nd.cc#L124

Frequently Asked Questions

What is CVE-2021-37687?

CVE-2021-37687 is a security vulnerability in Tensorflow involving Heap OOB in TFLite's `Gather*` implementations under CWE-125: Out-of-bounds Read.

Which Tensorflow versions are affected?

Affected versions include releases prior to 2.3.4, 2.4.3, 2.5.1 across affected packages: tensorflow, tensorflow-cpu, tensorflow-gpu.

What version fixes CVE-2021-37687?

Upgrading to 2.3.4, 2.4.3, 2.5.1 (or the latest patched release for your respective branch) resolves the vulnerability.

Is tensorflow affected?

Yes, tensorflow is affected in releases prior to 2.3.4, 2.4.3, 2.5.1. Upgrade your dependency to 2.3.4, 2.4.3, 2.5.1 or later.

Is tensorflow-cpu affected?

Yes, tensorflow-cpu is affected in releases prior to 2.3.4, 2.4.3, 2.5.1. Upgrade your dependency to 2.3.4, 2.4.3, 2.5.1 or later.

What is the CVSS score of CVE-2021-37687?

CVE-2021-37687 has a CVSS v3.1 base score of 5.5 (MODERATE) with vector string CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:N/A:N.

What CWE is associated with CVE-2021-37687?

CVE-2021-37687 is classified under CWE-125: Out-of-bounds Read (The product reads data past the end, or before the beginning, of the intended buffer.).

How can I check PyPI dependencies for CVE-2021-37687?

Run the dependency tree command: 'pip show tensorflow'. This command lists direct and transitive dependencies in your build configuration so you can verify if a vulnerable version is resolved.

Machine-readable OSV JSON OSV v1.5 Schema

Normalized schema output

View Raw JSON

{
  "schema_version" : "1.4.0",
  "id" : "GHSA-jwf9-w5xm-f437",
  "modified" : "2024-11-13T21:21:55Z",
  "published" : "2021-08-25T14:40:02Z",
  "aliases" : [ "CVE-2021-37687" ],
  "summary" : "Heap OOB in TFLite's `Gather*` implementations",
  "details" : "### Impact\nTFLite's [`GatherNd` implementation](https://github.com/tensorflow/tensorflow/blob/149562d49faa709ea80df1d99fc41d005b81082a/tensorflow/lite/kernels/gather_nd.cc#L124) does not support negative indices but there are no checks for this situation.\n\nHence, an attacker can read arbitrary data from the heap by carefully crafting a model with negative values in `indices`.\n\nSimilar issue exists in [`Gather` implementation](https://github.com/tensorflow/tensorflow/blob/149562d49faa709ea80df1d99fc41d005b81082a/tensorflow/lite/kernels/gather.cc).\n\n```python\nimport tensorflow as tf\nimport numpy as np\ntf.compat.v1.disable_v2_behavior()\n\nparams = tf.compat.v1.placeholder(name=\"params\", dtype=tf.int64, shape=(1,))\nindices = tf.compat.v1.placeholder(name=\"indices\", dtype=tf.int64, shape=())\n\nout = tf.gather(params, indices, name='out')\n\nwith tf.compat.v1.Session() as sess:\n   converter = tf.compat.v1.lite.TFLiteConverter.from_session(sess, [params, indices], [out])\n   tflite_model = converter.convert()\n\ninterpreter = tf.lite.Interpreter(model_content=tflite_model)\ninterpreter.allocate_tensors()\n\ninput_details = interpreter.get_input_details()\noutput_details = interpreter.get_output_details()\n\nparams_data = np.reshape(np.array([1], dtype=np.int64), newshape=(1,))\nindices_data = np.reshape(np.array(-10, dtype=np.int64), newshape=())\ninterpreter.set_tensor(input_details[0]['index'], params_data)\ninterpreter.set_tensor(input_details[1]['index'], indices_data)\n\ninterpreter.invoke()\n```\n\n### Patches\nWe have patched the issue in GitHub commits [bb6a0383ed553c286f87ca88c207f6774d5c4a8f](https://github.com/tensorflow/tensorflow/commit/bb6a0383ed553c286f87ca88c207f6774d5c4a8f) and [eb921122119a6b6e470ee98b89e65d721663179d](https://github.com/tensorflow/tensorflow/commit/eb921122119a6b6e470ee98b89e65d721663179d).\n\nThe 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.\n\n### For more information\nPlease consult [our security  guide](https://github.com/tensorflow/tensorflow/blob/master/SECURITY.md) for more information regarding the security model and how to contact us with issues and questions.\n\n### Attribution\nThis vulnerability has been reported by Yakun Zhang of Baidu Security.",
  "severity" : [ {
    "type" : "CVSS_V3",
    "score" : "CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:N/A:N"
  }, {
    "type" : "CVSS_V4",
    "score" : "CVSS:4.0/AV:L/AC:L/AT:N/PR:L/UI:N/VC:H/VI:N/VA:N/SC:N/SI:N/SA:N"
  } ],
  "affected" : [ {
    "package" : {
      "ecosystem" : "PyPI",
      "name" : "tensorflow"
    },
    "ranges" : [ {
      "type" : "ECOSYSTEM",
      "events" : [ {
        "introduced" : "0"
      }, {
        "fixed" : "2.3.4"
      } ]
    } ]
  }, {
    "package" : {
      "ecosystem" : "PyPI",
      "name" : "tensorflow"
    },
    "ranges" : [ {
      "type" : "ECOSYSTEM",
      "events" : [ {
        "introduced" : "2.4.0"
      }, {
        "fixed" : "2.4.3"
      } ]
    } ]
  }, {
    "package" : {
      "ecosystem" : "PyPI",
      "name" : "tensorflow"
    },
    "ranges" : [ {
      "type" : "ECOSYSTEM",
      "events" : [ {
        "introduced" : "2.5.0"
      }, {
        "fixed" : "2.5.1"
      } ]
    } ],
    "versions" : [ "2.5.0" ]
  }, {
    "package" : {
      "ecosystem" : "PyPI",
      "name" : "tensorflow-cpu"
    },
    "ranges" : [ {
      "type" : "ECOSYSTEM",
      "events" : [ {
        "introduced" : "0"
      }, {
        "fixed" : "2.3.4"
      } ]
    } ]
  }, {
    "package" : {
      "ecosystem" : "PyPI",
      "name" : "tensorflow-cpu"
    },
    "ranges" : [ {
      "type" : "ECOSYSTEM",
      "events" : [ {
        "introduced" : "2.4.0"
      }, {
        "fixed" : "2.4.3"
      } ]
    } ]
  }, {
    "package" : {
      "ecosystem" : "PyPI",
      "name" : "tensorflow-cpu"
    },
    "ranges" : [ {
      "type" : "ECOSYSTEM",
      "events" : [ {
        "introduced" : "2.5.0"
      }, {
        "fixed" : "2.5.1"
      } ]
    } ],
    "versions" : [ "2.5.0" ]
  }, {
    "package" : {
      "ecosystem" : "PyPI",
      "name" : "tensorflow-gpu"
    },
    "ranges" : [ {
      "type" : "ECOSYSTEM",
      "events" : [ {
        "introduced" : "0"
      }, {
        "fixed" : "2.3.4"
      } ]
    } ]
  }, {
    "package" : {
      "ecosystem" : "PyPI",
      "name" : "tensorflow-gpu"
    },
    "ranges" : [ {
      "type" : "ECOSYSTEM",
      "events" : [ {
        "introduced" : "2.4.0"
      }, {
        "fixed" : "2.4.3"
      } ]
    } ]
  }, {
    "package" : {
      "ecosystem" : "PyPI",
      "name" : "tensorflow-gpu"
    },
    "ranges" : [ {
      "type" : "ECOSYSTEM",
      "events" : [ {
        "introduced" : "2.5.0"
      }, {
        "fixed" : "2.5.1"
      } ]
    } ],
    "versions" : [ "2.5.0" ]
  } ],
  "references" : [ {
    "type" : "WEB",
    "url" : "https://github.com/tensorflow/tensorflow/security/advisories/GHSA-jwf9-w5xm-f437"
  }, {
    "type" : "ADVISORY",
    "url" : "https://nvd.nist.gov/vuln/detail/CVE-2021-37687"
  }, {
    "type" : "WEB",
    "url" : "https://github.com/tensorflow/tensorflow/commit/bb6a0383ed553c286f87ca88c207f6774d5c4a8f"
  }, {
    "type" : "WEB",
    "url" : "https://github.com/tensorflow/tensorflow/commit/eb921122119a6b6e470ee98b89e65d721663179d"
  }, {
    "type" : "WEB",
    "url" : "https://github.com/pypa/advisory-database/tree/main/vulns/tensorflow-cpu/PYSEC-2021-600.yaml"
  }, {
    "type" : "WEB",
    "url" : "https://github.com/pypa/advisory-database/tree/main/vulns/tensorflow-gpu/PYSEC-2021-798.yaml"
  }, {
    "type" : "WEB",
    "url" : "https://github.com/pypa/advisory-database/tree/main/vulns/tensorflow/PYSEC-2021-309.yaml"
  }, {
    "type" : "PACKAGE",
    "url" : "https://github.com/tensorflow/tensorflow"
  }, {
    "type" : "WEB",
    "url" : "https://github.com/tensorflow/tensorflow/blob/149562d49faa709ea80df1d99fc41d005b81082a/tensorflow/lite/kernels/gather.cc"
  }, {
    "type" : "WEB",
    "url" : "https://github.com/tensorflow/tensorflow/blob/149562d49faa709ea80df1d99fc41d005b81082a/tensorflow/lite/kernels/gather_nd.cc#L124"
  } ],
  "database_specific" : {
    "cwe_ids" : [ "CWE-125" ],
    "severity" : "MODERATE",
    "github_reviewed" : true,
    "github_reviewed_at" : "2021-08-24T17:49:13Z",
    "nvd_published_at" : "2021-08-12T23:15:00Z"
  }
}

CVSS V3

5.5 MODERATE

Attack Vector Local (AV:L)

Attack Complexity Low (AC:L)

Privileges Required Low (PR:L)

User Interaction None (UI:N)

Scope Unchanged (S:U)

Confidentiality High (C:H)

Integrity None (I:N)

Availability None (A:N)

Vector String

CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:N/A:N

Advisory Metadata

Source ID: GHSA-jwf9-w5xm-f437
Schema Version: 1.4.0
Published: 2021-08-25 14:40
Last Modified: 2024-11-13 21:21
Database Record File: advisories/github-reviewed/2021/08/GHSA-jwf9-w5xm-f437/GHSA-jwf9-w5xm-f437.json