Re: [iPAQ] cpu speed

From: Erik Mouw <J.A.K.Mouw.a.t.ITS.TUDelft.NL>
Date: Fri Oct 13 2000 - 16:24:59 EDT

On Fri, Oct 13, 2000 at 03:57:22PM -0400, ebunce@lhsl.com wrote:
> Sonny Mounicou wrote:
> > Looking at the code, we can change the 9 to a 10 and this would give us
> > our 206 MHz.
>
> The default speed specified there is for reference only, and it corresponds
> to the clock speed set by the iPAQ bootldr code. The clock scaling module
> does not set the speed without user intervention. I would prefer that the
> default value in scale.c was kept in sync with that set in bootldr. So
> when the Compaq folks modify the value for bootldr, it should get
> propagated to scale.c as well.

Why don't you just use the value at module insert time as default CPU
speed? I actually see no need to specify a default speed.

> A generalized hooking mechanism should probably be added to scale.c so that
> other drivers that need to adjust timing parameters can do so, including
> PCMCIA and video drivers.

Yup, that's on my ToDo list.

> There are two major open issues with the clock scaling support in scale.c
> on the iPAQ, that don't exist on the Intel Assabet board:
> 1. If you go to speed 0 the iPAQ frequently locks up.
> 2. If you try to do too large a speed increment in one step (say from 73.7
> MHz to 206MHz) you may experience a lockup (Your Mileage May Vary), but I
> havn't had trouble doing that type of change in two steps (say from 1 =
> 73.7MHz to 7 = 162.2MHz, and then transition to 10 = 206MHz).

Probably because of bad decoupling in the power supply. Never seen that
on the LART, I can easily go from 58 to 250 MHz in a single step. But
then again, the LART has an extremely good power supply design and
plenty of decoupling capacitors.

> Any clues/fixed to those two problems would be good.
> Discussion of what type of API should be setup to allow other drivers to
> register for notifications before/after speed modifications would be
> helpful.

I want to have something compatible with the current linux
powermanagement stuff. We have to anyway, otherwise it won't be
accepted by Linus. Or we have to come with a good proposal to change
the API.

Erik

-- 
J.A.K. (Erik) Mouw, Information and Communication Theory Group, Department
of Electrical Engineering, Faculty of Information Technology and Systems,
Delft University of Technology, PO BOX 5031,  2600 GA Delft, The Netherlands
Phone: +31-15-2783635  Fax: +31-15-2781843  Email: J.A.K.Mouw@its.tudelft.nl
WWW: http://www-ict.its.tudelft.nl/~erik/
Received on Fri Oct 13 13:21:13 2000

This archive was generated by hypermail 2.1.8 : Tue May 04 2004 - 09:43:44 EDT