Bad characters in Asus A8N-VM CSM

[Adriaan de Groot]

On Thursday 22 December 2005 11:55, Adriaan de Groot wrote:
> I just picked up an Asus A8N-VM CSM motherboard. It's a nForce 430 + GForge
> 6150 integrated mini-ATX job, amd64, very 1337 and new. That means trouble
> :) I knew that when buying it, and know now that it really needs work to
> get it working at all.
>
> ACPI-0397: *** Error: NSSearchAndEnter: Bad character in ACPI Name:
> 43035350 ACPI-0381: *** Error: Looking up [0x43035350] (NON-ASCII)
> in namespace, AE_BAD_CHARACTER

[Following up on myself, for documentation purposes, and CCing -amd64 to warn 
off potential users there as well.]

Well, an instructive but unproductive evening yields the following:

1) Default BIOS (rev. 0403) and FBSD 5.1 amd64 boots, produces ACPI warning 
messages above and works normally (no NIC, no SATA tried).

2) Newest BIOS (rev. 0506) and FBSD 5.1 amd64 boots and then page faults in 
kernel mode in vm_pageq_enqueue right after GEOM adds ad0. ACPI warning is 
still there.

3) I didn't try 6-STABLE with the newest BIOS, since having busted ACPI there 
means you can't get _anything_ done.



From the booted 5.1 environment I got an ACPI dump and after hacking utmisc.c 
in contrib/dev/acpica to accept a broken character 0x03, I got a disassembly. 
The bad character is in 0x43035350, which from the looks of the IASL should 
have been PSSC (0x43535350) but somehow isn't. It's in one of the processor 
sections (the other has a Name(PSSC, 0x0A) ).

Elsewhere in the IASL there's some undefined symbols ending in _HFZF (sorry, I 
don't have the machine under scrutiny on or functional at this very moment), 
which I suppose is semi-normal, and then a totally weird-ass

	Return( While(Local1) {
		...
	})

This chokes the compiler, of course. I suspect that it should read

	Return(0x00)
	While (Local1) ...

or so, but this suggests yet more broken ACPI tables.

Over in Linux land, http://www.nvnews.net/vbulletin/showthread.php?t=60596 
shows that with some effort and the latest of the latest everything, it is 
possible to get a functioning machine. If I can create a barely legal ASL 
file that compiles, I'll have to try out what's described in "11.16.5.3 
Overriding the Default AML " of the handbook.

Until then (or someone more competent than myself deals with the issues), I'd 
suggest avoiding this board entirely.

[ade]