Re: ACPI-Problems with NetBSD/i386 and a Compaq DL360 G1

[Jukka Ruohonen <jruohonen%iki.fi@localhost>]

On Thu, Aug 26, 2010 at 11:59:33AM -0600, Michael L. Hitch wrote:
> done anything more with it.  I had thought about defining ACPI_MADT_DEBUG 
> to try to see what was happening, but never got any farther than thinking.
> 
>   I've got a DL360 G1 that I could test with, and I could also update one 
> of my boot disks with -current if the acpidump(8) output can get the MADT 
> information.

I already got the MADT from you in a private mail regarding an unrelated
issue. But I recommend to file a bug report nevertheless, so that this won't
get lost.

As Andrew hinted, the APIC table(s) sure look a little funny. The enabled
CPUs are, in this order:

                        APIC ID         ACPI CPU ID
                CPU     0               0
                CPU     6               6
                CPU     1               1
                CPU     7               7

Unfortunately there is no way to tell from the tables alone which is the
boot processor. So the code is based on an educated guess, or more bluntly,
luck.

> Has there been any further progress on this problem?  I had noticed it 
> quite some time ago, and have always been meaning to try to get more 
> information about the problem and get a PR submitted about it, but booting 
> without ACPI has been a sufficient enough work-around that I had't ever 

The current MADT/APIC/ACPI configuration soup is somewhat iffy with some
known issues in the configuration of both local APICs and I/O APICs. I have
done some work to rewrite the whole thing, but this is not ready yet.

- Jukka.