Re: Build PST for AthlonXP

Re: Build PST for AthlonXP

[Lukas Schrangl]

> On Wed, Feb 04, 2004 at 11:45:25AM +0100, Lukas Schrangl wrote:
> > Hi!
> > 
> > I don't know wether this is the right mailinglist for my problem, but at
> > least I think here are some people who can help me.
> 
> Correct, the mailling list is the CPUFreq mailing list..
> 
> > Maybe someone remembers my problem: I have a AMD Athlon XP 2600+ mobile
> CPU
> > but neither ACPI- performance states nor cpufreq/powernow-k7 (no PST
> matching
> > this cpuid) do work.
> 
> Retrieving configuration from ACPI is a work in progress.  I do have
> some experimental patch for, but not publically available yet
> (it depends on others patches from Dominik, I'm not sure if that OK by
> now).
> 
> > Recently I browsed a bit through the powernow-k7 source and have now an
> idea
> > how it works. It shouldn't be too difficult to set up my own PST for my
> CPU,
> > should it?
> 
> Well, not so, especially if you can retrieve the configuration for the
> different
> states.
> The key is the control word, in the _PSS packages.
> 
> For example, suppose you have that:
> 
> Processor(..) {
> 	...
> 	Name (_PSS, Package () {	/* this may be a method, btw */
> 		Package () {
> 			0x0352,
> 			0x61A8,
> 			0x7D,
> 			0x009C414B,	/* that the control. */
> 			0x014B
> 		},
> 		Package () {
> 			...,
> 			...,
> 		}
> 	})
> 
> For the first one, this is 0x009C414B.
> The layout seems to be:
> bit 0-5: FID,
> bit 6-10: VID,
> bit 11-30: SGTC (or in other word 'latency' in 10 time micro second)
> bit 31-32: reserved.
> 
> Note that this is *not* confirmed by AMD people.
> 
> > I know about FIDs and VIDs and what they are good for, but can anyone
> > explain me the structure of a PST and the PSB (links to some documents,
> ...)?
> > The second thing: I could get the cpu- frequences that seem work (for my
> > FIDs) from my SSDT, but how can I calculate the voltage (for the VIDs)
> matching
> > the frequency, if there are only the mW given in the SSDT?
> > 
> 
> BTW, I send to CPUFreq ML yesterday a patch in order to get a sysctl
> interface for overwritting a powernow k7 table.
> 
> http://www.poupinou.org/cpufreq/powernow-k7/
> 
> Cheers,
> 
> -- 
> Bruno Ducrot
> 
> --  Which is worse:  ignorance or apathy?
> --  Don't know.  Don't care.
> 
> 
> -------------------------------------------------------
> The SF.Net email is sponsored by EclipseCon 2004
> Premiere Conference on Open Tools Development and Integration
> See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
> http://www.eclipsecon.org/osdn
> _______________________________________________
> Acpi-devel mailing list
> Acpi-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org
> https://lists.sourceforge.net/lists/listinfo/acpi-devel
> 

Sysctl? I just hard-coded a PST (thanks to your hint with the SSDT-
controldwords) and patched powernow-k7.c to make the driver use that PST instead of
BIOS PSTs. What shall I say? It simply works ;)

-Lukas-

PS.: I found one of the FIDs in the SSDT to be false, so maybe it's not save
to read the powernow- states from ACPI.

-- 
GMX ProMail (250 MB Mailbox, 50 FreeSMS, Virenschutz, 2,99 EUR/Monat...)
jetzt 3 Monate GRATIS + 3x DER SPIEGEL +++ http://www.gmx.net/derspiegel +++



-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: Build PST for AthlonXP

[Bruno Ducrot]

On Thu, Feb 05, 2004 at 08:17:53AM +0100, Lukas Schrangl wrote:
> > On Wed, Feb 04, 2004 at 11:45:25AM +0100, Lukas Schrangl wrote:
> > > Hi!
> > > 
> > > I don't know wether this is the right mailinglist for my problem, but at
> > > least I think here are some people who can help me.
> > 
> > Correct, the mailling list is the CPUFreq mailing list..
> > 
> > > Maybe someone remembers my problem: I have a AMD Athlon XP 2600+ mobile
> > CPU
> > > but neither ACPI- performance states nor cpufreq/powernow-k7 (no PST
> > matching
> > > this cpuid) do work.
> > 
> > Retrieving configuration from ACPI is a work in progress.  I do have
> > some experimental patch for, but not publically available yet
> > (it depends on others patches from Dominik, I'm not sure if that OK by
> > now).
> > 
> > > Recently I browsed a bit through the powernow-k7 source and have now an
> > idea
> > > how it works. It shouldn't be too difficult to set up my own PST for my
> > CPU,
> > > should it?
> > 
> > Well, not so, especially if you can retrieve the configuration for the
> > different
> > states.
> > The key is the control word, in the _PSS packages.
> > 
> > For example, suppose you have that:
> > 
> > Processor(..) {
> > 	...
> > 	Name (_PSS, Package () {	/* this may be a method, btw */
> > 		Package () {
> > 			0x0352,
> > 			0x61A8,
> > 			0x7D,
> > 			0x009C414B,	/* that the control. */
> > 			0x014B
> > 		},
> > 		Package () {
> > 			...,
> > 			...,
> > 		}
> > 	})
> > 
> > For the first one, this is 0x009C414B.
> > The layout seems to be:
> > bit 0-5: FID,
> > bit 6-10: VID,
> > bit 11-30: SGTC (or in other word 'latency' in 10 time micro second)
> > bit 31-32: reserved.
> > 
> > Note that this is *not* confirmed by AMD people.
> > 
> > > I know about FIDs and VIDs and what they are good for, but can anyone
> > > explain me the structure of a PST and the PSB (links to some documents,
> > ...)?
> > > The second thing: I could get the cpu- frequences that seem work (for my
> > > FIDs) from my SSDT, but how can I calculate the voltage (for the VIDs)
> > matching
> > > the frequency, if there are only the mW given in the SSDT?
> > > 
> > 
> > BTW, I send to CPUFreq ML yesterday a patch in order to get a sysctl
> > interface for overwritting a powernow k7 table.
> > 
> > http://www.poupinou.org/cpufreq/powernow-k7/
> > 
> > Cheers,
> > 
> > -- 
> > Bruno Ducrot
> > 
> > --  Which is worse:  ignorance or apathy?
> > --  Don't know.  Don't care.
> > 
> > 
> > -------------------------------------------------------
> > The SF.Net email is sponsored by EclipseCon 2004
> > Premiere Conference on Open Tools Development and Integration
> > See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
> > http://www.eclipsecon.org/osdn
> > _______________________________________________
> > Acpi-devel mailing list
> > Acpi-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org
> > https://lists.sourceforge.net/lists/listinfo/acpi-devel
> > 
> 
> Sysctl? I just hard-coded a PST (thanks to your hint with the SSDT-
> controldwords) and patched powernow-k7.c to make the driver use that PST instead of
> BIOS PSTs. What shall I say? It simply works ;)

I do not doubt that it work, just that the sysfs things is generic for
everybody (even if the driver is correctly loaded in fact), and then
you don't have to worry about to patch the kernel anymore.  Just
a little rc-script to write...

> PS.: I found one of the FIDs in the SSDT to be false, so maybe it's not save
> to read the powernow- states from ACPI.

Let me guess.  The one that is before the latest state, and actually
you meant the VID value which is a little too high?

If that's the case, it should be safe to go, but yes, I
have to workaround that though.  But if that's really the FID
which is wrong, well, I will be gratefull to get a report.

I saw also one example where the maximum speed was missing (and
of course, there is no BIOS PST)...

Note also that the acpi things is only done if and only if
the laptop is DMI blacklisted (and ACPI processor is compiled,
of course).

-- 
Bruno Ducrot

--  Which is worse:  ignorance or apathy?
--  Don't know.  Don't care.


-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: Build PST for AthlonXP

[Lukas Schrangl]

> On Thu, Feb 05, 2004 at 08:17:53AM +0100, Lukas Schrangl wrote:
> > > On Wed, Feb 04, 2004 at 11:45:25AM +0100, Lukas Schrangl wrote:
> > > > Hi!
> > > > 
> > > > I don't know wether this is the right mailinglist for my problem,
> but at
> > > > least I think here are some people who can help me.
> > > 
> > > Correct, the mailling list is the CPUFreq mailing list..
> > > 
> > > > Maybe someone remembers my problem: I have a AMD Athlon XP 2600+
> mobile
> > > CPU
> > > > but neither ACPI- performance states nor cpufreq/powernow-k7 (no PST
> > > matching
> > > > this cpuid) do work.
> > > 
> > > Retrieving configuration from ACPI is a work in progress.  I do have
> > > some experimental patch for, but not publically available yet
> > > (it depends on others patches from Dominik, I'm not sure if that OK by
> > > now).
> > > 
> > > > Recently I browsed a bit through the powernow-k7 source and have now
> an
> > > idea
> > > > how it works. It shouldn't be too difficult to set up my own PST for
> my
> > > CPU,
> > > > should it?
> > > 
> > > Well, not so, especially if you can retrieve the configuration for the
> > > different
> > > states.
> > > The key is the control word, in the _PSS packages.
> > > 
> > > For example, suppose you have that:
> > > 
> > > Processor(..) {
> > > 	...
> > > 	Name (_PSS, Package () {	/* this may be a method, btw */
> > > 		Package () {
> > > 			0x0352,
> > > 			0x61A8,
> > > 			0x7D,
> > > 			0x009C414B,	/* that the control. */
> > > 			0x014B
> > > 		},
> > > 		Package () {
> > > 			...,
> > > 			...,
> > > 		}
> > > 	})
> > > 
> > > For the first one, this is 0x009C414B.
> > > The layout seems to be:
> > > bit 0-5: FID,
> > > bit 6-10: VID,
> > > bit 11-30: SGTC (or in other word 'latency' in 10 time micro second)
> > > bit 31-32: reserved.
> > > 
> > > Note that this is *not* confirmed by AMD people.
> > > 
> > > > I know about FIDs and VIDs and what they are good for, but can
> anyone
> > > > explain me the structure of a PST and the PSB (links to some
> documents,
> > > ...)?
> > > > The second thing: I could get the cpu- frequences that seem work
> (for my
> > > > FIDs) from my SSDT, but how can I calculate the voltage (for the
> VIDs)
> > > matching
> > > > the frequency, if there are only the mW given in the SSDT?
> > > > 
> > > 
> > > BTW, I send to CPUFreq ML yesterday a patch in order to get a sysctl
> > > interface for overwritting a powernow k7 table.
> > > 
> > > http://www.poupinou.org/cpufreq/powernow-k7/
> > > 
> > > Cheers,
> > > 
> > > -- 
> > > Bruno Ducrot
> > > 
> > > --  Which is worse:  ignorance or apathy?
> > > --  Don't know.  Don't care.
> > > 
> > > 
> > > -------------------------------------------------------
> > > The SF.Net email is sponsored by EclipseCon 2004
> > > Premiere Conference on Open Tools Development and Integration
> > > See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
> > > http://www.eclipsecon.org/osdn
> > > _______________________________________________
> > > Acpi-devel mailing list
> > > Acpi-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org
> > > https://lists.sourceforge.net/lists/listinfo/acpi-devel
> > > 
> > 
> > Sysctl? I just hard-coded a PST (thanks to your hint with the SSDT-
> > controldwords) and patched powernow-k7.c to make the driver use that PST
> instead of
> > BIOS PSTs. What shall I say? It simply works ;)
> 
> I do not doubt that it work, just that the sysfs things is generic for
> everybody (even if the driver is correctly loaded in fact), and then
> you don't have to worry about to patch the kernel anymore.  Just
> a little rc-script to write...
> 
> > PS.: I found one of the FIDs in the SSDT to be false, so maybe it's not
> save
> > to read the powernow- states from ACPI.
> 
> Let me guess.  The one that is before the latest state, and actually
> you meant the VID value which is a little too high?
> 
> If that's the case, it should be safe to go, but yes, I
> have to workaround that though.  But if that's really the FID
> which is wrong, well, I will be gratefull to get a report.

I think it's really the FID in the state one before the last state which was
too low by one but I will look that up to be shure (I'm not working with my
computer). I will notify you as soon as I have looked that up and got to a
computer with internet access again.

> 
> I saw also one example where the maximum speed was missing (and
> of course, there is no BIOS PST)...

I have all states.

> 
> Note also that the acpi things is only done if and only if
> the laptop is DMI blacklisted (and ACPI processor is compiled,
> of course).

Don't think it is blacklisted, but didn't try your patch yet. Will do that
as soon as possible.

> 
> -- 
> Bruno Ducrot
> 
> --  Which is worse:  ignorance or apathy?
> --  Don't know.  Don't care.
> 

-- 
GMX ProMail (250 MB Mailbox, 50 FreeSMS, Virenschutz, 2,99 EUR/Monat...)
jetzt 3 Monate GRATIS + 3x DER SPIEGEL +++ http://www.gmx.net/derspiegel +++



-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: Build PST for AthlonXP

[Bruno Ducrot]

[-- Attachment #1: Type: text/plain, Size: 1563 bytes --]

On Thu, Feb 05, 2004 at 03:01:20PM +0100, Lukas Schrangl wrote:
> > On Thu, Feb 05, 2004 at 08:17:53AM +0100, Lukas Schrangl wrote:

> > > to read the powernow- states from ACPI.
> > 
> > Let me guess.  The one that is before the latest state, and actually
> > you meant the VID value which is a little too high?
> > 
> > If that's the case, it should be safe to go, but yes, I
> > have to workaround that though.  But if that's really the FID
> > which is wrong, well, I will be gratefull to get a report.
> 
> I think it's really the FID in the state one before the last state which was
> too low by one but I will look that up to be shure (I'm not working with my
> computer). I will notify you as soon as I have looked that up and got to a
> computer with internet access again.

Well, well. Ok.

> > the laptop is DMI blacklisted (and ACPI processor is compiled,
> > of course).
> 
> Don't think it is blacklisted, but didn't try your patch yet. Will do that
> as soon as possible.

If you want, you can test that dummy module instead.  Nothing will
be actually touched.  You need at least a 2.6.2-rc3-mm1, or a
recent ACPI test bk tree (this module is actually an hack of a module
made by Dominik).

Create a Makefile with that line:
obj-m = test-powernow.o
then
make -C /lib/modules/`uname -r`/build SUBDIRS=`pwd` modules
insmod ./test-powernow.ko

It will failed, but dmesg will give you the table.  You need to
have CONFIG_ACPI_PROCESSOR enabled, though.

Cheers,

-- 
Bruno Ducrot

--  Which is worse:  ignorance or apathy?
--  Don't know.  Don't care.

[-- Attachment #2: test-powernow.c --]
[-- Type: text/x-csrc, Size: 1919 bytes --]

/*
 */

#include <linux/config.h>
#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/init.h>
#include <linux/cpufreq.h>
#include <linux/proc_fs.h>
#include <linux/seq_file.h>
#include <asm/io.h>
#include <asm/delay.h>
#include <asm/uaccess.h>

#include <linux/acpi.h>
#include <acpi/processor.h>

struct acpi_processor_performance p;

union powernow_acpi_control_t {
	struct {
		unsigned long fid:5,
		vid:5,
		sgtc:20,
		res1:2;
	} bits;
	unsigned long val;
};

#define ACPI_PDC_REVISION_ID                   0x1
#define ACPI_PDC_CAPABILITY_ENHANCED_SPEEDSTEP 0x1

static int __init
acpi_cpufreq_init (void)
{
	unsigned int i;
	struct acpi_pct_register *r;
	union powernow_acpi_control_t pc;

	if (acpi_processor_register_performance(&p, 0))
		return -EIO;

	printk("number of states: %d\n", p.state_count);
	for (i=0; i< p.state_count; i++) {
		pc.val = (unsigned long) p.states[i].control;
		printk(KERN_INFO "powernow: %cP%d: %d MHz, %d mW, %d uS, control %08x, status %08x, vid: %02x fid: %02x SGTC: %d\n",
		       (i == p.state?'*':' '), i,
		       (u32) p.states[i].core_frequency,
		       (u32) p.states[i].power,
		       (u32) p.states[i].transition_latency,
		       (u32) p.states[i].control,
		       (u32) p.states[i].status,
		       pc.bits.vid,
		       pc.bits.fid,
		       pc.bits.sgtc);

	}

	printk("control_register:\n");
	r = &p.control_register;
	printk("%d %d %d %d %d %d %lld\n", r->descriptor, r->length, r->space_id,
	       r->bit_width, r->bit_offset, r->reserved, r->address);

	printk("status_register:\n");
	r = &p.status_register;
	printk("%d %d %d %d %d %d %lld\n", r->descriptor, r->length, r->space_id,
	       r->bit_width, r->bit_offset, r->reserved, r->address);

	acpi_processor_unregister_performance(&p, 0);

	return -ENODEV;
}


static void __exit
acpi_cpufreq_exit (void)
{
	return;
}


late_initcall(acpi_cpufreq_init);
module_exit(acpi_cpufreq_exit);

Build PST for AthlonXP

[Lukas Schrangl]

Hi!

I don't know wether this is the right mailinglist for my problem, but at
least I think here are some people who can help me.

Maybe someone remembers my problem: I have a AMD Athlon XP 2600+ mobile CPU
but neither ACPI- performance states nor cpufreq/powernow-k7 (no PST matching
this cpuid) do work.

Recently I browsed a bit through the powernow-k7 source and have now an idea
how it works. It shouldn't be too difficult to set up my own PST for my CPU,
should it?
I know about FIDs and VIDs and what they are good for, but can anyone
explain me the structure of a PST and the PSB (links to some documents, ...)?
The second thing: I could get the cpu- frequences that seem work (for my
FIDs) from my SSDT, but how can I calculate the voltage (for the VIDs) matching
the frequency, if there are only the mW given in the SSDT?

-Lukas-

-- 
GMX ProMail (250 MB Mailbox, 50 FreeSMS, Virenschutz, 2,99 EUR/Monat...)
jetzt 3 Monate GRATIS + 3x DER SPIEGEL +++ http://www.gmx.net/derspiegel +++



-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: Build PST for AthlonXP

[Bruno Ducrot]

On Wed, Feb 04, 2004 at 11:45:25AM +0100, Lukas Schrangl wrote:
> Hi!
> 
> I don't know wether this is the right mailinglist for my problem, but at
> least I think here are some people who can help me.

Correct, the mailling list is the CPUFreq mailing list..

> Maybe someone remembers my problem: I have a AMD Athlon XP 2600+ mobile CPU
> but neither ACPI- performance states nor cpufreq/powernow-k7 (no PST matching
> this cpuid) do work.

Retrieving configuration from ACPI is a work in progress.  I do have
some experimental patch for, but not publically available yet
(it depends on others patches from Dominik, I'm not sure if that OK by
now).

> Recently I browsed a bit through the powernow-k7 source and have now an idea
> how it works. It shouldn't be too difficult to set up my own PST for my CPU,
> should it?

Well, not so, especially if you can retrieve the configuration for the different
states.
The key is the control word, in the _PSS packages.

For example, suppose you have that:

Processor(..) {
	...
	Name (_PSS, Package () {	/* this may be a method, btw */
		Package () {
			0x0352,
			0x61A8,
			0x7D,
			0x009C414B,	/* that the control. */
			0x014B
		},
		Package () {
			...,
			...,
		}
	})

For the first one, this is 0x009C414B.
The layout seems to be:
bit 0-5: FID,
bit 6-10: VID,
bit 11-30: SGTC (or in other word 'latency' in 10 time micro second)
bit 31-32: reserved.

Note that this is *not* confirmed by AMD people.

> I know about FIDs and VIDs and what they are good for, but can anyone
> explain me the structure of a PST and the PSB (links to some documents, ...)?
> The second thing: I could get the cpu- frequences that seem work (for my
> FIDs) from my SSDT, but how can I calculate the voltage (for the VIDs) matching
> the frequency, if there are only the mW given in the SSDT?
> 

BTW, I send to CPUFreq ML yesterday a patch in order to get a sysctl
interface for overwritting a powernow k7 table.

http://www.poupinou.org/cpufreq/powernow-k7/

Cheers,

-- 
Bruno Ducrot

--  Which is worse:  ignorance or apathy?
--  Don't know.  Don't care.


-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn