Access to invalid memory during shape inference in `Cudnn*` ops
tensorflow
Affected versions
| Product | Affected | Fixed in |
|---|---|---|
| tensorflow PyPI |
|
|
tensorflow
| Product | Affected | Fixed in |
|---|---|---|
| tensorflow PyPI |
|
|
| tensorflow-cpu PyPI |
|
|
| tensorflow-gpu PyPI |
|
|
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.
From the CVE record · GitHub_M
Shape inference for TensorFlow’s `Cudnn*` operations does not validate the ranks of `input`, `input_h`, and `input_c`. An attacker who can supply inputs with unexpected ranks can trigger a heap buffer overflow and invalid memory access. The input does not specify network access, authentication, or user-interaction requirements.
Anyone running affected versions of the PyPI packages `tensorflow`, `tensorflow-cpu`, or `tensorflow-gpu`: versions below 2.4.4, 2.5.0–2.5.1, 2.6.0, or 2.7.0 release candidates.