This sounds like a good idea: put support into the kernel where it is faster and has access to more of the insides of the kernel (rather than going through interfaces). It is like programming on the bare metal instead of using the standard toolbox calls (or API).
However, the downside is not often reported: this will make the kernel even larger and opens up security risks as well. I tend more to believe in the small kernel more than anything, and Solaris already appears to be one of the largest kernels in the open source arena.
Time will tell whether this was a good idea or not.