Looks fine.
>
> Bug: https://bugs.openjdk.java.net/browse/JDK-8130737
> Webrev : http://cr.openjdk.java.net/~prr/8130737/
>
> If a child raster is created with a non-zero x-offset then subsequently
> in the code that validates the raster for imaging operations we may
> incorrectly calculate that the image exceeds the data buffer.
> This resolves that and a test has been provided that exercises
> each of the code paths with the buggy logic.
>
> This also reorders the arguments to a macro to use width and height
> consistently,
> and use signed values instead of unsigned in the macros.
>
> -phil.
