CVE-2021-29591
Stack overflow due to looping TFLite subgraph
TensorFlow is an end-to-end open source platform for machine learning. TFlite graphs must not have loops between nodes. However, this condition was not checked and an attacker could craft models that would result in infinite loop during evaluation. In certain cases, the infinite loop would be replaced by stack overflow due to too many recursive calls. For example, the While implementation(https://github.com/tensorflow/tensorflow/blob/106d8f4fb89335a2c52d7c895b7a7485465ca8d9/tensorflow/lite/kernels/while.cc) could be tricked into a scneario where both the body and the loop subgraphs are the same. Evaluating one of the subgraphs means calling the Eval function for the other and this quickly exhaust all stack space. 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. 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.
Attack profile
How easy it is to exploit
- Attack vector
- Local
- Complexity
- Low
- Privileges needed
- Low
- User interaction
- None
What an attacker can affect
- Confidentiality
- High
- Integrity
- Low
- Availability
- High
Exploit likelihood
What each source reports
CVE List Score used
Stack overflow due to looping TFLite subgraph
TensorFlow is an end-to-end open source platform for machine learning. TFlite graphs must not have loops between nodes. However, this condition was not checked and an attacker could craft models that would result in infinite loop during evaluation. In certain cases, the infinite loop would be replaced by stack overflow due to too many recursive calls. For example, the While implementation(https://github.com/tensorflow/tensorflow/blob/106d8f4fb89335a2c52d7c895b7a7485465ca8d9/tensorflow/lite/kernels/while.cc) could be tricked into a scneario where both the body and the loop subgraphs are the same. Evaluating one of the subgraphs means calling the Eval function for the other and this quickly exhaust all stack space. 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. 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.
- Record
CVE-2021-29591- CVSS 3.1
- 7.3 High
CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:L/A:H - Weakness
- CWE-835
- Updated
- 2026-11-02
EUVDEUVD-2021-0332
TensorFlow is an end-to-end open source platform for machine learning. TFlite graphs must not have loops between nodes. However, this condition was not checked and an attacker could craft models that would result in infinite loop during evaluation. In certain cases, the infinite loop would be replaced by stack overflow due to too many recursive calls. For example, the While implementation(https://github.com/tensorflow/tensorflow/blob/106d8f4fb89335a2c52d7c895b7a7485465ca8d9/tensorflow/lite/kernels/while.cc) could be tricked into a scneario where both the body and the loop subgraphs are the same. Evaluating one of the subgraphs means calling the Eval function for the other and this quickly exhaust all stack space. 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. 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.
- Record
EUVD-2021-0332- CVSS 3.1
- 7.3 High
CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:L/A:H - Updated
- 2026-11-03
GitHub AdvisoriesGHSA-cwv3-863g-39vx
Stack overflow due to looping TFLite subgraph
Impact
TFlite graphs must not have loops between nodes. However, this condition was not checked and an attacker could craft models that would result in infinite loop during evaluation. In certain cases, the infinite loop would be replaced by stack overflow due to too many recursive calls.
For example, the While implementation could be tricked into a scneario where both the body and the loop subgraphs are the same. Evaluating one of the subgraphs means calling the Eval function for the other and this quickly exhaust all stack space.
Patches
We have patched the issue in GitHub commit 9c1dc920d8ffb4893d6c9d27d1f039607b326743 (for the While operator) and in GitHub commit c6173f5fe66cdbab74f4f869311fe6aae2ba35f4 (in general).
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 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.
- Record
GHSA-cwv3-863g-39vx- CVSS 4.0
- 7.0 High
CVSS:4.0/AV:L/AC:L/AT:N/PR:L/UI:N/VC:H/VI:L/VA:H/SC:N/SI:N/SA:N - CVSS 3.1
- 7.3 High
CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:L/A:H - Weakness
- CWE-674, CWE-835
- Updated
- 2026-11-02
NVD
TensorFlow is an end-to-end open source platform for machine learning. TFlite graphs must not have loops between nodes. However, this condition was not checked and an attacker could craft models that would result in infinite loop during evaluation. In certain cases, the infinite loop would be replaced by stack overflow due to too many recursive calls. For example, the While implementation(https://github.com/tensorflow/tensorflow/blob/106d8f4fb89335a2c52d7c895b7a7485465ca8d9/tensorflow/lite/kernels/while.cc) could be tricked into a scneario where both the body and the loop subgraphs are the same. Evaluating one of the subgraphs means calling the Eval function for the other and this quickly exhaust all stack space. 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. 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.
- Record
CVE-2021-29591- CVSS 3.1
- 7.3 High
CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:L/A:H - CVSS 3.1
- 7.8 High
CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H - CVSS 2.0
- 4.6 Medium
AV:L/AC:L/Au:N/C:P/I:P/A:P - Weakness
- CWE-835, CWE-674
- Updated
- 2026-11-02
OSV
TensorFlow is an end-to-end open source platform for machine learning. TFlite graphs must not have loops between nodes. However, this condition was not checked and an attacker could craft models that would result in infinite loop during evaluation. In certain cases, the infinite loop would be replaced by stack overflow due to too many recursive calls. For example, the While implementation(https://github.com/tensorflow/tensorflow/blob/106d8f4fb89335a2c52d7c895b7a7485465ca8d9/tensorflow/lite/kernels/while.cc) could be tricked into a scneario where both the body and the loop subgraphs are the same. Evaluating one of the subgraphs means calling the Eval function for the other and this quickly exhaust all stack space. 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. 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.
- Record
CVE-2021-29591- CVSS 3.1
- 7.8 High
CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H - Updated
- 2026-11-02
OSVGHSA-cwv3-863g-39vx
Stack overflow due to looping TFLite subgraph
Impact
TFlite graphs must not have loops between nodes. However, this condition was not checked and an attacker could craft models that would result in infinite loop during evaluation. In certain cases, the infinite loop would be replaced by stack overflow due to too many recursive calls.
For example, the While implementation could be tricked into a scneario where both the body and the loop subgraphs are the same. Evaluating one of the subgraphs means calling the Eval function for the other and this quickly exhaust all stack space.
Patches
We have patched the issue in GitHub commit 9c1dc920d8ffb4893d6c9d27d1f039607b326743 (for the While operator) and in GitHub commit c6173f5fe66cdbab74f4f869311fe6aae2ba35f4 (in general).
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 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.
- Record
GHSA-cwv3-863g-39vx- CVSS 4.0
- 7.0 High
CVSS:4.0/AV:L/AC:L/AT:N/PR:L/UI:N/VC:H/VI:L/VA:H/SC:N/SI:N/SA:N - CVSS 3.1
- 7.3 High
CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:L/A:H - Weakness
- CWE-674, CWE-835
- Updated
- 2026-11-02
Show 4 moreShow less
OSVBIT-tensorflow-2021-29591 No score
Stack overflow due to looping TFLite subgraph
TensorFlow is an end-to-end open source platform for machine learning. TFlite graphs must not have loops between nodes. However, this condition was not checked and an attacker could craft models that would result in infinite loop during evaluation. In certain cases, the infinite loop would be replaced by stack overflow due to too many recursive calls. For example, the While implementation(https://github.com/tensorflow/tensorflow/blob/106d8f4fb89335a2c52d7c895b7a7485465ca8d9/tensorflow/lite/kernels/while.cc) could be tricked into a scneario where both the body and the loop subgraphs are the same. Evaluating one of the subgraphs means calling the Eval function for the other and this quickly exhaust all stack space. 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. 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.
- Record
BIT-tensorflow-2021-29591- Updated
- 2026-11-02
OSVPYSEC-2021-228 No score
TensorFlow is an end-to-end open source platform for machine learning. TFlite graphs must not have loops between nodes. However, this condition was not checked and an attacker could craft models that would result in infinite loop during evaluation. In certain cases, the infinite loop would be replaced by stack overflow due to too many recursive calls. For example, the While implementation(https://github.com/tensorflow/tensorflow/blob/106d8f4fb89335a2c52d7c895b7a7485465ca8d9/tensorflow/lite/kernels/while.cc) could be tricked into a scneario where both the body and the loop subgraphs are the same. Evaluating one of the subgraphs means calling the Eval function for the other and this quickly exhaust all stack space. 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. 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.
- Record
PYSEC-2021-228- Updated
- 2026-11-02
OSVPYSEC-2021-519 No score
TensorFlow is an end-to-end open source platform for machine learning. TFlite graphs must not have loops between nodes. However, this condition was not checked and an attacker could craft models that would result in infinite loop during evaluation. In certain cases, the infinite loop would be replaced by stack overflow due to too many recursive calls. For example, the While implementation(https://github.com/tensorflow/tensorflow/blob/106d8f4fb89335a2c52d7c895b7a7485465ca8d9/tensorflow/lite/kernels/while.cc) could be tricked into a scneario where both the body and the loop subgraphs are the same. Evaluating one of the subgraphs means calling the Eval function for the other and this quickly exhaust all stack space. 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. 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.
- Record
PYSEC-2021-519- Updated
- 2026-11-02
OSVPYSEC-2021-717 No score
TensorFlow is an end-to-end open source platform for machine learning. TFlite graphs must not have loops between nodes. However, this condition was not checked and an attacker could craft models that would result in infinite loop during evaluation. In certain cases, the infinite loop would be replaced by stack overflow due to too many recursive calls. For example, the While implementation(https://github.com/tensorflow/tensorflow/blob/106d8f4fb89335a2c52d7c895b7a7485465ca8d9/tensorflow/lite/kernels/while.cc) could be tricked into a scneario where both the body and the loop subgraphs are the same. Evaluating one of the subgraphs means calling the Eval function for the other and this quickly exhaust all stack space. 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. 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.
- Record
PYSEC-2021-717- Updated
- 2026-11-02
Affected products
Affected packages
| Package | Affected | Fixed |
|---|---|---|
tensorflowPyPI |
before 2.1.4 | 2.1.4 |
tensorflowPyPI |
2.2.0 up to 2.2.3 | 2.2.3 |
tensorflowPyPI |
2.3.0 up to 2.3.3 | 2.3.3 |
tensorflowPyPI |
2.4.0 up to 2.4.2 | 2.4.2 |
tensorflow-cpuPyPI |
before 2.1.4 | 2.1.4 |
tensorflow-cpuPyPI |
2.2.0 up to 2.2.3 | 2.2.3 |
tensorflow-cpuPyPI |
2.3.0 up to 2.3.3 | 2.3.3 |
tensorflow-cpuPyPI |
2.4.0 up to 2.4.2 | 2.4.2 |
tensorflow-gpuPyPI |
before 2.1.4 | 2.1.4 |
tensorflow-gpuPyPI |
2.2.0 up to 2.2.3 | 2.2.3 |
Show 18 moreShow less
| Package | Affected | Fixed |
|---|---|---|
tensorflow-gpuPyPI |
2.3.0 up to 2.3.3 | 2.3.3 |
tensorflow-gpuPyPI |
2.4.0 up to 2.4.2 | 2.4.2 |
tensorflowBitnami |
before 2.1.4, 2.2.0 up to 2.2.3, 2.3.0 up to 2.3.3, 2.4.0 up to 2.4.2 | 2.1.42.2.32.3.32.4.2 |
tensorflowPyPI |
before 2.1.4, plus listed versions 0.12.00.12.11.0.01.0.11.1.0 Show 35 moreShow less1.10.01.10.11.11.01.12.01.12.21.12.31.13.11.13.21.14.01.15.01.15.21.15.31.15.41.15.51.2.01.2.11.3.01.4.01.4.11.5.01.5.11.6.01.7.01.7.11.8.01.9.02.0.02.0.12.0.22.0.32.0.42.1.02.1.12.1.22.1.3 |
2.1.4 |
tensorflowPyPI |
2.2.0 up to 2.2.3, plus listed versions 2.2.12.2.2 |
2.2.3 |
tensorflowPyPI |
2.3.0 up to 2.3.3, plus listed versions 2.3.12.3.2 |
2.3.3 |
tensorflowPyPI |
2.4.0 up to 2.4.2, plus listed versions 2.4.1 |
2.4.2 |
tensorflow-cpuPyPI |
before 2.1.4, plus listed versions 1.15.02.1.02.1.12.1.22.1.3 |
2.1.4 |
tensorflow-cpuPyPI |
2.2.0 up to 2.2.3, plus listed versions 2.2.12.2.2 |
2.2.3 |
tensorflow-cpuPyPI |
2.3.0 up to 2.3.3, plus listed versions 2.3.12.3.2 |
2.3.3 |
tensorflow-cpuPyPI |
2.4.0 up to 2.4.2, plus listed versions 2.4.1 |
2.4.2 |
tensorflow-gpuPyPI |
before 2.1.4, plus listed versions 0.12.00.12.11.0.01.0.11.1.0 Show 35 moreShow less1.10.01.10.11.11.01.12.01.12.21.12.31.13.11.13.21.14.01.15.01.15.21.15.31.15.41.15.51.2.01.2.11.3.01.4.01.4.11.5.01.5.11.6.01.7.01.7.11.8.01.9.02.0.02.0.12.0.22.0.32.0.42.1.02.1.12.1.22.1.3 |
2.1.4 |
tensorflow-gpuPyPI |
2.2.0 up to 2.2.3, plus listed versions 2.2.12.2.2 |
2.2.3 |
tensorflow-gpuPyPI |
2.3.0 up to 2.3.3, plus listed versions 2.3.12.3.2 |
2.3.3 |
tensorflow-gpuPyPI |
2.4.0 up to 2.4.2, plus listed versions 2.4.1 |
2.4.2 |
tensorflowPyPI |
before 2.1.4, 2.2.0 up to 2.2.3, 2.3.0 up to 2.3.3, 2.4.0 up to 2.4.2, plus listed versions 0.12.00.12.0rc00.12.0rc10.12.11.0.0 Show 45 moreShow less1.0.11.1.01.1.0rc01.1.0rc11.1.0rc21.10.01.10.0rc01.10.0rc11.10.11.11.01.11.0rc01.11.0rc11.11.0rc21.12.01.12.0rc01.12.0rc11.12.0rc21.12.21.12.31.13.0rc01.13.0rc11.13.0rc21.13.11.13.21.14.01.14.0rc01.14.0rc11.15.01.15.0rc01.15.0rc11.15.0rc21.15.0rc31.15.21.15.31.15.41.15.51.2.01.2.0rc01.2.0rc11.2.0rc21.2.11.3.01.3.0rc01.3.0rc11.3.0rc2 |
2.1.42.2.32.3.32.4.2 |
tensorflow-cpuPyPI |
before 2.1.4, 2.2.0 up to 2.2.3, 2.3.0 up to 2.3.3, 2.4.0 up to 2.4.2, plus listed versions 1.15.02.1.02.1.12.1.22.1.3 Show 5 moreShow less2.2.12.2.22.3.12.3.22.4.1 |
2.1.42.2.32.3.32.4.2 |
tensorflow-gpuPyPI |
before 2.1.4, 2.2.0 up to 2.2.3, 2.3.0 up to 2.3.3, 2.4.0 up to 2.4.2, plus listed versions 0.12.00.12.11.0.01.0.11.1.0 Show 40 moreShow less1.10.01.10.11.11.01.12.01.12.21.12.31.13.11.13.21.14.01.15.01.15.21.15.31.15.41.15.51.2.01.2.11.3.01.4.01.4.11.5.01.5.11.6.01.7.01.7.11.8.01.9.02.0.02.0.12.0.22.0.32.0.42.1.02.1.12.1.22.1.32.2.12.2.22.3.12.3.22.4.1 |
2.1.42.2.32.3.32.4.2 |
History
| Recorded | Changed | Sources |
|---|---|---|
| aliases | EUVD | |
| first indexed | CVE List, GitHub Advisories, NVD, OSV |
Dates are when Every CVE recorded each change. Upstream dates may differ.
In short
CVE-2021-29591 is a high vulnerability (CVSS 7.3) in Google Tensorflow, published May 14, 2021. No source lists public exploit code. EPSS puts the chance of exploitation in the next 30 days at 0.26%, the 18th percentile. Fixed in tensorflow 2.1.4.
References
github.com/tensorflow/tensorflow/security/advisories/GHSA-cwv3-863g-39vx
nvd.nist.gov/vuln/detail/CVE-2021-29591
Left out here: 45 versions. The JSON record has them.