Heap OOB in shape inference for `QuantizeV2`

Impact

The shape inference code for QuantizeV2 can trigger a read outside of bounds of heap allocated array:

import tensorflow as tf

@tf.function
def test():
  data=tf.raw_ops.QuantizeV2(
    input=[1.0,1.0],
    min_range=[1.0,10.0],
    max_range=[1.0,10.0],
    T=tf.qint32,
    mode='MIN_COMBINED',
    round_mode='HALF_TO_EVEN',
    narrow_range=False,
    axis=-100,
    ensure_minimum_range=10)
  return data

test()

This occurs whenever axis is a negative value less than -1. In this case, we are accessing data before the start of a heap buffer:

int axis = -1;
Status s = c->GetAttr("axis", &axis);
if (!s.ok() && s.code() != error::NOT_FOUND) {
  return s;
}   
... 
if (axis != -1) {
  ...
  TF_RETURN_IF_ERROR(
      c->Merge(c->Dim(minmax, 0), c->Dim(input, axis), &depth));
}

The code allows axis to be an optional argument (s would contain an error::NOT_FOUND error code). Otherwise, it assumes that axis is a valid index into the dimensions of the input tensor. If axis is less than -1 then this results in a heap OOB read.

Patches

We have patched the issue in GitHub commit a0d64445116c43cf46a5666bd4eee28e7a82f244.

The fix will be included in TensorFlow 2.7.0. We will also cherrypick this commit on TensorFlow 2.6.1, as this version is the only one that is also affected.

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 members of the Aivul Team from Qihoo 360.