[Freesurfer] smoothing kernel inconsistency

Dear Freesurfer experts,

I am trying to do the group analysis, before this I tried to smooth the data with the different smoothing kernel. However, when I checked the output dir, it turned out that data were smoothed with a larger smoothing kernel than I expected. For example, images intended to be smoothed with 5mm were actually smoothed with 8mm. So does anyone know why?

Another thing is about the surface area in the analysis. I am a little confused which surface does the default cmd or the area in qdec indicate? Does it mean the area I got is white surface? If so, can I also get a pial surface? Technically, is the middle between the white surface and the pial surface the more reasonable one? Thank you in advance.

Best regards, Ting