Advisoriesby IbisHeap buffer overflow in `FractionalAvgPoolGrad`
Impact
The implementation for tf.raw_ops.FractionalAvgPoolGrad can be tricked into accessing data outside of bounds of heap allocated buffers:
```python
import tensorflow as tf
tf.raw_ops.FractionalAvgPoolGrad(
orig_input_tensor_shape=[0,1,2,3],
out_backprop = np.array([[[[541],[541]],[[541],[541]]]]),
row_pooling_sequence=[0, 0, 0, 0, 0],
col_pooling_sequence=[-2, 0, 0, 2, 0],
overlapping=True)
`
The [implementation](https://github.com/tensorflow/tensorflow/blob/f24faa153ad31a4b51578f8181d3aaab77a1ddeb/tensorflow/core/kernels/fractional_avg_pool_op.cc#L205) does not validate that the input tensor is non-empty. Thus, code constructs an empty EigenDoubleMatrixMap and then accesses this buffer with indices that are outside of the empty area.
Patches
We have patched the issue in GitHub commit [0f931751fb20f565c4e94aa6df58d54a003cdb30](https://github.com/tensorflow/tensorflow/commit/0f931751fb20f565c4e94aa6df58d54a003cdb30).
The fix will be included in TensorFlow 2.6.0. We will also cherrypick this commit on TensorFlow 2.5.1, TensorFlow 2.4.3, and TensorFlow 2.3.4, as these are also affected and still in supported range.
For more information
Please consult [our security guide](https://github.com/tensorflow/tensorflow/blob/master/SECURITY.md) for more information regarding the security model and how to contact us with issues and questions.
###...
Affected versions
| Package | Affected | Fixed in |
|---|---|---|
| tensorflow PyPI | < 2.3.4 | 2.3.4 |
| >= 2.4.0, < 2.4.3 | 2.4.3 | |
| >= 2.5.0, < 2.5.1 | 2.5.1 |
Details and references
### Impact The implementation for `tf.raw_ops.FractionalAvgPoolGrad` can be tricked into accessing data outside of bounds of heap allocated buffers: ```python import tensorflow as tf tf.raw_ops.FractionalAvgPoolGrad( orig_input_tensor_shape=[0,1,2,3], out_backprop = np.array([[[[541],[541]],[[541],[541]]]]), row_pooling_sequence=[0, 0, 0, 0, 0], col_pooling_sequence=[-2, 0, 0, 2, 0], overlapping=True) ``` The [implementation](https://github.com/tensorflow/tensorflow/blob/f24faa153ad31a4b51578f8181d3aaab77a1ddeb/tensorflow/core/kernels/fractional_avg_pool_op.cc#L205) does not validate that the input tensor is non-empty. Thus, code constructs an empty `EigenDoubleMatrixMap` and then accesses this buffer with indices that are outside of the empty area. ### Patches We have patched the issue in GitHub commit [0f931751fb20f565c4e94aa6df58d54a003cdb30](https://github.com/tensorflow/tensorflow/commit/0f931751fb20f565c4e94aa6df58d54a003cdb30). The fix will be included in TensorFlow 2.6.0. We will also cherrypick this commit on TensorFlow 2.5.1, TensorFlow 2.4.3, and TensorFlow 2.3.4, as these are also affected and still in supported range. ### For more information Please consult [our security guide](https://github.com/tensorflow/tensorflow/blob/master/SECURITY.md) for more information regarding the security model and how to contact us with issues and questions. ### Attribution This vulnerability has been reported by members of the Aivul Team from Qihoo 360.
- CVSS 3.1
- CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:N
- Severity from
- GitHub (reviewed advisory)
- Weakness
- CWE-125, CWE-787
- Also known as
- BIT-tensorflow-2021-37651, CVE-2021-37651, PYSEC-2021-273, PYSEC-2021-564, PYSEC-2021-762
- github.com/tensorflow/tensorflow/security/advisories/GHSA-hpv4-7p9c-mvfr
- nvd.nist.gov/vuln/detail/CVE-2021-37651
- github.com/tensorflow/tensorflow/commit/0f931751fb20f565c4e94aa6df58d54a003cdb30
- github.com/pypa/advisory-database/tree/main/vulns/tensorflow-cpu/PYSEC-2021-564.yaml
- github.com/pypa/advisory-database/tree/main/vulns/tensorflow-gpu/PYSEC-2021-762.yaml
- github.com/pypa/advisory-database/tree/main/vulns/tensorflow/PYSEC-2021-273.yaml
- github.com/tensorflow/tensorflow
More TensorFlow advisories
All TensorFlow
Heap out of bounds access in sparse reduction operationsCVE-2021-37635fixed in 2.3.4 +2Floating point exception in `SparseDenseCwiseDiv`CVE-2021-37636fixed in 2.3.4 +2Null pointer dereference in `CompressElement`CVE-2021-37637fixed in 2.3.4 +2Null pointer dereference in `RaggedTensorToTensor`CVE-2021-37638fixed in 2.3.4 +2Null pointer dereference and heap OOB read in operations restoring tensorsCVE-2021-37639fixed in 2.3.4 +2Integer division by 0 in sparse reshapingCVE-2021-37640fixed in 2.3.4 +2