Hello, just some wild guesses here: (don't have CS6 here) convert the mask to a layer. Apply AWA on a smart object containing the layer. Change layer content to the "fake mask layer", copy the mask.
Do I understand you correctly to say that you have an image as a Smart Object in a document, then you added a mask to the Smart Object layer in the main document.
These things are in two different places, and of course manipulations to one won't be coordinated in the other. As it turns out the Adaptive Wide Angle filter won't affect a layer mask even if no Smart Object is used.
It seems to me you'd either want to do your position-sensitive image manipulation before making the mask in the first place, or make a new mask after the AWA run.
When you have a SO and run a filter, that mask appears below the layer.
If you use that mask with black marks in it, using lens correction or AWA
does not apply to the mask simultaneously. I think #1 above may be the
way to go?