Re: cpuinfo shows wrong MHz value

[Thomas Renninger <trenn@suse.de>]

On Monday 16 February 2009 19:19:36 Matthew Garrett wrote:
> On Mon, Feb 16, 2009 at 07:13:15PM +0100, Thomas Renninger wrote:
> > On Monday 16 February 2009 17:50:08 Matthew Garrett wrote:
> > > On Mon, Feb 16, 2009 at 02:19:50PM +0000, Thomas Renninger wrote:
> > > 
> > > > There should be a message when this driver is loaded like:
> > > > "This driver is broken. Don't use it, don't complain."
> > > 
> > > Please don't. It's perfectly valid (if dumb) for machines to
> > > depend on the CPU for passive cooling even if they don't expose any P
> > > or T states.
> > No it's not valid. If the BIOS does not export these, it could be for
> > a reason.
> 
> We have to deal with insane BIOSes on a regular basis.
In this case I will agree with BIOS developers that they have to deal
with insane OS implementations...

> > > In that case p4-clockmod is the only code that can manage it. The 
> > > removal of the user-visible cpufreq interface should be a strong
> > > enough hint that it's not intended for speed control.
> > AFAIK p4-clockmode is still not synchronized with ACPI throttling?
> 
> It ought to work fine with any systems using MSR-based throttling.

This may come from the fact that throttling via ACPI would only kick in
in emergency temperature case. Exactly then, when throttling should be
used, it won't work. I tried it once and I could easily stall the system
by throttling via ACPI and p4_clockmod.

   Thomas