CVE-2021-29540: Tensorflow Heap Buffer Overflow Vulnerability

LOW
Advisory Identifier: GHSA-xgc3-m89p-vr3x
Share
CVSS 2.5
CWE-120 CWE-787
Published
Advisory Summary

CVE-2021-29540 is a LOW (CVSS 2.5) heap buffer overflow vulnerability (CWE-120: Buffer Copy without Checking Size of Input ('Classic Buffer Overflow')) affecting Tensorflow libraries in the PyPI ecosystem. The issue involves heap buffer overflow in `conv2dbackpropfilter`. Remediation is available by upgrading affected packages to version 2.1.4, 2.2.3, 2.3.3, 2.4.2.

CVE-2021-29540 at a Glance

CVE Identifier CVE-2021-29540
GHSA Identifier GHSA-xgc3-m89p-vr3x
Severity LOW
CVSS v3.1 Base Score 2.5
CWE Classification CWE-120: Buffer Copy without Checking Size of Input ('Classic Buffer Overflow')
Affected Product Tensorflow
Package Ecosystem PyPI
Published Date
Last Updated

Security Impact

CVSS: 2.5

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

Attack Vector Local (AV:L)
Attack Complexity High (AC:H)
Privileges Required Low (PR:L)
User Interaction None (UI:N)
Scope Unchanged (S:U)
Confidentiality None (C:N)
Integrity None (I:N)
Availability Low (A:L)

How to Fix CVE-2021-29540

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: 2.1.4 
pip install tensorflow>=2.1.4
tensorflow fixed: 2.2.3 
pip install tensorflow>=2.2.3
tensorflow fixed: 2.3.3 
pip install tensorflow>=2.3.3
tensorflow fixed: 2.4.2 
pip install tensorflow>=2.4.2
tensorflow-cpu fixed: 2.1.4 
pip install tensorflow-cpu>=2.1.4
tensorflow-cpu fixed: 2.2.3 
pip install tensorflow-cpu>=2.2.3
tensorflow-cpu fixed: 2.3.3 
pip install tensorflow-cpu>=2.3.3
tensorflow-cpu fixed: 2.4.2 
pip install tensorflow-cpu>=2.4.2
tensorflow-gpu fixed: 2.1.4 
pip install tensorflow-gpu>=2.1.4
tensorflow-gpu fixed: 2.2.3 
pip install tensorflow-gpu>=2.2.3
tensorflow-gpu fixed: 2.3.3 
pip install tensorflow-gpu>=2.3.3
tensorflow-gpu fixed: 2.4.2 
pip install tensorflow-gpu>=2.4.2
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-2021-29540, 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 2.1.4), 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-120: Buffer Copy without Checking Size of Input ('Classic Buffer Overflow'), where the product copies an input buffer to an output buffer without verifying that the size of the input buffer is less than the size of the output 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 An attacker can cause a heap buffer overflow to occur in `Conv2DBackpropFilter`: ```python import tensorflow as tf input_tensor = tf.constant([386.078431372549, 386.07843139643234], shape=[1, 1, 1, 2], dtype=tf.float32) filter_sizes = tf.constant([1, 1, 1, 1], shape=[4], dtype=tf.int32) out_backprop = tf.constant([386.078431372549], shape=[1, 1, 1, 1], dtype=tf.float32) tf.raw_ops.Conv2DBackpropFilter( input=input_tensor, filter_sizes=filter_sizes, out_backprop=out_backprop, strides=[1, 66, 49, 1], use_cudnn_on_gpu=True, padding='VALID', explicit_paddings=[], data_format='NHWC', dilations=[1, 1, 1, 1] ) ``` Alternatively, passing empty tensors also results in similar behavior: ```python import tensorflow as tf input_tensor = tf.constant([], shape=[0, 1, 1, 5], dtype=tf.float32) filter_sizes = tf.constant([3, 8, 1, 1], shape=[4], dtype=tf.int32) out_backprop = tf.constant([], shape=[0, 1, 1, 1], dtype=tf.float32) tf.raw_ops.Conv2DBackpropFilter( input=input_tensor, filter_sizes=filter_sizes, out_backprop=out_backprop, strides=[1, 66, 49, 1], use_cudnn_on_gpu=True, padding='VALID', explicit_paddings=[], data_format='NHWC', dilations=[1, 1, 1, 1] ) ``` This is because the [implementation](https://github.com/tensorflow/tensorflow/blob/1b0296c3b8dd9bd948f924aa8cd62f87dbb7c3da/tensorflow/core/kernels/conv_grad_filter_ops.cc#L495-L497) computes the size of the filter tensor but does not validate that it matches the number of elements in `filter_sizes`. Later, when reading/writing to this buffer, code uses the value computed here, instead of the number of elements in the tensor. ### Patches We have patched the issue in GitHub commit [c570e2ecfc822941335ad48f6e10df4e21f11c96](https://github.com/tensorflow/tensorflow/commit/c570e2ecfc822941335ad48f6e10df4e21f11c96). 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. ### 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 and Ying Wang of Baidu X-Team.

CVE-2021-29540 Affected Tensorflow Versions

12 component(s)
Fixed Versions:
2.1.42.2.32.3.32.4.2
Product / Package Ecosystem Affected Versions Fixed Version
tensorflow
pypi/tensorflow
PyPI
< 2.1.4
2.1.4
tensorflow
pypi/tensorflow
PyPI
≥ 2.2.0
< 2.2.3
2.2.3
tensorflow
pypi/tensorflow
PyPI
≥ 2.3.0
< 2.3.3
2.3.3
tensorflow
pypi/tensorflow
PyPI
≥ 2.4.0
< 2.4.2
2.4.2
tensorflow-cpu
pypi/tensorflow-cpu
PyPI
< 2.1.4
2.1.4
tensorflow-cpu
pypi/tensorflow-cpu
PyPI
≥ 2.2.0
< 2.2.3
2.2.3
tensorflow-cpu
pypi/tensorflow-cpu
PyPI
≥ 2.3.0
< 2.3.3
2.3.3
tensorflow-cpu
pypi/tensorflow-cpu
PyPI
≥ 2.4.0
< 2.4.2
2.4.2
tensorflow-gpu
pypi/tensorflow-gpu
PyPI
< 2.1.4
2.1.4
tensorflow-gpu
pypi/tensorflow-gpu
PyPI
≥ 2.2.0
< 2.2.3
2.2.3
tensorflow-gpu
pypi/tensorflow-gpu
PyPI
≥ 2.3.0
< 2.3.3
2.3.3
tensorflow-gpu
pypi/tensorflow-gpu
PyPI
≥ 2.4.0
< 2.4.2
2.4.2

References

GitHub Advisory / Wiki ADVISORY
https://github.com/tensorflow/tensorflow/security/advisories/GHSA-xgc3-m89p-vr3x
NVD — CVE-2021-29540 NVD
https://nvd.nist.gov/vuln/detail/CVE-2021-29540
Security Fix Commit FIX COMMIT
https://github.com/tensorflow/tensorflow/commit/c570e2ecfc822941335ad48f6e10df4e21f11c96
Tensorflow Source Repository REPOSITORY
https://github.com/pypa/advisory-database/tree/main/vulns/tensorflow-cpu/PYSEC-2021-468.yaml
Tensorflow Source Repository REPOSITORY
https://github.com/pypa/advisory-database/tree/main/vulns/tensorflow-gpu/PYSEC-2021-666.yaml
Tensorflow Source Repository REPOSITORY
https://github.com/pypa/advisory-database/tree/main/vulns/tensorflow/PYSEC-2021-177.yaml
Tensorflow Source Repository REPOSITORY
https://github.com/tensorflow/tensorflow

Frequently Asked Questions

What is CVE-2021-29540?
CVE-2021-29540 is a security vulnerability in Tensorflow involving Heap buffer overflow in `Conv2DBackpropFilter` under CWE-120: Buffer Copy without Checking Size of Input ('Classic Buffer Overflow').
Which Tensorflow versions are affected?
Affected versions include releases prior to 2.1.4, 2.2.3, 2.3.3, 2.4.2 across affected packages: tensorflow, tensorflow-cpu, tensorflow-gpu.
What version fixes CVE-2021-29540?
Upgrading to 2.1.4, 2.2.3, 2.3.3, 2.4.2 (or the latest patched release for your respective branch) resolves the vulnerability.
Is tensorflow affected?
Yes, tensorflow is affected in releases prior to 2.1.4, 2.2.3, 2.3.3, 2.4.2. Upgrade your dependency to 2.1.4, 2.2.3, 2.3.3, 2.4.2 or later.
Is tensorflow-cpu affected?
Yes, tensorflow-cpu is affected in releases prior to 2.1.4, 2.2.3, 2.3.3, 2.4.2. Upgrade your dependency to 2.1.4, 2.2.3, 2.3.3, 2.4.2 or later.
What is the CVSS score of CVE-2021-29540?
CVE-2021-29540 has a CVSS v3.1 base score of 2.5 (LOW) with vector string CVSS:3.1/AV:L/AC:H/PR:L/UI:N/S:U/C:N/I:N/A:L.
What CWE is associated with CVE-2021-29540?
CVE-2021-29540 is classified under CWE-120: Buffer Copy without Checking Size of Input ('Classic Buffer Overflow') (The product copies an input buffer to an output buffer without verifying that the size of the input buffer is less than the size of the output buffer.).
How can I check PyPI dependencies for CVE-2021-29540?
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-xgc3-m89p-vr3x",
  "modified" : "2024-10-30T23:28:15Z",
  "published" : "2021-05-21T14:23:09Z",
  "aliases" : [ "CVE-2021-29540" ],
  "summary" : "Heap buffer overflow in `Conv2DBackpropFilter`",
  "details" : "### Impact\nAn attacker can cause a heap buffer overflow to occur in `Conv2DBackpropFilter`:\n\n```python\nimport tensorflow as tf\n\ninput_tensor = tf.constant([386.078431372549, 386.07843139643234],\n                           shape=[1, 1, 1, 2], dtype=tf.float32)\nfilter_sizes = tf.constant([1, 1, 1, 1], shape=[4], dtype=tf.int32)\nout_backprop = tf.constant([386.078431372549], shape=[1, 1, 1, 1],\n                           dtype=tf.float32)\n  \ntf.raw_ops.Conv2DBackpropFilter(\n  input=input_tensor,\n  filter_sizes=filter_sizes,\n  out_backprop=out_backprop,\n  strides=[1, 66, 49, 1],\n  use_cudnn_on_gpu=True,\n  padding='VALID',\n  explicit_paddings=[],\n  data_format='NHWC',\n  dilations=[1, 1, 1, 1]\n)\n```\n\nAlternatively, passing empty tensors also results in similar behavior: \n\n```python\nimport tensorflow as tf\n\ninput_tensor = tf.constant([], shape=[0, 1, 1, 5], dtype=tf.float32)\nfilter_sizes = tf.constant([3, 8, 1, 1], shape=[4], dtype=tf.int32)\nout_backprop = tf.constant([], shape=[0, 1, 1, 1], dtype=tf.float32)\n\ntf.raw_ops.Conv2DBackpropFilter(\n  input=input_tensor,\n  filter_sizes=filter_sizes, \n  out_backprop=out_backprop,\n  strides=[1, 66, 49, 1], \n  use_cudnn_on_gpu=True,\n  padding='VALID',\n  explicit_paddings=[],\n  data_format='NHWC',\n  dilations=[1, 1, 1, 1]\n)\n```\n\nThis is because the [implementation](https://github.com/tensorflow/tensorflow/blob/1b0296c3b8dd9bd948f924aa8cd62f87dbb7c3da/tensorflow/core/kernels/conv_grad_filter_ops.cc#L495-L497) computes the size of the filter tensor but does not validate that it matches the number of elements in `filter_sizes`. Later, when reading/writing to this buffer, code uses the value computed here, instead of the number of elements in the tensor.\n\n### Patches \nWe have patched the issue in GitHub commit [c570e2ecfc822941335ad48f6e10df4e21f11c96](https://github.com/tensorflow/tensorflow/commit/c570e2ecfc822941335ad48f6e10df4e21f11c96).\n\nThe 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.\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 and Ying Wang of Baidu X-Team.",
  "severity" : [ {
    "type" : "CVSS_V3",
    "score" : "CVSS:3.1/AV:L/AC:H/PR:L/UI:N/S:U/C:N/I:N/A:L"
  }, {
    "type" : "CVSS_V4",
    "score" : "CVSS:4.0/AV:L/AC:L/AT:P/PR:L/UI:N/VC:N/VI:N/VA:L/SC:N/SI:N/SA:N"
  } ],
  "affected" : [ {
    "package" : {
      "ecosystem" : "PyPI",
      "name" : "tensorflow"
    },
    "ranges" : [ {
      "type" : "ECOSYSTEM",
      "events" : [ {
        "introduced" : "0"
      }, {
        "fixed" : "2.1.4"
      } ]
    } ]
  }, {
    "package" : {
      "ecosystem" : "PyPI",
      "name" : "tensorflow"
    },
    "ranges" : [ {
      "type" : "ECOSYSTEM",
      "events" : [ {
        "introduced" : "2.2.0"
      }, {
        "fixed" : "2.2.3"
      } ]
    } ]
  }, {
    "package" : {
      "ecosystem" : "PyPI",
      "name" : "tensorflow"
    },
    "ranges" : [ {
      "type" : "ECOSYSTEM",
      "events" : [ {
        "introduced" : "2.3.0"
      }, {
        "fixed" : "2.3.3"
      } ]
    } ]
  }, {
    "package" : {
      "ecosystem" : "PyPI",
      "name" : "tensorflow"
    },
    "ranges" : [ {
      "type" : "ECOSYSTEM",
      "events" : [ {
        "introduced" : "2.4.0"
      }, {
        "fixed" : "2.4.2"
      } ]
    } ]
  }, {
    "package" : {
      "ecosystem" : "PyPI",
      "name" : "tensorflow-cpu"
    },
    "ranges" : [ {
      "type" : "ECOSYSTEM",
      "events" : [ {
        "introduced" : "0"
      }, {
        "fixed" : "2.1.4"
      } ]
    } ]
  }, {
    "package" : {
      "ecosystem" : "PyPI",
      "name" : "tensorflow-cpu"
    },
    "ranges" : [ {
      "type" : "ECOSYSTEM",
      "events" : [ {
        "introduced" : "2.2.0"
      }, {
        "fixed" : "2.2.3"
      } ]
    } ]
  }, {
    "package" : {
      "ecosystem" : "PyPI",
      "name" : "tensorflow-cpu"
    },
    "ranges" : [ {
      "type" : "ECOSYSTEM",
      "events" : [ {
        "introduced" : "2.3.0"
      }, {
        "fixed" : "2.3.3"
      } ]
    } ]
  }, {
    "package" : {
      "ecosystem" : "PyPI",
      "name" : "tensorflow-cpu"
    },
    "ranges" : [ {
      "type" : "ECOSYSTEM",
      "events" : [ {
        "introduced" : "2.4.0"
      }, {
        "fixed" : "2.4.2"
      } ]
    } ]
  }, {
    "package" : {
      "ecosystem" : "PyPI",
      "name" : "tensorflow-gpu"
    },
    "ranges" : [ {
      "type" : "ECOSYSTEM",
      "events" : [ {
        "introduced" : "0"
      }, {
        "fixed" : "2.1.4"
      } ]
    } ]
  }, {
    "package" : {
      "ecosystem" : "PyPI",
      "name" : "tensorflow-gpu"
    },
    "ranges" : [ {
      "type" : "ECOSYSTEM",
      "events" : [ {
        "introduced" : "2.2.0"
      }, {
        "fixed" : "2.2.3"
      } ]
    } ]
  }, {
    "package" : {
      "ecosystem" : "PyPI",
      "name" : "tensorflow-gpu"
    },
    "ranges" : [ {
      "type" : "ECOSYSTEM",
      "events" : [ {
        "introduced" : "2.3.0"
      }, {
        "fixed" : "2.3.3"
      } ]
    } ]
  }, {
    "package" : {
      "ecosystem" : "PyPI",
      "name" : "tensorflow-gpu"
    },
    "ranges" : [ {
      "type" : "ECOSYSTEM",
      "events" : [ {
        "introduced" : "2.4.0"
      }, {
        "fixed" : "2.4.2"
      } ]
    } ]
  } ],
  "references" : [ {
    "type" : "WEB",
    "url" : "https://github.com/tensorflow/tensorflow/security/advisories/GHSA-xgc3-m89p-vr3x"
  }, {
    "type" : "ADVISORY",
    "url" : "https://nvd.nist.gov/vuln/detail/CVE-2021-29540"
  }, {
    "type" : "WEB",
    "url" : "https://github.com/tensorflow/tensorflow/commit/c570e2ecfc822941335ad48f6e10df4e21f11c96"
  }, {
    "type" : "WEB",
    "url" : "https://github.com/pypa/advisory-database/tree/main/vulns/tensorflow-cpu/PYSEC-2021-468.yaml"
  }, {
    "type" : "WEB",
    "url" : "https://github.com/pypa/advisory-database/tree/main/vulns/tensorflow-gpu/PYSEC-2021-666.yaml"
  }, {
    "type" : "WEB",
    "url" : "https://github.com/pypa/advisory-database/tree/main/vulns/tensorflow/PYSEC-2021-177.yaml"
  }, {
    "type" : "PACKAGE",
    "url" : "https://github.com/tensorflow/tensorflow"
  } ],
  "database_specific" : {
    "cwe_ids" : [ "CWE-120", "CWE-787" ],
    "severity" : "LOW",
    "github_reviewed" : true,
    "github_reviewed_at" : "2021-05-18T22:05:40Z",
    "nvd_published_at" : "2021-05-14T20:15:00Z"
  }
}