tfsa-2020-020.md

TFSA-2020-020: Heap buffer overflow in weighted sparse count ops

CVE Number

CVE-2020-15196

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:

    if (use_weights) {
      OP_REQUIRES(
          context, weights.shape() == data.shape(),
          errors::InvalidArgument(
              "Weights and data must have the same shape. Weight shape: ",
              weights.shape().DebugString(),
              "; data shape: ", data.shape().DebugString()));
    }

In the sparse and ragged count weights are still accessed in parallel with the data:

    for (int idx = 0; idx < num_values; ++idx) {
      int batch = is_1d ? 0 : indices_values(idx, 0);
      const auto& value = values_values(idx);
      per_batch_counts[batch][value] += weight_values(idx);
   }

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.

Vulnerable Versions

TensorFlow 2.3.0.

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 has been discovered through a variant analysis of a vulnerability reported by members of the Aivul Team from Qihoo 360.