Re: acpica5 issues

[YONETANI Tomokazu <qhwt+dragonfly-kernel@xxxxxxxxxx>]

On Wed, May 26, 2004 at 11:42:55PM -0400, Rahul Siddharthan wrote:
> YONETANI Tomokazu said on May 27, 2004 at 08:44:27:
> > On Tue, May 25, 2004 at 11:21:02PM -0400, Rahul Siddharthan wrote:
> > > If I use the module, it doesn't load with vanilla sources.  With your
> > > patch of May 22, the module loads, but the machine hangs (locks up)
> > > before it can finish booting.  First it hung at the dhclient stage,
> > > second time it hung during the fsck (necessitated by the first hang).
> > > This is with the madt line commented out.
> > 
> > I'm not sure if I can help you, but can you put your dmesg
> > from boot into singleuser mode with verbose flag somewhere
> > where we can download it?
> 
> I can't seem to get a dmesg with acpi: it hangs before writing any
> logs.  The dmesg from boot -v without acpi is at
> http://rsidd.online.fr/dmesg_noacpi.gz

As I thought I can't help you just by looking at the dmesg :(
Ok, there is a variable called debug.acpi.disabled (in non-patched
version it's debug.acpi.disable) to disable individual acpi subsystems.
When you have time, can you try to find out what subsystem triggers
the hang? The variable is a space-separated list and to be set at the
boot loader; for instance, if you want disable "cpu" and "lid",
type as follows from the boot loader:

OK set debug.acpi.disabled="cpu lid"

If you include "acpi", acpi itself is disabled. The valid subsystem
names are(you can find out by grep'ing acpi_disabled in /sys/dev/acpica5):

"acad" "acpi" "bus" "button" "children" "cmbat" "cpu" "ec"
"isa" "lid" "pci" "pci_link" "sysresource" "thermal" "timer"

Regards.