An attacker can pass an invalid axis
value to tf.quantization.quantize_and_dequantize
:
tf.quantization.quantize_and_dequantize(
input=[2.5, 2.5], input_min=[0,0], input_max=[1,1], axis=10)
This results in accessing a dimension outside the rank of the input tensor in the C++ kernel implementation:
const int depth = (axis_ == -1) ? 1 : input.dim_size(axis_);
However, dim_size
only does a DCHECK
to validate the argument and then uses it to access the corresponding element of an array:
int64 TensorShapeBase<Shape>::dim_size(int d) const {
DCHECK_GE(d, 0);
DCHECK_LT(d, dims());
DoStuffWith(dims_[d]);
}
Since in normal builds, DCHECK
-like macros are no-ops, this results in segfault and access out of bounds of the array.
We have patched the issue in eccb7ec454e6617738554a255d77f08e60ee0808 and will release TensorFlow 2.4.0 containing the patch. TensorFlow nightly packages after this commit will also have the issue resolved.
Please consult our security guide for more information regarding the security model and how to contact us with issues and questions.
This vulnerability has been reported in #42105
{ "nvd_published_at": "2020-10-21T21:15:00Z", "cwe_ids": [ "CWE-125" ], "severity": "HIGH", "github_reviewed": true, "github_reviewed_at": "2020-11-13T17:11:27Z" }