TensorFlow vulnerable to null dereference on MLIR on empty function attributes

Overview

Impact

When mlir::tfg::ConvertGenericFunctionToFunctionDef is given empty function attributes, it gives a null dereference.

// Import the function attributes with a `tf.` prefix to match the current
// infrastructure expectations.
for (const auto& namedAttr : func.attr()) {
  const std::string& name = "tf." + namedAttr.first;
  const AttrValue& tf_attr = namedAttr.second;
  TF_ASSIGN_OR_RETURN(Attribute attr,
                      ConvertAttributeValue(tf_attr, builder, tfgDialect));
  attrs.append(name, attr);
}

If namedAttr.first is empty, it will crash.

Patches

We have patched the issue in GitHub commit 1cf45b831eeb0cab8655c9c7c5d06ec6f45fc41b.

The fix will be included in TensorFlow 2.10.0. We will also cherrypick this commit on TensorFlow 2.9.1, TensorFlow 2.8.1, and TensorFlow 2.7.2, 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.

Affected packages

• tensorflow < 2.7.2
• tensorflow >= 2.8.0, < 2.8.1
• tensorflow >= 2.9.0, < 2.9.1
• tensorflow-cpu < 2.7.2
• tensorflow-cpu >= 2.8.0, < 2.8.1
• tensorflow-cpu >= 2.9.0, < 2.9.1
• tensorflow-gpu < 2.7.2
• tensorflow-gpu >= 2.8.0, < 2.8.1
• tensorflow-gpu >= 2.9.0, < 2.9.1

Remediation

Upgrade to a patched release:

• tensorflow 2.7.2
• tensorflow 2.8.1
• tensorflow 2.9.1
• tensorflow-cpu 2.7.2
• tensorflow-cpu 2.8.1
• tensorflow-cpu 2.9.1
• tensorflow-gpu 2.7.2
• tensorflow-gpu 2.8.1
• tensorflow-gpu 2.9.1