Masao Uebayashi said;

> I think having a common name helps read code.

I vote against the common name.  The reason is such the
"on-chip bus" would need arbitral amount of locators in arbitral
ways.  The demand will wildly vary simply because "it's the SoC
whatever HW designers can do."

Your proposal sounds me that you will define a hypothetical Super
SoC will cover all the need human civilization will want.  I'm reluctant
to introduce such the software-driven illusion out of kernel
programmers' head.

At the same time, I deeply express skeptical to PPC405 NetBSD
implementation of bus construct, which apparently intends to mimic/expose
HW defined internal.  On-chip interconnect topology is quite likely
changed in the _next_ product release.  Then adapting old code to new
SoC would get bitter and bitter.  I would even stand sliently with a placard
saying, "SoC is incompatible in any ways, can not be ruled in a single
discipline."

Toru Nishimura / ALKYL Technology