Subject: Re: acpi SCI interrupt override bug
To: Frank van der Linden <fvdl@netbsd.org>
From: Cherry G. Mathew <cherry.g.mathew@gmail.com>
List: port-i386
Date: 03/02/2007 10:28:05
On 3/2/07, Frank van der Linden <fvdl@netbsd.org> wrote:
> Cherry G. Mathew wrote:
> > Patch below is commit-worthy, IMHO. rtk0 works for me now.
> That does look better. However, I think we need to look at one more
> thing: if the override should basically be ignored, as this patch
> effectively does, then we might have an ACPI implementation where the
> override is bad, and should already be ignored (marked with a quirk)
> earlier. I know that other OSs do have suck quirks in their tables, and
> I think this might be the same scenario. Maybe you can verify that (I
> don't have much time right now).

I'm not sure I understand. What's happening here is that the intel
acpica code is (incorrectly ?) calling
acpi_md_OsInstallInterruptHandler() with the initially probed FADT
interrupt value. What I'm doing is to match up the old FADT value
which has been saved in mip->bus_pin (
mpacpi.c:mpacpi_nonpci_intr():254 ) and if they match-up, then patch
the requested Interrupt to the MADT global_int value ( which has
already been parsed and saved. ). Are you suggesting that the MADT
overrides themselves may be wrong ? If so how would the OS know and
what would be the canonical setting ?


-- 
~Cherry