CVE-2021-37679: TensorFlow CPU - Uninitialized Memory Leak in tf.map_fn
Summary
CVE-2021-37679: TensorFlow is an end-to-end open source platform for machine learning. Affects tensorflow-cpu across 3 version ranges.
>= 2.3.0, < 2.3.4
>= 2.4.0, < 2.4.3
< 4e2565483d0ffcadc719bd44893fb7f609bb5f12
Fixed in 2.3.4 or 2.4.3, depending on the release line you run. How to fix ↓
Also tracked for
- tensorflow · Bitnami
- tensorflow · PyPI
- tensorflow · PyPI
- tensorflow-gpu · PyPI
The same CVE-2021-37679 is tracked separately for each ecosystem and vendor. Its fix information is the same; the package and affected versions differ.
Check if this vulnerability affects your application
Upload a manifest or SBOM and we'll show whether you're affected, plus every other risky package in the same scan.
No signup required · Results in about 60 seconds
Habilume
Habit tracker with a companion that grows with you
Complete habits, earn Lume Points, and watch your Lume grow as you stay consistent.
Background
TensorFlow is an end-to-end open source platform for machine learning. It provides a comprehensive ecosystem of tools, libraries, and resources for developing and deploying machine learning models. TensorFlow is widely used in various domains, including research, industry, and academia. The vulnerability described here affects the TensorFlow CPU package, which is a Python package for running TensorFlow on CPUs. This package is commonly used when GPU acceleration is not available or not required.
Vulnerability Detail
The vulnerability (CVE-2021-37679) arises from the nesting of a tf.map_fn within another tf.map_fn call when the input tensor is a RaggedTensor and no function signature is provided. In this scenario, the code assumes that the output is a fully specified tensor and fills the output buffer with uninitialized contents from the heap. As a result, the t and z outputs, which should be identical, differ. The last row of t contains data from the heap, which can be exploited to leak other memory information. The bug lies in the conversion from a Variant tensor to a RaggedTensor, where the implementation fails to check that all inner shapes match, leading to additional dimensions. This vulnerability can also result in data loss if the input tensor is manipulated. The issue has been patched in the GitHub commit 4e2565483d0ffcadc719bd44893fb7f609bb5f12 and will be included in TensorFlow 2.6.0. The fix will also be backported to TensorFlow 2.5.1, TensorFlow 2.4.3, and TensorFlow 2.3.4, as these versions are also affected and within the supported range.
How To Fix
Latest published version: 2.21.0 as of 22 Oct 2026
pip install --upgrade "tensorflow-cpu==2.21.0"Upgrading to it may include breaking changes; the versions below are the minimum fixed release for this vulnerability.
If you can't move to the latest
The minimum fixed version depends on the release line you run.
-
>= 2.3.0, < 2.3.42.3.4pip install --upgrade "tensorflow-cpu==2.3.4" -
>= 2.4.0, < 2.4.32.4.3pip install --upgrade "tensorflow-cpu==2.4.3"
Tip — Not sure which release line you're on? Vulert can find the fix for the version you are using. Scan free →
Workaround
As a workaround, you can avoid nesting tf.map_fn calls with RaggedTensor inputs. Instead, consider using alternative approaches or restructuring your code to achieve the desired functionality without relying on this specific nesting pattern. Here's an example of how you can modify your code to avoid the vulnerable pattern:
# Original code
output = tf.map_fn(lambda x: tf.map_fn(lambda y: y + 1, x), input_tensor)
# Modified code
output = tf.map_fn(lambda x: x + 1, input_tensor)
Conclusion
The vulnerability (CVE-2021-37679) in TensorFlow CPU's tf.map_fn nesting with RaggedTensor inputs has been identified and addressed. It is crucial to update to the patched versions to mitigate the risk of uninitialized memory leaks and potential data loss. Follow the recommended steps provided in the 'how_to_fix' section to rectify the vulnerability. For further support on vulnerability remediation, contact DevNack.com.
See every vulnerability recorded for tensorflow-cpu →
FAQ
TensorFlow is an open source platform for machine learning that provides a comprehensive ecosystem of tools and resources for developing and deploying machine learning models.
The TensorFlow CPU package is a Python package that allows running TensorFlow on CPUs. It is commonly used when GPU acceleration is not available or not required.
The vulnerability occurs when a tf.map_fn is nested within another tf.map_fn call, and the input tensor is a RaggedTensor without a provided function signature. This leads to uninitialized memory being used to fill the output buffer, resulting in a memory leak and potential data loss.
The vulnerability can lead to uninitialized memory leaks and potential data loss. Additionally, an attacker could exploit the leaked memory information to gather sensitive data from the system.
To fix the vulnerability, update the TensorFlow CPU package to the patched versions. Follow the official TensorFlow documentation or the package manager's instructions for updating the package. Additionally, ensure that your code does not nest tf.map_fn calls with RaggedTensor inputs without providing a function signature.
References
Related vulnerabilities
Recent vulnerabilities recorded for tensorflow-cpu.
Habilume
Habit tracker with a companion that grows with you
Complete habits, earn Lume Points, and watch your Lume grow as you stay consistent.
