Vulnerability · Published Nov 5, 2021 · Assigned by GitHub_M
CVE-2021-41221Access to invalid memory during shape inference in `Cudnn*` ops
High 7.8
In plain English
A high-severity buffer overflow in tensorflow from tensorflow. There are no confirmed reports of exploitation so far. The chance it is exploited in the next 30 days is estimated at 0.2% (higher than 13% of all known vulnerabilities). Fixed versions are available - upgrade the affected package.
Official description
TensorFlow is an open source platform for machine learning. In affected versions the shape inference code for the `Cudnn*` operations in TensorFlow can be tricked into accessing invalid memory, via a heap buffer overflow. This occurs because the ranks of the `input`, `input_h` and `input_c` parameters are not validated, but code assumes they have certain values. The fix will be included in TensorFlow 2.7.0. We will also cherrypick this commit on TensorFlow 2.6.1, TensorFlow 2.5.2, and TensorFlow 2.4.4, as these are also affected and still in supported range.
Affected packages & fixed versions
| Ecosystem | Package | Fixed in | Advisory |
|---|---|---|---|
| PyPI | tensorflow | 2.4.4, 2.5.2, 2.6.1, 2.7.0 | PYSEC-2021-413 |
| PyPI | tensorflow | 2.6.1 | GHSA-cqv6-3phm-hcwx |
| PyPI | tensorflow | 2.5.2 | GHSA-cqv6-3phm-hcwx |
| PyPI | tensorflow | 2.4.4 | GHSA-cqv6-3phm-hcwx |
| PyPI | tensorflow-cpu | 2.5.2 | GHSA-cqv6-3phm-hcwx |
| PyPI | tensorflow-cpu | 2.4.4, 2.5.2, 2.6.1, 2.7.0 | PYSEC-2021-630 |
| PyPI | tensorflow-cpu | 2.6.1 | GHSA-cqv6-3phm-hcwx |
| PyPI | tensorflow-cpu | 2.4.4 | GHSA-cqv6-3phm-hcwx |
| PyPI | tensorflow-gpu | 2.4.4, 2.5.2, 2.6.1, 2.7.0 | PYSEC-2021-828 |
| PyPI | tensorflow-gpu | 2.6.1 | GHSA-cqv6-3phm-hcwx |
| PyPI | tensorflow-gpu | 2.5.2 | GHSA-cqv6-3phm-hcwx |
| PyPI | tensorflow-gpu | 2.4.4 | GHSA-cqv6-3phm-hcwx |
Affected products (as reported)
| Vendor | Product | Versions |
|---|---|---|
| tensorflow | tensorflow | >= 2.6.0, < 2.6.1>= 2.5.0, < 2.5.2< 2.4.4 |
Weakness: CWE-120 · Vector: CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H