2008-02-03 19:01:51

by cecco

[permalink] [raw]
Subject: hpet on nforce4/MCP51 in asus a6t

Hello,
I have a asus a6t with nforce4/MCP51 chipset. I pass to kernel 2.6.24
32 bit the options acpi_use_timer_override and hpet=force, in this way
and the timer IRQ ends up routed as IO-APIC-edge instead of XT-PIC.
With acpi_use_timer_override I just get:

TIMER: vector=0x30 apic1=0 pin1=2 apic2=-1 pin2=-1

and it works through the IO-APIC.

Instead the hpet is disabled and I just get :

Time: acpi_pm clocksource has been installed.
Clocksource tsc unstable (delta = -85010975 ns)

Is possible to have hpet working also with if the BIOS doesn't support
it ?

I wish to be personally CC'ed the answers/comments posted to the list
in response to my posting.

Cecco






____________________________________________________
Tiscali.Fax: ricevi gratis sulla tua email e invii
a 12 cent per pagina senza scatto alla risposta
http://vas.tiscali.it/fax//


2008-02-04 12:30:41

by Carlos Corbacho

[permalink] [raw]
Subject: Re: hpet on nforce4/MCP51 in asus a6t

[Adding tglx to CC as he knows more about HPET's than I]

On Sunday 03 February 2008 18:33:57 [email protected] wrote:
> Hello,
> I have a asus a6t with nforce4/MCP51 chipset. I pass to kernel 2.6.24
> 32 bit the options acpi_use_timer_override and hpet=force, in this way
> and the timer IRQ ends up routed as IO-APIC-edge instead of XT-PIC.
> With acpi_use_timer_override I just get:
>
> TIMER: vector=0x30 apic1=0 pin1=2 apic2=-1 pin2=-1
>
> and it works through the IO-APIC.
>
> Instead the hpet is disabled and I just get :
>
> Time: acpi_pm clocksource has been installed.
> Clocksource tsc unstable (delta = -85010975 ns)
>
> Is possible to have hpet working also with if the BIOS doesn't support
> it ?

One point before we start - MCP51 is nForce 5, not nForce 4 (which is CK804).

If my understanding is correct, acpi_use_timer_override is only a workaround
required for those nForce 5 boards that don't enable the HPET for you.

So IIUC, you should be able to use either 'hpet=force'
or 'acpi_use_timer_override', either of which should solve any timing issues
you see, but you shouldn't need both (which should solve your problem of
acpi_use_timer_override ignoring the HPET - it shouldn't be needed
if 'hpet=force' enables the HPET).

-Carlos
--
E-Mail: [email protected]
Web: strangeworlds.co.uk
GPG Key ID: 0x23EE722D

2008-02-04 18:10:59

by cecco

[permalink] [raw]
Subject: Re: hpet on nforce4/MCP51 in asus a6t

Carlos Corbacho <carlos <at> strangeworlds.co.uk> writes:

>
> [Adding tglx to CC as he knows more about HPET's than I]
>
> On Sunday 03 February 2008 18:33:57 f.cecco77 <at> tiscali.it wrote:
> > Hello,
> > I have a asus a6t with nforce4/MCP51 chipset. I pass to kernel 2.6.24
> > 32 bit the options acpi_use_timer_override and hpet=force, in this way
> > and the timer IRQ ends up routed as IO-APIC-edge instead of XT-PIC.
> > With acpi_use_timer_override I just get:
> >
> > TIMER: vector=0x30 apic1=0 pin1=2 apic2=-1 pin2=-1
> >
> > and it works through the IO-APIC.
> >
> > Instead the hpet is disabled and I just get :
> >
> > Time: acpi_pm clocksource has been installed.
> > Clocksource tsc unstable (delta = -85010975 ns)
> >
> > Is possible to have hpet working also with if the BIOS doesn't support
> > it ?
>
> One point before we start - MCP51 is nForce 5, not nForce 4 (which is CK804).
>
> If my understanding is correct, acpi_use_timer_override is only a workaround
> required for those nForce 5 boards that don't enable the HPET for you.
>
> So IIUC, you should be able to use either 'hpet=force'
> or 'acpi_use_timer_override', either of which should solve any timing issues
> you see, but you shouldn't need both (which should solve your problem of
> acpi_use_timer_override ignoring the HPET - it shouldn't be needed
> if 'hpet=force' enables the HPET).
>
> -Carlos

Thanks for your answer. In my case only acpi_use_timer_override helps with
timing issues, enabling apic. Unfortunately, either the options hpet=force or
acpi_use_timer_override don't enable HPET on asus a6t. The situation is always
the same:

cecco:/home/cecco# grep Clock /proc/timer_list
Clock Event Device: pit
Clock Event Device: lapic
Clock Event Device: lapic

I don't understand because with Turion X2 or amd64 is so difficult to have hpet
and dynticks working. Is important to have a cool notebook, but Amd doesn't seem
to understand it. Amd has introduced C1e state for a good powersaving, saying
it is equivalent to C3 state but I haven't any prove about this.

Cecco