Heap buffer overflow in `UnsortedSegmentSum` in TensorFlow
tensorflow
Affected versions
| Product | Affected | Fixed in |
|---|---|---|
| tensorflow PyPI |
|
|
| tensorflow-cpu PyPI |
|
|
tensorflow
| Product | Affected | Fixed in |
|---|---|---|
| tensorflow PyPI |
|
|
| tensorflow-cpu PyPI |
|
|
| tensorflow-gpu PyPI |
|
|
In TensorFlow before 1.15, a heap buffer overflow in UnsortedSegmentSum can be produced when the Index template argument is int32. In this case data_size and num_segments fields are truncated from int64 to int32 and can produce negative numbers, resulting in accessing out of bounds heap memory. This is unlikely to be exploitable and was detected and fixed internally in TensorFlow 1.15 and 2.0.
From the CVE record · GitHub_M
TensorFlow before 1.15 has a heap buffer overflow in UnsortedSegmentSum when the Index template argument is int32. Truncation of data_size and num_segments can produce negative values and out-of-bounds heap access. The input does not specify how an attacker would reach this code, whether authentication or user interaction is needed, or a practical exploit.
Engineers running PyPI tensorflow, tensorflow-cpu, or tensorflow-gpu versions 1.0.0 through 1.14.x should check their installations.