Re: [Xen-devel] Freeze with 2.6.32.19 and xen-4.0.1rc5, acpi problems

[Pasi Kärkkäinen <pasik@xxxxxx>]

On Wed, Aug 25, 2010 at 10:02:59AM +0200, Claus Rosenberger wrote:
> Am 25.08.2010 01:22, schrieb Claus Rosenberger:
> >  I checked out these options, following result
> >>>> - acpi=ht noapic     => worked
> >>>> - noapic             => didn't work, disk errors while booting, no 
> >>>> functional system
> >>>> - acpi=ht apic=debug => worked
> >>>>
> >>>> So it seems i definitly need the option acpi=ht or acpi=off or something 
> >>>> like that.
> >>>>
> >>> Keir, Jeremy: Any idea why acpi=ht is needed to make Intel DQ45CB (Q45 
> >>> chipset) work?
> >>>
> >>> Without "acpi=ht" the system freezes randomly without any 
> >>> log/debug/serialconsole output..
> >> At a random guess, there's something strange about the SCI interrupt. 
> >> When booting native, does "acpi" get irq 9 or something else?  Is there
> >> a BIOS setting relating to SCI or ACPI interrupts?
> >>
> > 
> > The kernel was running native without that problems, i attached dmesg. I
> > didn't find any BIOS options like that.
> > 
> 
> 
> a new test was done loading libata with option noacpi=1
> 
> - acpi=ht noapic     => worked
> - noapic             => didn't work, disk errors while booting, no functional 
> system
> - acpi=ht apic=debug => worked
> - xen without options but loading libata with option noacpi=1    => worked
> 

So only setting libata.noacpi=1 fixes (or workarounds) the problem.. hmm.
Or setting "acpi=ht" fixes it aswell.

Hopefully someone has ideas about the cause..

-- Pasi


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel