Описание
Heap buffer overflow in Tensorflow
Impact
The SparseCountSparseOutput and RaggedCountSparseOutput implementations don't validate that the weights tensor has the same shape as the data. The check exists for DenseCountSparseOutput, where both tensors are fully specified:
https://github.com/tensorflow/tensorflow/blob/0e68f4d3295eb0281a517c3662f6698992b7b2cf/tensorflow/core/kernels/count_ops.cc#L110-L117
In the sparse and ragged count weights are still accessed in parallel with the data: https://github.com/tensorflow/tensorflow/blob/0e68f4d3295eb0281a517c3662f6698992b7b2cf/tensorflow/core/kernels/count_ops.cc#L199-L201
But, since there is no validation, a user passing fewer weights than the values for the tensors can generate a read from outside the bounds of the heap buffer allocated for the weights.
Patches
We have patched the issue in 3cbb917b4714766030b28eba9fb41bb97ce9ee02 and will release a patch release.
We recommend users to upgrade to TensorFlow 2.3.1.
For more information
Please consult our security guide for more information regarding the security model and how to contact us with issues and questions.
Attribution
This vulnerability is a variant of GHSA-p5f8-gfw5-33w4
Ссылки
- https://github.com/tensorflow/tensorflow/security/advisories/GHSA-pg59-2f92-5cph
- https://nvd.nist.gov/vuln/detail/CVE-2020-15196
- https://github.com/tensorflow/tensorflow/commit/3cbb917b4714766030b28eba9fb41bb97ce9ee02
- https://github.com/pypa/advisory-database/tree/main/vulns/tensorflow-cpu/PYSEC-2020-276.yaml
- https://github.com/pypa/advisory-database/tree/main/vulns/tensorflow-gpu/PYSEC-2020-311.yaml
- https://github.com/pypa/advisory-database/tree/main/vulns/tensorflow/PYSEC-2020-119.yaml
- https://github.com/tensorflow/tensorflow/releases/tag/v2.3.1
Пакеты
tensorflow
= 2.3.0
2.3.1
tensorflow-cpu
= 2.3.0
2.3.1
tensorflow-gpu
= 2.3.0
2.3.1
EPSS
5.8 Medium
CVSS4
8.5 High
CVSS3
CVE ID
Дефекты
Связанные уязвимости
In Tensorflow version 2.3.0, the `SparseCountSparseOutput` and `RaggedCountSparseOutput` implementations don't validate that the `weights` tensor has the same shape as the data. The check exists for `DenseCountSparseOutput`, where both tensors are fully specified. In the sparse and ragged count weights are still accessed in parallel with the data. But, since there is no validation, a user passing fewer weights than the values for the tensors can generate a read from outside the bounds of the heap buffer allocated for the weights. The issue is patched in commit 3cbb917b4714766030b28eba9fb41bb97ce9ee02 and is released in TensorFlow version 2.3.1.
In Tensorflow version 2.3.0, the `SparseCountSparseOutput` and `Ragged ...
EPSS
5.8 Medium
CVSS4
8.5 High
CVSS3