{"id":"CVE-2021-37645","aliases":["GHSA-9w2p-5mgw-p94c","BIT-tensorflow-2021-37645","PYSEC-2021-267","PYSEC-2021-558","PYSEC-2021-756"],"title":"Integer overflow due to conversion to unsigned","summary":"Integer overflow due to conversion to unsigned","severity":"medium","cvss":5.5,"cvssVector":"CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H","vendor":"tensorflow","product":"tensorflow","ecosystem":"pip","affected":["tensorflow < 2.4.3","tensorflow >= 2.5.0, < 2.5.1","tensorflow-cpu < 2.4.3","tensorflow-cpu >= 2.5.0, < 2.5.1","tensorflow-gpu < 2.4.3","tensorflow-gpu >= 2.5.0, < 2.5.1"],"patched":["tensorflow 2.4.3","tensorflow 2.5.1","tensorflow-cpu 2.4.3","tensorflow-cpu 2.5.1","tensorflow-gpu 2.4.3","tensorflow-gpu 2.5.1"],"published":"2021-08-25","updated":"2026-07-08","source":"OSV","sourceUrl":"https://osv.dev/vulnerability/GHSA-9w2p-5mgw-p94c","references":[{"url":"https://github.com/tensorflow/tensorflow/security/advisories/GHSA-9w2p-5mgw-p94c"},{"url":"https://nvd.nist.gov/vuln/detail/CVE-2021-37645"},{"url":"https://github.com/tensorflow/tensorflow/commit/96f364a1ca3009f98980021c4b32be5fdcca33a1"},{"url":"https://github.com/pypa/advisory-database/tree/main/vulns/tensorflow-cpu/PYSEC-2021-558.yaml"},{"url":"https://github.com/pypa/advisory-database/tree/main/vulns/tensorflow-gpu/PYSEC-2021-756.yaml"},{"url":"https://github.com/pypa/advisory-database/tree/main/vulns/tensorflow/PYSEC-2021-267.yaml"},{"url":"https://github.com/tensorflow/tensorflow"}],"tags":["osv","pip"],"epss":0.00152,"epssPercentile":0.04761,"ingestedAt":"2026-07-08T18:25:48.100Z","slug":"CVE-2021-37645","body":"## Overview\n\n### Impact\nThe implementation of `tf.raw_ops.QuantizeAndDequantizeV4Grad` is vulnerable to an integer overflow issue caused by converting a signed integer value to an unsigned one and then allocating memory based on this value.\n\n```python\nimport tensorflow as tf\n\ntf.raw_ops.QuantizeAndDequantizeV4Grad(\n  gradients=[1.0,2.0],\n  input=[1.0,1.0],\n  input_min=[0.0],\n  input_max=[10.0],\n  axis=-100)\n```\n\nThe [implementation](https://github.com/tensorflow/tensorflow/blob/8d72537c6abf5a44103b57b9c2e22c14f5f49698/tensorflow/core/kernels/quantize_and_dequantize_op.cc#L126) uses the `axis` value as the size argument to `absl::InlinedVector` constructor. But, the constructor uses an unsigned type for the argument, so the implicit conversion transforms the negative value to a large integer.\n\n### Patches\nWe have patched the issue in GitHub commit [96f364a1ca3009f98980021c4b32be5fdcca33a1](https://github.com/tensorflow/tensorflow/commit/96f364a1ca3009f98980021c4b32be5fdcca33a1).\n\nThe fix will be included in TensorFlow 2.6.0. We will also cherrypick this commit on TensorFlow 2.5.1, and TensorFlow 2.4.3, 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 members of the Aivul Team from Qihoo 360.\n\n## Affected packages\n\n- `tensorflow < 2.4.3`\n- `tensorflow >= 2.5.0, < 2.5.1`\n- `tensorflow-cpu < 2.4.3`\n- `tensorflow-cpu >= 2.5.0, < 2.5.1`\n- `tensorflow-gpu < 2.4.3`\n- `tensorflow-gpu >= 2.5.0, < 2.5.1`\n\n## Remediation\n\nUpgrade to a patched release:\n\n- `tensorflow 2.4.3`\n- `tensorflow 2.5.1`\n- `tensorflow-cpu 2.4.3`\n- `tensorflow-cpu 2.5.1`\n- `tensorflow-gpu 2.4.3`\n- `tensorflow-gpu 2.5.1`","depth":"sunlit","depthScore":30,"depthScoreParts":{"impact":30.3,"likelihood":0,"exploitation":0,"ransomware":0},"changes":[]}