Advisory

PYSEC-2026-953

Also known as: BIT-tensorflow-2022-29196, CVE-2022-29196, GHSA-5v77-j66x-4c4g, PYSEC-2026-3110, PYSEC-2026-3278

Summary

Missing validation causes denial of service via Conv3DBackpropFilterV2

Details

Impact
The implementation of `tf.raw_ops.Conv3DBackpropFilterV2` does not fully validate the input arguments. This results in a CHECK-failure which can be used to trigger a denial of service attack:

import tensorflow as tf

tf.raw_ops.Conv3DBackpropFilterV2(
  input=tf.constant(.5053710941, shape=[2,2,2,2,1], dtype=tf.float16),
  filter_sizes=tf.constant(0, shape=[], dtype=tf.int32),
  out_backprop=tf.constant(.5053710941, shape=[2,2,2,2,1], dtype=tf.float16),
  strides=[1, 1, 1, 1, 1],
  padding="VALID",
  data_format="NDHWC",
  dilations=[1, 1, 1, 1, 1])

The code does not validate that the filter_sizes argument is a vector.

Patches
We have patched the issue in GitHub commit 174c5096f303d5be7ed2ca2662b08371bff4ab88.

The fix will be included in TensorFlow 2.9.0. We will also cherrypick this commit on TensorFlow 2.8.1, TensorFlow 2.7.2, and TensorFlow 2.6.4, 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 Neophytos Christou from Secure Systems Lab at Brown University.

Severity

Medium Basis: CVSS v3 score 5.5, vector CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H

Dates

Published
2026-07-06 08:03 UTC
Modified
2026-07-13 16:43 UTC

Affected packages and fixes

tensorflow-gpu

PyPI · All systems

Affected

  • introduced 0, fixed 2.6.4, introduced 2.7.0, fixed 2.7.2, introduced 2.8.0, fixed 2.8.1 (ecosystem)

Fixed versions

2.6.4, 2.7.2, 2.8.1

Fix

  • Fixed version 2.6.4
    pip install --upgrade tensorflow-gpu==2.6.4
  • Fixed version 2.7.2
    pip install --upgrade tensorflow-gpu==2.7.2
  • Fixed version 2.8.1
    pip install --upgrade tensorflow-gpu==2.8.1

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

View this record on OSV.dev

Source

Source: PyPA Advisory Database · CC BY 4.0

Sources & licences