{"id":"CVE-2021-29544","aliases":["GHSA-6g85-3hm8-83f9","BIT-tensorflow-2021-29544","PYSEC-2021-181","PYSEC-2021-472","PYSEC-2021-670"],"title":"CHECK-fail in `QuantizeAndDequantizeV4Grad`","summary":"CHECK-fail in `QuantizeAndDequantizeV4Grad`","severity":"low","cvss":2.5,"cvssVector":"CVSS:3.1/AV:L/AC:H/PR:L/UI:N/S:U/C:N/I:N/A:L","vendor":"tensorflow","product":"tensorflow","ecosystem":"pip","affected":["tensorflow >= 2.4.0, < 2.4.2","tensorflow-cpu >= 2.4.0, < 2.4.2","tensorflow-gpu >= 2.4.0, < 2.4.2"],"patched":["tensorflow 2.4.2","tensorflow-cpu 2.4.2","tensorflow-gpu 2.4.2"],"published":"2021-05-21","updated":"2026-07-08","source":"OSV","sourceUrl":"https://osv.dev/vulnerability/GHSA-6g85-3hm8-83f9","references":[{"url":"https://github.com/tensorflow/tensorflow/security/advisories/GHSA-6g85-3hm8-83f9"},{"url":"https://nvd.nist.gov/vuln/detail/CVE-2021-29544"},{"url":"https://github.com/tensorflow/tensorflow/commit/20431e9044cf2ad3c0323c34888b192f3289af6b"},{"url":"https://github.com/pypa/advisory-database/tree/main/vulns/tensorflow-cpu/PYSEC-2021-472.yaml"},{"url":"https://github.com/pypa/advisory-database/tree/main/vulns/tensorflow-gpu/PYSEC-2021-670.yaml"},{"url":"https://github.com/pypa/advisory-database/tree/main/vulns/tensorflow/PYSEC-2021-181.yaml"},{"url":"https://github.com/tensorflow/tensorflow"},{"url":"https://github.com/tensorflow/tensorflow/blob/95078c145b5a7a43ee046144005f733092756ab5/tensorflow/core/kernels/quantize_and_dequantize_op.cc#L162-L163"},{"url":"https://github.com/tensorflow/tensorflow/blob/95078c145b5a7a43ee046144005f733092756ab5/tensorflow/core/kernels/quantize_and_dequantize_op.h#L295-L306"}],"tags":["osv","pip"],"epss":0.0031,"epssPercentile":0.24025,"ingestedAt":"2026-07-08T18:25:46.461Z","slug":"CVE-2021-29544","body":"## Overview\n\n### Impact\nAn attacker can trigger a denial of service via a `CHECK`-fail in `tf.raw_ops.QuantizeAndDequantizeV4Grad`:\n\n```python\nimport tensorflow as tf\n\ngradient_tensor = tf.constant([0.0], shape=[1])\ninput_tensor = tf.constant([0.0], shape=[1])\ninput_min = tf.constant([[0.0]], shape=[1, 1])\ninput_max = tf.constant([[0.0]], shape=[1, 1])\n\ntf.raw_ops.QuantizeAndDequantizeV4Grad(\n  gradients=gradient_tensor, input=input_tensor,\n  input_min=input_min, input_max=input_max, axis=0)\n```                     \n                        \nThis is because the [implementation](https://github.com/tensorflow/tensorflow/blob/95078c145b5a7a43ee046144005f733092756ab5/tensorflow/core/kernels/quantize_and_dequantize_op.cc#L162-L163) does not validate the rank of the `input_*` tensors. In turn, this results in the tensors being passes as they are to [`QuantizeAndDequantizePerChannelGradientImpl`](https://github.com/tensorflow/tensorflow/blob/95078c145b5a7a43ee046144005f733092756ab5/tensorflow/core/kernels/quantize_and_dequantize_op.h#L295-L306):\n\n```cc \ntemplate <typename Device, typename T>\nstruct QuantizeAndDequantizePerChannelGradientImpl {\n  static void Compute(const Device& d,\n                      typename TTypes<T, 3>::ConstTensor gradient,\n                      typename TTypes<T, 3>::ConstTensor input,\n                      const Tensor* input_min_tensor,\n                      const Tensor* input_max_tensor,\n                      typename TTypes<T, 3>::Tensor input_backprop,\n                      typename TTypes<T>::Flat input_min_backprop,\n                      typename TTypes<T>::Flat input_max_backprop) {\n    ...\n    auto input_min = input_min_tensor->vec<T>();\n    auto input_max = input_max_tensor->vec<T>();\n    ...\n}\n```\n\nHowever, the `vec<T>` method, requires the rank to 1 and triggers a `CHECK` failure otherwise.\n\n### Patches\nWe have patched the issue in GitHub commit [20431e9044cf2ad3c0323c34888b192f3289af6b](https://github.com/tensorflow/tensorflow/commit/20431e9044cf2ad3c0323c34888b192f3289af6b).\n\nThe fix will be included in TensorFlow 2.5.0. We will also cherrypick this commit on TensorFlow 2.4.2 as this is the only other affected version.\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.\n\n## Affected packages\n\n- `tensorflow >= 2.4.0, < 2.4.2`\n- `tensorflow-cpu >= 2.4.0, < 2.4.2`\n- `tensorflow-gpu >= 2.4.0, < 2.4.2`\n\n## Remediation\n\nUpgrade to a patched release:\n\n- `tensorflow 2.4.2`\n- `tensorflow-cpu 2.4.2`\n- `tensorflow-gpu 2.4.2`","depth":"sunlit","depthScore":14,"depthScoreParts":{"impact":13.8,"likelihood":0.1,"exploitation":0,"ransomware":0},"changes":[]}