Re: 2.6.25 pci=noacpi

[Thomas Renninger <trenn@suse.de>]

On Fri, 2008-05-02 at 14:11 +0100, Richard wrote:
> Andrew Morton wrote:
> > On Fri, 02 May 2008 12:23:01 +0100 Richard <tuxbox.guru@gmail.com> wrote:
> >
> >   
> >> Len Brown wrote:
> >>     
> >>> On Thursday 01 May 2008, Richard wrote:
> >>>   
> >>>       
> >>>> Len Brown wrote:
> >>>>     
> >>>>         
> >>>>> On Tuesday 29 April 2008, Richard wrote:
> >>>>>   
> >>>>>       
> >>>>>           
> >>>>>> Hi there,
> >>>>>>
> >>>>>> Now I am at a loss.. I cant get any logs from netconsole. With ACPI
> >>>>>> enabled it drops dead just after the kernel hands control to Init. When
> >>>>>> I reboot with acpi disabled there is no trace of the previous boot in
> >>>>>> /var/log/message or kernel logs.
> >>>>>>
> >>>>>> The network card I have is a Tigon3 and its not initialised before the
> >>>>>> system dies... or so it looks (altho I cant seem to get any netconsole
> >>>>>> working - I suspect pebkac)
> >>>>>>
> >>>>>> I have tried to find an option where the kernel boot is logged
> >>>>>> explicitly to disk, but I cant seem to find the option. There is so much
> >>>>>> chaff I cant seem to get my brain to find the REAL options list thats up
> >>>>>> to date.
> >>>>>>
> >>>>>> Any pointers are appreciated
> >>>>>>     
> >>>>>>         
> >>>>>>             
> >>>>> go fishing with this bait:
> >>>>>
> >>>>> idle=poll
> >>>>> acpi=noirq
> >>>>>
> >>>>> -Len
> >>>>>
> >>>>>   
> >>>>>       
> >>>>>           
> >>>> Thanks :-) With help from Thomas (and his team) , noapictimer seems to
> >>>> have fixed the symptom.. now its to hunt for the cure :-)
> >>>>     
> >>>>         
> >>> if the problem is the apic timer stopping during C-states,
> >>> then you should find that idle=poll will also work around the problem.
> >>> If idle=poll does not work around it but noapictimer does, then
> >>> the problem is the apic timer when the system is non-idle.
> >>>
> >>> -Len
> >>>
> >>>   
> >>>       
> >> Hi Len,
> >>
> >> thanks - the idle=poll works to fix the problem.
> >>     
> >
> > Well, that's hardly a "fix".  The power consumption will be very high.
> >
> >   
> Hehehe, its a start. A booting kernel is definitely a good start (even 
> one what drinks battery). I have some avenues in the source to 
> investigate, but being a ACPI n00bie its trying to get my head around 
> the mass of code thats a daunting task.

Richard, could you try processor.max_cstate=2 (or even 1), pls (without
noapictimer).
Or you can double check whether you have cstates at all first(when
booting with noapictimer):
cat /proc/acpi/processor/*/power

Vendors often invalidate deeper sleep states with an invalid latency.
Maybe this has been forgotten here, Windows does not use deeper C-state
or it works there for some reason.
If this works for you, please send dmidecode output.
IMO we should then blacklist this machine in
drivers/acpi/processor_idle.c to not use the offending sleep state.
There already is a blacklist.

Thanks,

   Thomas