Re: Changing ibm4xx device tree.

[Shigeyuki Fukushima <shige%netbsd.org@localhost>]

Toru Nishimura wrote:
>>   Is it practical that there are many #include directives
>>   for all supported cpu in plb.c, opb.c, and so on.
>>   Such as:
> That's a hard fact of life.  IBM/AMCC had reasons to make variants
> mostly same but different in many little aspects.  Having different
> files which include others to fulfill commons and differences would
> not help.

First, I'm sorry that my plan wasn't clear and essential and
has no persuation.

Perhaps you might thought that shige told something stupid.
Surely I was confused myself. ;-(


I have collected my idea...

What I achieve the following things at codes of powerpc/ibm4xx:

1. we, all developers of evbppc ports depend on powerpc/ibm4xx codes
   implement their own port (board or machine) without making some
   impact on others ports as much as we can.
   Therefore, almost developers don't have others eval-board
   or machine, and can't verify their codes on others board.
   This occurs when pulling up the releng-codes.

   For example, when port A(cpu x) depends on opb.c and
   opb was updated to support new cpu y for port B
   in the -current code,
   developer of port B must obtain a port A developer's consent
   to pull up opb.c.

2. However it may be inconsistent with 1., we, all developers
   share the same codes as much as we can.
   we can't share board-dependent codes, but we can share
   cpu-dependent codes such as plb and opb drivers and
   cpu devices declaration code.


What we should do? There is no way to solve these?
Please give me your ideas or hints.

-- 
Kind Regards,
--- shige
Shigeyuki Fukushima <shige@{FreeBSD,jp.FreeBSD,NetBSD}.org>