Re: keyboard and USB problems (Re: 2.6.2-rc1-mm2)

[Vojtech Pavlik]

On Fri, Jan 23, 2004 at 11:27:41AM -0800, john stultz wrote:
> On Fri, 2004-01-23 at 08:19, Vojtech Pavlik wrote:
> > On Fri, Jan 23, 2004 at 05:01:52PM +0100, Rudo Thomas wrote: 
> > > BogoMIPS is figured out to be 8.19 (this was already reported by another user),
> > 
> > ... this the root cause of the following problems.
> > 
> > > and i8042.c complaints with this:
> > > i8042.c: Can't write CTR while closing AUX.
> > 
> > ... bogomips is used in udelay() and that's used for waiting. If
> > bogomips is measured lower than real, the wait takes shorter and the
> > hardware doesn't do what it should in that short time.
> 
> Well, loops_per_jiffy is actually being measured correctly as we're
> using the acpi pm timesource to time udelay(). However there is a loss
> of resolution using the slower time source, so udelay(1) might take
> longer then 1 us. 

Longer udelay shouldn't cause trouble. Shorter one definitely would.

> If that is going to cause problems, then we'll need to pull out the
> use-pmtmr-for-delay_pmtmr patch. I guess our only option is then to use
> the TSC for delay_pmtrm() (as a loop based delay fails in other cases).
> I'll write that up and send it your way, Andrew. 

I've seen the PM timer breaking the mouse operation rather badly in the
past, the lost-sync check was triggering for many people when the PM
timer was used. This implies time inacurracy in the range of 0.5
seconds. Could that happen somehow?

-- 
Vojtech Pavlik
SuSE Labs, SuSE CR
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/