Advisory
GHSA-xm2v-8rrw-w9pm
Also known as: BIT-tensorflow-2021-29525, CVE-2021-29525, PYSEC-2021-162, PYSEC-2021-453, PYSEC-2021-651
Summary
Division by 0 in Conv2DBackpropInput
Details
Impact
An attacker can trigger a division by 0 in tf.raw_ops.Conv2DBackpropInput:
import tensorflow as tf
input_tensor = tf.constant([52, 1, 1, 5], shape=[4], dtype=tf.int32)
filter_tensor = tf.constant([], shape=[0, 1, 5, 0], dtype=tf.float32)
out_backprop = tf.constant([], shape=[52, 1, 1, 0], dtype=tf.float32)
tf.raw_ops.Conv2DBackpropInput(input_sizes=input_tensor, filter=filter_tensor,
out_backprop=out_backprop, strides=[1, 1, 1, 1],
use_cudnn_on_gpu=True, padding='SAME',
explicit_paddings=[], data_format='NHWC',
dilations=[1, 1, 1, 1])
This is because the implementation does a division by a quantity that is controlled by the caller:
const size_t size_A = output_image_size * dims.out_depth;
const size_t size_B = filter_total_size * dims.out_depth;
const size_t size_C = output_image_size * filter_total_size;
const size_t work_unit_size = size_A + size_B + size_C;
...
const size_t shard_size =
use_parallel_contraction ? 1 :
(target_working_set_size + work_unit_size - 1) / work_unit_size;
Patches
We have patched the issue in GitHub commit 2be2cdf3a123e231b16f766aa0e27d56b4606535.
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.
Severity
Low Basis: The source's own band: LOW
Dates
- Published
- 2021-05-21 14:21 UTC
- Modified
- 2026-09-10 03:49 UTC
Affected packages and fixes
| Package | Affected | Fix |
|---|---|---|
| tensorflow | Earlier versions | pip install --upgrade tensorflow==2.1.4 |
| tensorflow | introduced 2.2.0, fixed 2.2.3 | pip install --upgrade tensorflow==2.2.3 |
| tensorflow | introduced 2.3.0, fixed 2.3.3 | pip install --upgrade tensorflow==2.3.3 |
| tensorflow | introduced 2.4.0, fixed 2.4.2 | pip install --upgrade tensorflow==2.4.2 |
| tensorflow-cpu | Earlier versions | pip install --upgrade tensorflow-cpu==2.1.4 |
| tensorflow-cpu | introduced 2.2.0, fixed 2.2.3 | pip install --upgrade tensorflow-cpu==2.2.3 |
| tensorflow-cpu | introduced 2.3.0, fixed 2.3.3 | pip install --upgrade tensorflow-cpu==2.3.3 |
| tensorflow-cpu | introduced 2.4.0, fixed 2.4.2 | pip install --upgrade tensorflow-cpu==2.4.2 |
| tensorflow-gpu | Earlier versions | pip install --upgrade tensorflow-gpu==2.1.4 |
| tensorflow-gpu | introduced 2.2.0, fixed 2.2.3 | pip install --upgrade tensorflow-gpu==2.2.3 |
| tensorflow-gpu | introduced 2.3.0, fixed 2.3.3 | pip install --upgrade tensorflow-gpu==2.3.3 |
| tensorflow-gpu | introduced 2.4.0, fixed 2.4.2 | pip install --upgrade tensorflow-gpu==2.4.2 |
Fix commands are the export's own generic commands. Ojuju never runs them. Check them against your setup first.
Check your own computer: guard packages check
Links
- https://github.com/tensorflow/tensorflow/security/advisories/GHSA-xm2v-8rrw-w9pm
- https://nvd.nist.gov/vuln/detail/CVE-2021-29525
- https://github.com/tensorflow/tensorflow/commit/2be2cdf3a123e231b16f766aa0e27d56b4606535
- https://github.com/pypa/advisory-database/tree/main/vulns/tensorflow-cpu/PYSEC-2021-453.yaml
- https://github.com/pypa/advisory-database/tree/main/vulns/tensorflow-gpu/PYSEC-2021-651.yaml
- https://github.com/pypa/advisory-database/tree/main/vulns/tensorflow/PYSEC-2021-162.yaml
- https://github.com/tensorflow/tensorflow
Source
Source: GitHub Advisory Database (reviewed advisories) · CC BY 4.0