----- Original Message ----- 
From: <itojun@iijlab.net>
To: "Perry E. Metzger" <perry@piermont.com>
Cc: "TAKEMURA Shin" <takemura@netbsd.org>; <tech-kern@netbsd.org>
Sent: Monday, June 12, 2000 9:04 AM
Subject: Re: ACPI


> >> Is anyone working for ACPI support ?
> >> Hpcmips needs power management mechanism. And I found
> >> that ACPI is informative and interesting to me. Legacy APM spec
> >> and NetBSD's APM driver depend on i386 CPU too much and,
> >> therefore, other platforms including Hpcmips can't use them.
> >> However, ACPI might work on Hpcmips as well as 386.
> 
> I'm not sure why ACPI works on hpcmips while APM does not.
> Does WinCE use ACPI and its pseudo instructions?

No. Hpcmis hardwares don't have ACPI BIOS (for now).

ACPI specifies it's BIOS features and interfaces between BIOS and
OS. ACPI requires OS to support P-code method execution. If you
add some flexibility to OS to support native method execution 
as well as P-code execution, the OS can work very well on hpcmips.
This does not violate ACPI specification because ACPI have no
limitation about OS though ACPI have some requirements for OS.

Anyway, you should implement ACPI support like MI/MD. It will
be separated into 3 (or more) parts:

  1) ACPI and IA-PC(Intel Architecture PC) depend part
    (ex. system address map manipulation)
  2) ACPI depend but not IA-PC depend part
    (ex. virtual machine for p-code execution)
  3) ACPI power management abstruction part
    this will be a largest part in three

Because ACPI is very rich and flexible specification, 3) is work as
generic power management frame work. So hpcmips and other
port could use it.

Well, I feel that I can't explain what I think because my english
is poor. Please feel free to ask me any question about this.

> There are group of Japanese who are doing ACPI pseudo instruction
> parser for FreeBSD.
> http://home.jp.FreeBSD.org/mail-list/acpi-jp/
> Should be good as starting point.

I will check it. Thank you for your good information.

Takemura