Segfault on strings tensors with mistmatched dimensions in TensorFlow
Key facts
- Attack vector
- Local
- Privileges required
- Low
- User interaction
- None
- Attack complexity
- Low
- Impact type
- —
- Confidentiality · integrity · availability
- None · None · High
- Fix status
- Fixed
Attack profile from CVSS 3.1 · CNA. Fix status from CNA and GitHub.
Summary
TensorFlow is an end-to-end open source platform for machine learning. In affected versions under certain conditions, Go code can trigger a segfault in string deallocation. For string tensors, `C.TF_TString_Dealloc` is called during garbage collection within a finalizer function. However, tensor structure isn't checked until encoding to avoid a performance penalty. The current method for dealloc assumes that encoding succeeded, but segfaults when a string tensor is garbage collected whose encoding failed (e.g., due to mismatched dimensions). To fix this, the call to set the finalizer function is deferred until `NewTensor` returns and, if encoding failed for a string tensor, deallocs are determined based on bytes written. We have patched the issue in GitHub commit 8721ba96e5760c229217b594f6d2ba332beedf22. The fix will be included in TensorFlow 2.6.0. We will also cherrypick this commit on TensorFlow 2.5.1, which is the other affected version.
Why it matters now
SSVC decision points
Not available: CISA has not published SSVC decision points for this record.
Affected products
| Vendor | Product | Affected versions | Fixed in | Status | Source |
|---|---|---|---|---|---|
| tensorflow | 2.5.0 to before 2.5.1 | 2.5.1 | Patch available | CNA | |
| — | tensorflow (PyPI) | 2.5.0rc0 to before 2.5.1 | 2.5.1 | Patch available | GHSA-cmgw-8vpc-rc59 |
| — | tensorflow-cpu (PyPI) | 2.5.0rc0 to before 2.5.1 | 2.5.1 | Patch available | GHSA-cmgw-8vpc-rc59 |
| — | tensorflow-gpu (PyPI) | 2.5.0rc0 to before 2.5.1 | 2.5.1 | Patch available | GHSA-cmgw-8vpc-rc59 |
- tensorflow VendorGoogle Affected 2.5.0 to before 2.5.1 Fixed in 2.5.1 Patch available CNA
- tensorflow (PyPI) Vendor— Affected 2.5.0rc0 to before 2.5.1 Fixed in 2.5.1 Patch available GHSA-cmgw-8vpc-rc59
- tensorflow-cpu (PyPI) Vendor— Affected 2.5.0rc0 to before 2.5.1 Fixed in 2.5.1 Patch available GHSA-cmgw-8vpc-rc59
- tensorflow-gpu (PyPI) Vendor— Affected 2.5.0rc0 to before 2.5.1 Fixed in 2.5.1 Patch available GHSA-cmgw-8vpc-rc59
Remediation
- Fix · CNA tensorflow (tensorflow) Fixed in 2.5.1
- Fix · GHSA-cmgw-8vpc-rc59 tensorflow (PyPI) Fixed in 2.5.1 Advisory →
- Fix · GHSA-cmgw-8vpc-rc59 tensorflow-cpu (PyPI) Fixed in 2.5.1 Advisory →
- Fix · GHSA-cmgw-8vpc-rc59 tensorflow-gpu (PyPI) Fixed in 2.5.1 Advisory →
Timeline
- 2026-10-02 Threatist record created from CVE-2021-37692
- 2026-10-03 Alias added: GHSA-cmgw-8vpc-rc59
References
Patch 2
Advisory 2
Other 4
- https://github.com/pypa/advisory-database/tree/main/vulns/tensorflow-cpu/PYSEC-2021-605.yaml
- https://github.com/pypa/advisory-database/tree/main/vulns/tensorflow-gpu/PYSEC-2021-803.yaml
- https://github.com/pypa/advisory-database/tree/main/vulns/tensorflow/PYSEC-2021-314.yaml
- https://github.com/tensorflow/tensorflow
More Google Tensorflow vulnerabilities
- TI-2022-20241 FractionalMaxPool and FractionalAVGPool heap out-of-bounds acess in Tensorflow CVSS 3.1CRITICAL 9.8
- TI-2022-07640 Integer overflow in Tensorflow CVSS 3.1CRITICAL 9.8
- TI-2022-07627 Out of bounds read and write in Tensorflow CVSS 3.1HIGH 8.8
- TI-2022-07619 Out of bounds write in Tensorflow CVSS 3.1HIGH 8.8
- TI-2022-07613 Read and Write outside of bounds in TFLite CVSS 3.1HIGH 8.8
- TI-2022-07612 Integer overflow in TFLite CVSS 3.1HIGH 8.8
Sources
- CVE-2021-37692 · CVE Program
- CVE-2021-37692 · NVD
- FIRST EPSS
- GHSA-cmgw-8vpc-rc59 · GitHub Advisory Database · CC BY 4.0
This product uses the NVD API but is not endorsed or certified by the NVD. Sources and licences