Re: speedstep-centrino does not load on an Intel(R) Pentium(R) M processor 1.70 GHz

Re: speedstep-centrino does not load on an Intel(R) Pentium(R) M processor 1.70 GHz

[Jeremy Fitzhardinge]

On Sun, 2004-08-01 at 21:29 -0400, Thomas Tuttle wrote:
> Okay... here's the scoop.
> 
> My laptop has one of the new Dothan processors, and neither Jeremy's
> version of speedstep-centrino.c or the patch that ignores invalid
> frequency/voltage table entries seems to fix it.  I've tried it as a
> module and compiled in, and with and without the ACPI tables option
> turned on, and I get the following results:
> 
> When I modprobe speedstep-centrino module, I get 'No such device'. 
> dmesg reveals a message stating that my processor is unsupported.
> 
> When I modprobe acpi (the "ACPI Processor Power States" module, which,
> I assume, is a way of accessing speedstep-like stuff through the ACPI
> BIOS rather than through speedstep's interface--or maybe speedstep
> uses that), I get 'No such device' also, with no messages in dmesg to
> reveal further problems.
> 
> Jeremy: I tried enabling the DEBUG option (uncommented the #define) in
> speedstep-centrino.c, but there's no more insight into the problem.

Hm.  What messages do appear in the dmesg output?  Is it "no support for
CPU model ...", "no table support for CPU model ..." or "found
unsupported CPU with Enhanced SpeedStep: ...".  Or something else?

	J

Re: Re: speedstep-centrino does not load on an Intel(R) Pentium(R) M processor 1.70 GHz

[Thomas Tuttle]

If it is compiled in to the kernel, I get the "Found unsupported
CPU..." message.
If it is a module with the original code or with the patch you gave
me, I get "No support for CPU model...email [you]".
If I make a module with the full replacement for speedstep-centrino.c
you gave me, I get "No table support for CPU model".

I'm not sure why, but it doesn't seem to be able to retrieve the
frequency/voltage tables from the BIOS, no matter how I compile it.

Are there any tests I can run to just display the tables or see if
they can be read?

On Sun, 01 Aug 2004 21:57:35 -0700, Jeremy Fitzhardinge <jeremy@goop.org> wrote:
> On Sun, 2004-08-01 at 21:29 -0400, Thomas Tuttle wrote:
> > Okay... here's the scoop.
> >
> > My laptop has one of the new Dothan processors, and neither Jeremy's
> > version of speedstep-centrino.c or the patch that ignores invalid
> > frequency/voltage table entries seems to fix it.  I've tried it as a
> > module and compiled in, and with and without the ACPI tables option
> > turned on, and I get the following results:
> >
> > When I modprobe speedstep-centrino module, I get 'No such device'.
> > dmesg reveals a message stating that my processor is unsupported.
> >
> > When I modprobe acpi (the "ACPI Processor Power States" module, which,
> > I assume, is a way of accessing speedstep-like stuff through the ACPI
> > BIOS rather than through speedstep's interface--or maybe speedstep
> > uses that), I get 'No such device' also, with no messages in dmesg to
> > reveal further problems.
> >
> > Jeremy: I tried enabling the DEBUG option (uncommented the #define) in
> > speedstep-centrino.c, but there's no more insight into the problem.
> 
> Hm.  What messages do appear in the dmesg output?  Is it "no support for
> CPU model ...", "no table support for CPU model ..." or "found
> unsupported CPU with Enhanced SpeedStep: ...".  Or something else?
> 
>         J
> 
> 


-- 
/"\                              | Thomas Tuttle
\ /    ASCII Ribbon Campaign     | Email: thinkinginbinary@gmail.com
 X   against HTML email          | AIM:   thinkinginbinary
/ \                              | Web:   http://thinkinginbinary.webhop.net
---------------------------------+

Want a Gmail invite?  Send a message to thinkinginbinary@gmail.com
with "request_gmail_invite" in the subject (without the quotes, with
the underscore) and you will be added to my waiting list.

Re: Re: speedstep-centrino does not load on an Intel(R) Pentium(R) M processor 1.70 GHz

[Dominik Brodowski]

On Mon, Aug 02, 2004 at 07:57:40AM -0400, Thomas Tuttle wrote:
> Are there any tests I can run to just display the tables or see if
> they can be read?

There's a test module at:
http://www.brodo.de/patches/2004-04-06/cpufreq-acpi_pdump.c

please put it in arch/i386/kernel/cpu/cpufreq/ ,
append arch/i386/kernel/cpu/cpufreq/Makefile
with the following line:
obj-m += cpufreq-acpi_pdump.o

re-compile, make modules_install, modprobe cpufreq-acpi_pdump
[it'll fail], and check for the entries in dmesg.

Also, please post a /proc/cpuinfo

Thanks,

	Dominik