Optional mmap(2) support for geom(4) providers

[Ivan Voras]

Maxim Sobolev wrote:

> I am not sure that I understand what do you mean. In the case of md(4) 
> mmap makes huge sense, since for preloaded images your disk in fact is 
> just range of kernel memory, so that doing mmap is trivial.

You're right. I was thinking of making mmap() of "raw" devices possible 
(like it is in Linux), but I see that it would be extremely nontrivial 
:) (block devices on Linux...)