Re: [2.6.6] Synaptics driver is 'jumpy'

[Dmitry Torokhov <dtor_core@ameritech.net>]

On Monday 17 May 2004 01:32 am, Jan De Luyck wrote:
> On Sunday 16 May 2004 21:29, Dmitry Torokhov wrote:
> > On Sunday 16 May 2004 02:06 pm, Jan De Luyck wrote:
> > > On Sunday 16 May 2004 19:18, Dmitry Torokhov wrote:
> > > > Hmm.. there was no changes to PS/2 processing between 2.6.5 and 2.6.6
> > > > except for some Logitech tweaking, but it should not affect Synaptics
> > > > handling in any way...
> > > >
> > > > Could you check if you still have DMA enabled on your disks, check your
> > > > time source (TSC, ACPI PM timer, etc) and probably boot with acpi off?
> > > >
> > > > Thank you.
> > >
> > > Dmitry,
> > >
> > > Booting with acpi=off fixes the problem, although I'm curious to what the
> > > problem actually is.
> > >
> > > I've attached the dmesgs from 2.6.6, 2.6.5, and 2.6.6 with acpi=off.
> > >
> > > There is a line that says "Invalid control registers" that I wonder where
> > > it comes from, but you might see something more here than I do.
> >
> > It comes from speedstep-centrino module, could you please try booting with
> > ACPI but without speedstep-centrino loaded? Also, does it help if you do
> > not compile/ load ACPI battery module?
> 
> Neither has any effect.
> 
> I've also disabled CONFIG_X86_SPEEDSTEP_CENTRINO_ACPI, figuring there might be
> some problem with that, but that doesn't change a thing.
>

Perhaps we need to check with ACPI guys as it seems that you are affected by
this code and there were some changes to ACPI subsystem between 2.6.5 and 2.6.6.

> Also, I'm wondering if there's any point in using the CONFIG_X86_PM_TIMER
> timesource?
>

It is supposed to be unaffected by frequency changes which is a good thing. I am
running with it, but I think I've seen couple of reports that it caused time go
twice as fast.
 
-- 
Dmitry