[2.6.17.8] noapic and /proc/acpi/event

[2.6.17.8] noapic and /proc/acpi/event

[Andreas Steinmetz]

I do have a problem with a new laptop (Acer Ferrari 4006):

It does suspend either to disk or to ram only when I do boot with
"noapic". So far, so good.

If, however, I do boot with "noapic" no events are delivered to
/proc/acpi/event so lid switch and power button can't be used to suspend
anymore.

The strange thing is, that at least in /proc/acpi/button/lid/LID/state I
can view the lid switch state.

Can anybody shed some light on this?
-- 
Andreas Steinmetz                       SPAMmers use robotrap@domdv.de

Re: [2.6.17.8] noapic and /proc/acpi/event

[Len Brown]

On Thursday 07 September 2006 14:52, Andreas Steinmetz wrote:
> I do have a problem with a new laptop (Acer Ferrari 4006):
> 
> It does suspend either to disk or to ram only when I do boot with
> "noapic". So far, so good.

Well no, that isn't so good either.  You shouldn't need "noapic"
for anything, either normal operation or suspend/resume.

Do ACPI events work properly w/o noapic if you don't suspend/resume?

You should be able to kill acpid, and cat /proc/acpi/event
and open/close your lid and watch events appear --
same for power button.
You should also be able to see the acpi line in /proc/interrupts
increment for each of these events.

> If, however, I do boot with "noapic" no events are delivered to
> /proc/acpi/event so lid switch and power button can't be used to suspend
> anymore.

Does noapic work properly before the suspend?
(test the same way as w/o noapic above)

> The strange thing is, that at least in /proc/acpi/button/lid/LID/state I
> can view the lid switch state.

The problem with your system is that it isn't getting ACPI interrupts.
The lid state in /proc is immune to that problem because when
you read that file Linux asks the hardware for its state on demand.

cheers,
-Len
 

Re: [2.6.17.8] noapic and /proc/acpi/event

[Andreas Steinmetz]

Len Brown wrote:
> On Thursday 07 September 2006 14:52, Andreas Steinmetz wrote:
> 
>>I do have a problem with a new laptop (Acer Ferrari 4006):
>>
>>It does suspend either to disk or to ram only when I do boot with
>>"noapic". So far, so good.
> 
> 
> Well no, that isn't so good either.  You shouldn't need "noapic"
> for anything, either normal operation or suspend/resume.
> 
> Do ACPI events work properly w/o noapic if you don't suspend/resume?
> 

Yes, they do, that's how I tested.

> You should be able to kill acpid, and cat /proc/acpi/event
> and open/close your lid and watch events appear --
> same for power button.
> You should also be able to see the acpi line in /proc/interrupts
> increment for each of these events.
> 
> 
>>If, however, I do boot with "noapic" no events are delivered to
>>/proc/acpi/event so lid switch and power button can't be used to suspend
>>anymore.
> 
> 
> Does noapic work properly before the suspend?
> (test the same way as w/o noapic above)
> 

No events, no ACPI interrupts.

> 
>>The strange thing is, that at least in /proc/acpi/button/lid/LID/state I
>>can view the lid switch state.
> 
> 
> The problem with your system is that it isn't getting ACPI interrupts.
> The lid state in /proc is immune to that problem because when
> you read that file Linux asks the hardware for its state on demand.
> 

I see.

> cheers,
> -Len
>  


-- 
Andreas Steinmetz                       SPAMmers use robotrap@domdv.de