Alienware acpi problem

[John Baldwin]

On Tuesday 25 July 2006 16:15, Steve Kargl wrote:
> On Mon, Jul 24, 2006 at 06:23:24PM -0400, John Baldwin wrote:
> > On Monday 24 July 2006 17:50, Steve Kargl wrote:
> > > ACPI gurus,
> > > 
> > > My colleague has installed FreeBSD 6.1-table onto his
> > > Alienware MJ-12 laptop.  During the boot process, we
> > > see
> > > 
> > > acpi0: [MPSAFE]
> > >     ACPI-0438: *** Error: Looking up [\\_SB_.PCI0.PIB_.LNKH] in 
namespace, 
> > AE_NOT_FOUND
> > > SearchNode 0xffffff000091c240 StartNode 0xffffff000091c240 ReturnNode 0
> > >     ACPI-0438: *** Error: Looking up [\\_SB_.PCI0.LNKH] in namespace, 
> > AE_NOT_FOUND
> > > SearchNode 0xffffff00009084c0 StartNode 0xffffff00009084c0 ReturnNode 0
> > >     ACPI-0438: *** Error: Looking up [\\_SB_.PCI0.LNKH] in namespace, 
> > AE_NOT_FOUND
> > > SearchNode 0xffffff00008eb200 StartNode 0xffffff00008eb200 ReturnNode 0
> > >     ACPI-0438: *** Error: Looking up [\\_SB_.PCI0.LNKH] in namespace, 
> > AE_NOT_FOUND
> > > SearchNode 0xffffff0000909d40 StartNode 0xffffff0000909d40 ReturnNode 0
> > 
> > Well, that's probably why cardbus is broken.  Can you get the output of 
> > acpidump?  Specifically I'd be interested in the parts that mention LNKH 
(as 
> > well as where it actually lives in the device tree).
> > 
> 
> John, 
> 
> The pattern LNKH appears in multiple places throughout the asl file.
> Instead of sending seemly random chucks to the list, the entire asl
> can be found at (uncompressed and gzipped)
> 
> http://troutmask.apl.washington.edu/~kargl/alienware.asl
> http://troutmask.apl.washington.edu/~kargl/alienware.asl.gz

You have some of the most incompetent BIOS writers I've ever seen working on 
your machine I'm afraid.  First off, there is no LNKH device on your system.  
You do have various pci_link devices.  Four for APIC: ALKA, ALKB, ALKC, and 
ALKD, and four for non-APIC: LNKA, LNKB, LNKC, LNKD.  Not only that, but they 
all live in \_SB_.PCI0.PCIB.  Some places refer to LNKH via 
\_SB_.PCI0.PCIB.LNKH (as do all places for LNKA - LNKD it seems) but others 
refer to it as \_SB_.PCI0.LNKH.  Anyways, I think you might be fine if you 
always use 'device apic', as the PCI interrupt routing tables for the APIC 
case don't seem to reference LNKH.  I think it will only go down in flames 
this way if you disable APIC (via hint or not including 'device apic' in your 
kernel config).

-- 
John Baldwin