> 	Well, I just vote for kernfs extension or adding  a dedicated fs that
> 	will map devices  detected in  the  kernel boot stage into  hierarchy
> 	stored somewhere in the filesystem.  Consider we extend the kernfs:
> 

It's been done before, and in fact, it ultimately is the wrong idea because:

a) At varying points it is ambiguous what the actual bus interconnects are.

b) It cannot handle architectures where there may be either no root (or
a desire to have multiple 'root' nodes). It also cannot handle multipath
architectures where you may have separate bus interconnects to the same
end node.

c) It is a lot of effort to implement and no clear preferred access mechanism
or policy for leaf or bus nodes. Like, what does it buy you?