Uninitialized memory access in Eigen types in TensorFlow
tensorflow
Affected versions
| Product | Affected | Fixed in |
|---|---|---|
| tensorflow PyPI |
|
|
tensorflow
| Product | Affected | Fixed in |
|---|---|---|
| tensorflow PyPI |
|
|
2.3.2| tensorflow-cpu PyPI |
|
|
| tensorflow-gpu PyPI |
|
|
In affected versions of TensorFlow under certain cases a saved model can trigger use of uninitialized values during code execution. This is caused by having tensor buffers be filled with the default value of the type but forgetting to default initialize the quantized floating point types in Eigen. This is fixed in versions 1.15.5, 2.0.4, 2.1.3, 2.2.2, 2.3.2, and 2.4.0.
From the CVE record · GitHub_M
In affected TensorFlow versions, a saved model can trigger use of uninitialized values during code execution because Eigen does not default-initialize quantized floating-point types. The input does not specify how an attacker could provide the model or whether authentication or user interaction is required.
Teams using PyPI tensorflow, tensorflow-cpu, or tensorflow-gpu versions below 1.15.5, 2.0.0–2.0.3, 2.1.0–2.1.2, 2.2.0–2.2.1, or 2.3.0–2.3.1 should check their installations.