Write to immutable memory region 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 the tf.raw_ops.ImmutableConst operation returns a constant tensor created from a memory mapped file which is assumed immutable. However, if the type of the tensor is not an integral type, the operation crashes the Python interpreter as it tries to write to the memory area. If the file is too small, TensorFlow properly returns an error as the memory area has fewer bytes than what is needed for the tensor it creates. However, as soon as there are enough bytes, the above snippet causes a segmentation fault. This is because the allocator used to return the buffer data is not marked as returning an opaque handle since the needed virtual method is not overridden. 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
TensorFlow’s tf.raw_ops.ImmutableConst can crash the Python interpreter when it creates a non-integral tensor from a memory-mapped file containing enough bytes for that tensor. An attacker who can trigger the operation under those conditions can cause a segmentation fault; the input does not specify network access or authentication requirements.
Engineers using the PyPI tensorflow, tensorflow-cpu, or tensorflow-gpu packages in affected releases before 1.15.5, or in the 2.0.x through 2.3.x ranges before their respective fixes, should check their installations.