RE: SSDT on Sony VAIO

RE: SSDT on Sony VAIO

[Grover, Andrew]

> From: Nathan Conrad [mailto:conrad-aYgNHrD+FxfR7s880joybQ@public.gmane.org] 
> I am using kernel 2.5.55 on a Sony Vaio PCG-FXA49. Right now,
> /proc/acpi/processor/CPU0/info says that performance management is
> not available. My CPU (1.2 GHz mobile Athlon4) supports PowerNow, and
> I would expect this would qualify as performance management.
> 
> I have looked at my ACPI tables and there is a listing of 5 P-states
> in my SSDT (1200 MHz,1000,800,600,500 MHz).
> 
> Why are these not listed in my /proc/acpi/processor/CPU0 directory?
> 
> Does the current Linux ACPI driver look at the SSDT tables, or does it
> only look at the DSDT for AML?

It should consume both SSDT and DSDT tables.

You might try taking a look at the "performance management" section of
drivers/acpi/processor.c. Stick some printk's in
acpi_processor_get_performance_states and see where things are going
awry.

Also, if _PCT isn't present in the namespace as well, then it flags perf
support as not available.

Regards -- Andy


-------------------------------------------------------
This SF.NET email is sponsored by: A Thawte Code Signing Certificate 
is essential in establishing user confidence by providing assurance of 
authenticity and code integrity. Download our Free Code Signing guide:
http://ads.sourceforge.net/cgi-bin/redirect.pl?thaw0028en

Re: SSDT on Sony VAIO

[Dominik Brodowski]

On Wed, Jan 15, 2003 at 05:08:40PM -0800, Grover, Andrew wrote:
> > From: Nathan Conrad [mailto:conrad-aYgNHrD+FxfR7s880joybQ@public.gmane.org] 
> > I am using kernel 2.5.55 on a Sony Vaio PCG-FXA49. Right now,
> > /proc/acpi/processor/CPU0/info says that performance management is
> > not available. My CPU (1.2 GHz mobile Athlon4) supports PowerNow, and
> > I would expect this would qualify as performance management.
> > 
> > I have looked at my ACPI tables and there is a listing of 5 P-states
> > in my SSDT (1200 MHz,1000,800,600,500 MHz).
> > 
> > Why are these not listed in my /proc/acpi/processor/CPU0 directory?
> > 
> > Does the current Linux ACPI driver look at the SSDT tables, or does it
> > only look at the DSDT for AML?
> 
> It should consume both SSDT and DSDT tables.
> 
> You might try taking a look at the "performance management" section of
> drivers/acpi/processor.c. Stick some printk's in
> acpi_processor_get_performance_states and see where things are going
> awry.
> 
> Also, if _PCT isn't present in the namespace as well, then it flags perf
> support as not available.

Also, do you have CONFIG_CPU_FREQ and CONFIG_ACPI_PROCESSOR_PERF enabled? If
not, support for "performance management" will be disabled.

	Dominik


-------------------------------------------------------
This SF.NET email is sponsored by: A Thawte Code Signing Certificate 
is essential in establishing user confidence by providing assurance of 
authenticity and code integrity. Download our Free Code Signing guide:
http://ads.sourceforge.net/cgi-bin/redirect.pl?thaw0028en

Re: SSDT on Sony VAIO

[Nathan Conrad]

[-- Attachment #1.1: Type: text/plain, Size: 2055 bytes --]

I did not have CONFIG_CPU_FREQ and CONFIG_ACPI_PROCESSOR_PERF
enabled. Once I enabled them, a performance file appeared in my
/proc/acpi/processor/CPU0 directory. This file contained '<not
supported>'. I also noticed the following message when I boot:

acpi_processor-0959 [17] acpi_processor_get_per: Unsupported address
space [127] (control_register)

Is there a way for me to examine the ACPI namespace in order to
determine what the parser has put in my CPU0 structure?

In what way are cpufreq and ACPI related? My laptop has an Athlon4 in
it, which is not supported by the kernel's cpufreq (yet).

My SSDT is attached.

-Nathan

On Thu, Jan 16, 2003 at 08:46:24AM +0100, Dominik Brodowski wrote:
> On Wed, Jan 15, 2003 at 05:08:40PM -0800, Grover, Andrew wrote:
> > > From: Nathan Conrad [mailto:conrad-aYgNHrD+FxfR7s880joybQ@public.gmane.org] 
> > > I am using kernel 2.5.55 on a Sony Vaio PCG-FXA49. Right now,
> > > /proc/acpi/processor/CPU0/info says that performance management is
> > > not available. My CPU (1.2 GHz mobile Athlon4) supports PowerNow, and
> > > I would expect this would qualify as performance management.
> > > 
> > > I have looked at my ACPI tables and there is a listing of 5 P-states
> > > in my SSDT (1200 MHz,1000,800,600,500 MHz).
> > > 
> > > Why are these not listed in my /proc/acpi/processor/CPU0 directory?
> > > 
> > > Does the current Linux ACPI driver look at the SSDT tables, or does it
> > > only look at the DSDT for AML?
> > 
> > It should consume both SSDT and DSDT tables.
> > 
> > You might try taking a look at the "performance management" section of
> > drivers/acpi/processor.c. Stick some printk's in
> > acpi_processor_get_performance_states and see where things are going
> > awry.
> > 
> > Also, if _PCT isn't present in the namespace as well, then it flags perf
> > support as not available.
> 
> Also, do you have CONFIG_CPU_FREQ and CONFIG_ACPI_PROCESSOR_PERF enabled? If
> not, support for "performance management" will be disabled.
> 
> 	Dominik

[-- Attachment #1.2: ssdt --]
[-- Type: application/octet-stream, Size: 280 bytes --]

[-- Attachment #1.3: dsdt --]
[-- Type: application/octet-stream, Size: 20627 bytes --]

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

RE: SSDT on Sony VAIO

[Adachi, Kenichi]

Hi Nathan, 
Can you please take a look at the following msg that I posted before? It
tells all about the unsupported address space 127. 

http://marc.theaimsgroup.com/?l=acpi4linux&m=104022580902620&w=2

Incidentally I didn't know the existence of CpuFreq project when I made
the above statement. Dominik, if BIOS tells OS via ACPI _PCT to access
FFH to initiate FID/VID change (in AMD PowerNow!), how current code
coordinates it? I'm going to look into code soon, but I'd appreciate
very much if you give me some insight in advance.

Thanks,
- Adachi, Kenichi

> -----Original Message-----
> From: acpi-devel-admin-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org 
> [mailto:acpi-devel-admin-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org] On Behalf Of 
> Nathan Conrad
> Sent: Friday, January 17, 2003 1:51 AM
> To: Dominik Brodowski
> Cc: acpi-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org; andrew.grover@intel.com
> Subject: Re: [ACPI] SSDT on Sony VAIO
> 
> 
> I did not have CONFIG_CPU_FREQ and CONFIG_ACPI_PROCESSOR_PERF 
> enabled. Once I enabled them, a performance file appeared in 
> my /proc/acpi/processor/CPU0 directory. This file contained '<not
> supported>'. I also noticed the following message when I boot:
> 
> acpi_processor-0959 [17] acpi_processor_get_per: Unsupported 
> address space [127] (control_register)
> 
> Is there a way for me to examine the ACPI namespace in order 
> to determine what the parser has put in my CPU0 structure?
> 
> In what way are cpufreq and ACPI related? My laptop has an 
> Athlon4 in it, which is not supported by the kernel's cpufreq (yet).
> 
> My SSDT is attached.
> 
> -Nathan
> 
> On Thu, Jan 16, 2003 at 08:46:24AM +0100, Dominik Brodowski wrote:
> > On Wed, Jan 15, 2003 at 05:08:40PM -0800, Grover, Andrew wrote:
> > > > From: Nathan Conrad [mailto:conrad-aYgNHrD+FxfR7s880joybQ@public.gmane.org]
> > > > I am using kernel 2.5.55 on a Sony Vaio PCG-FXA49. Right now,
> > > > /proc/acpi/processor/CPU0/info says that performance 
> management is
> > > > not available. My CPU (1.2 GHz mobile Athlon4) supports 
> PowerNow, and
> > > > I would expect this would qualify as performance management.
> > > > 
> > > > I have looked at my ACPI tables and there is a listing of 5 
> > > > P-states in my SSDT (1200 MHz,1000,800,600,500 MHz).
> > > > 
> > > > Why are these not listed in my /proc/acpi/processor/CPU0 
> > > > directory?
> > > > 
> > > > Does the current Linux ACPI driver look at the SSDT tables, or 
> > > > does it only look at the DSDT for AML?
> > > 
> > > It should consume both SSDT and DSDT tables.
> > > 
> > > You might try taking a look at the "performance 
> management" section 
> > > of drivers/acpi/processor.c. Stick some printk's in 
> > > acpi_processor_get_performance_states and see where 
> things are going 
> > > awry.
> > > 
> > > Also, if _PCT isn't present in the namespace as well, 
> then it flags 
> > > perf support as not available.
> > 
> > Also, do you have CONFIG_CPU_FREQ and CONFIG_ACPI_PROCESSOR_PERF 
> > enabled? If not, support for "performance management" will be 
> > disabled.
> > 
> > 	Dominik
> 




-------------------------------------------------------
This SF.NET email is sponsored by: Thawte.com
Understand how to protect your customers personal information by implementing
SSL on your Apache Web Server. Click here to get our FREE Thawte Apache 
Guide: http://ads.sourceforge.net/cgi-bin/redirect.pl?thaw0029en

RE: SSDT on Sony VAIO

[Adachi, Kenichi]

Hi Nathan, 
Can you please take a look at the following msg that I posted before? It
tells all about the unsupported address space 127. 

http://marc.theaimsgroup.com/?l=acpi4linux&m=104022580902620&w=2

Incidentally I didn't know the existence of CpuFreq project when I made
the above statement. Dominik, if BIOS tells OS via ACPI _PCT to access
FFH to initiate FID/VID change (in AMD PowerNow!), how current code
coordinates it? I'm going to look into code soon, but I'd appreciate
very much if you give me some insight in advance.

Thanks,
- Adachi, Kenichi

> -----Original Message-----
> From: acpi-devel-admin-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org 
> [mailto:acpi-devel-admin-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org] On Behalf Of 
> Nathan Conrad
> Sent: Friday, January 17, 2003 1:51 AM
> To: Dominik Brodowski
> Cc: acpi-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org; andrew.grover@intel.com
> Subject: Re: [ACPI] SSDT on Sony VAIO
> 
> 
> I did not have CONFIG_CPU_FREQ and CONFIG_ACPI_PROCESSOR_PERF 
> enabled. Once I enabled them, a performance file appeared in 
> my /proc/acpi/processor/CPU0 directory. This file contained '<not
> supported>'. I also noticed the following message when I boot:
> 
> acpi_processor-0959 [17] acpi_processor_get_per: Unsupported 
> address space [127] (control_register)
> 
> Is there a way for me to examine the ACPI namespace in order 
> to determine what the parser has put in my CPU0 structure?
> 
> In what way are cpufreq and ACPI related? My laptop has an 
> Athlon4 in it, which is not supported by the kernel's cpufreq (yet).
> 
> My SSDT is attached.
> 
> -Nathan
> 
> On Thu, Jan 16, 2003 at 08:46:24AM +0100, Dominik Brodowski wrote:
> > On Wed, Jan 15, 2003 at 05:08:40PM -0800, Grover, Andrew wrote:
> > > > From: Nathan Conrad [mailto:conrad-aYgNHrD+FxfR7s880joybQ@public.gmane.org]
> > > > I am using kernel 2.5.55 on a Sony Vaio PCG-FXA49. Right now,
> > > > /proc/acpi/processor/CPU0/info says that performance 
> management is
> > > > not available. My CPU (1.2 GHz mobile Athlon4) supports 
> PowerNow, and
> > > > I would expect this would qualify as performance management.
> > > > 
> > > > I have looked at my ACPI tables and there is a listing of 5 
> > > > P-states in my SSDT (1200 MHz,1000,800,600,500 MHz).
> > > > 
> > > > Why are these not listed in my /proc/acpi/processor/CPU0 
> > > > directory?
> > > > 
> > > > Does the current Linux ACPI driver look at the SSDT tables, or 
> > > > does it only look at the DSDT for AML?
> > > 
> > > It should consume both SSDT and DSDT tables.
> > > 
> > > You might try taking a look at the "performance 
> management" section 
> > > of drivers/acpi/processor.c. Stick some printk's in 
> > > acpi_processor_get_performance_states and see where 
> things are going 
> > > awry.
> > > 
> > > Also, if _PCT isn't present in the namespace as well, 
> then it flags 
> > > perf support as not available.
> > 
> > Also, do you have CONFIG_CPU_FREQ and CONFIG_ACPI_PROCESSOR_PERF 
> > enabled? If not, support for "performance management" will be 
> > disabled.
> > 
> > 	Dominik
> 




-------------------------------------------------------
This SF.NET email is sponsored by: Thawte.com
Understand how to protect your customers personal information by implementing
SSL on your Apache Web Server. Click here to get our FREE Thawte Apache 
Guide: http://ads.sourceforge.net/cgi-bin/redirect.pl?thaw0029en

Re: SSDT on Sony VAIO

[Dominik Brodowski]

Hi,

On Fri, Jan 17, 2003 at 02:22:42AM +0900, Adachi, Kenichi wrote:
> Incidentally I didn't know the existence of CpuFreq project when I made
> the above statement. Dominik, if BIOS tells OS via ACPI _PCT to access
> FFH to initiate FID/VID change (in AMD PowerNow!), how current code
> coordinates it? I'm going to look into code soon, but I'd appreciate
> very much if you give me some insight in advance.

Currently, it fails with this error message. If you add your patch,
frequency transitions will still fail - the ACPI processor driver doesn't
know anything about the PowerNow internals. So loading of teh ACPI cpufreq
driver will be aborted at latest in acpi_cpufreq_init, if not earlier.

So, actually I can't think of any way to use the ACPI processr driver for
CPU frequency transitions in these cases - you need to use the "legacy" 
powernow-k7.c driver instead (which is currently developed by Dave Jones).

	Dominik


-------------------------------------------------------
This SF.NET email is sponsored by: Thawte.com
Understand how to protect your customers personal information by implementing
SSL on your Apache Web Server. Click here to get our FREE Thawte Apache 
Guide: http://ads.sourceforge.net/cgi-bin/redirect.pl?thaw0029en

Re: SSDT on Sony VAIO

[Dominik Brodowski]

On Thu, Jan 16, 2003 at 11:51:28AM -0500, Nathan Conrad wrote:
> I did not have CONFIG_CPU_FREQ and CONFIG_ACPI_PROCESSOR_PERF
> enabled. Once I enabled them, a performance file appeared in my
> /proc/acpi/processor/CPU0 directory. This file contained '<not
> supported>'. I also noticed the following message when I boot:
> 
> acpi_processor-0959 [17] acpi_processor_get_per: Unsupported address
> space [127] (control_register)
> 
> Is there a way for me to examine the ACPI namespace in order to
> determine what the parser has put in my CPU0 structure?
> 
> In what way are cpufreq and ACPI related? My laptop has an Athlon4 in
> it, which is not supported by the kernel's cpufreq (yet).

cpufreq is a generic, cross-architecture kernel infrastructure for CPU
frequency and voltage scaling support. This has several advantages:

- the user shouldn't have to care what technique is used to scale the CPU
frequency and voltage, whether that is a legacy interface like speedstep, 
powernow-k6, longrun etc. or whether the interface is "ACPI P-States
support". All these techniques can be managed through one unified userspace 
interface ( http://www.brodo.de/cpufreq/interface.html )

- on almost all architectures, timing codes need to be updated (so that
"sleep 1" still takes one second, and not two when switching to low 
frequency). The cpufreq core takes care of that, independent of what
scaling technique is used.

- dynamic voltage scaling only has to be implemented at one place, and not
for each "cpufreq driver" on its own. ( DVS isn't yet ready for prime time, 
but that will change soon. If you're interested, you can follow the 
development at http://www.brodo.de/cpufreq/advanced.html )

- because of these reasons, ACPI P-States use the cpufreq infrastructure in
2.5. So you need to enable the cpufreq core if you want to use ACPI
P-States. No need for the legacy drivers, though, in case ACPI P-States
work.

Concerning the powernow-k7 / Athlon driver: Dave Jones is working on it. 

	Dominik


-------------------------------------------------------
This SF.NET email is sponsored by: Thawte.com
Understand how to protect your customers personal information by implementing
SSL on your Apache Web Server. Click here to get our FREE Thawte Apache 
Guide: http://ads.sourceforge.net/cgi-bin/redirect.pl?thaw0029en

Re: SSDT on Sony VAIO

[Jens Haug]

> On Fri, Jan 17, 2003 at 02:22:42AM +0900, Adachi, Kenichi wrote:
> > Incidentally I didn't know the existence of CpuFreq project when I made
> > the above statement. Dominik, if BIOS tells OS via ACPI _PCT to access
> > FFH to initiate FID/VID change (in AMD PowerNow!), how current code
> > coordinates it? I'm going to look into code soon, but I'd appreciate
> > very much if you give me some insight in advance.
> 
> Currently, it fails with this error message. If you add your patch,
> frequency transitions will still fail - the ACPI processor driver doesn't
> know anything about the PowerNow internals. So loading of teh ACPI cpufreq
> driver will be aborted at latest in acpi_cpufreq_init, if not earlier.
> 
> So, actually I can't think of any way to use the ACPI processr driver for
> CPU frequency transitions in these cases - you need to use the "legacy" 
> powernow-k7.c driver instead (which is currently developed by Dave Jones).

Did I understand this? I'm not sure.
ACPI P-States use the cpufreq infrastructure (in 2.5.x). Cpufreq
itself uses either the ACPI processor driver (for supported CPUs) 
or a special driver (like powernow-k7).
So, in the future, ACPI P-States will use Cpufreq, Cpufreq will
use powernow-k7, and everybody will be happy. Or can't we use
P-States here at all, and everything will be done by Cpufreq like
on a non-ACPI system?


Jens

-- 
Jens Haug
IKFF Universität Stuttgart              Tel. 0711/685-6422
Pfaffenwaldring 9                       Fax  0711/685-6356
70550 Stuttgart	                haug-X6ztD3ggwzuBAmxm6OvjtTjhTm2NLCe8@public.gmane.org



-------------------------------------------------------
This SF.NET email is sponsored by: Thawte.com
Understand how to protect your customers personal information by implementing
SSL on your Apache Web Server. Click here to get our FREE Thawte Apache
Guide: http://ads.sourceforge.net/cgi-bin/redirect.pl?thaw0029en

Re: SSDT on Sony VAIO

[Jens Haug]

> > acpi_processor-0959 [17] acpi_processor_get_per: Unsupported address
> > space [127] (control_register)
> > 
> > Is there a way for me to examine the ACPI namespace in order to
> > determine what the parser has put in my CPU0 structure?
> > 
> > In what way are cpufreq and ACPI related? My laptop has an Athlon4 in
> > it, which is not supported by the kernel's cpufreq (yet).
(...)
> Concerning the powernow-k7 / Athlon driver: Dave Jones is working on it. 

If you're impatient and your CPU is a Mobile Athlon 4 (not XP) with
1200MHz or lower, then you might want to try rtdvs with the 
patch by David Kimdom. With a newer CPU you'd have to do some 
hacking in the driver.
http://nervous.oltrelinux.com/hw/linux_aspire1300xv.html#PowerNow
For me it works fine. It's not a very smart solution (like CPUfreq
will be) and I only set the voltage and frequency manually, but I
can save a lot of power (50%).


Jens

-- 
Jens Haug
IKFF Universität Stuttgart              Tel. 0711/685-6422
Pfaffenwaldring 9                       Fax  0711/685-6356
70550 Stuttgart	                haug-X6ztD3ggwzuBAmxm6OvjtTjhTm2NLCe8@public.gmane.org



-------------------------------------------------------
This SF.NET email is sponsored by: Thawte.com
Understand how to protect your customers personal information by implementing
SSL on your Apache Web Server. Click here to get our FREE Thawte Apache
Guide: http://ads.sourceforge.net/cgi-bin/redirect.pl?thaw0029en

SSDT on Sony VAIO

[Nathan Conrad]

[-- Attachment #1: Type: text/plain, Size: 990 bytes --]

Hi,

I am using kernel 2.5.55 on a Sony Vaio PCG-FXA49. Right now,
/proc/acpi/processor/CPU0/info says that performance management is
not available. My CPU (1.2 GHz mobile Athlon4) supports PowerNow, and
I would expect this would qualify as performance management.

I have looked at my ACPI tables and there is a listing of 5 P-states
in my SSDT (1200 MHz,1000,800,600,500 MHz).

Why are these not listed in my /proc/acpi/processor/CPU0 directory?

Does the current Linux ACPI driver look at the SSDT tables, or does it
only look at the DSDT for AML?

One thing to note is that my DSDT has a very minimal \_PR section:

    Scope (\_PR)
    {
        Processor (CPU0, 0x01, 0x00008010, 0x06) {}
    }

The SSDT has a more filled out one. I do not know how to disassemble
it (iasl does not know how to handle a SSDT).

-Nathan

--
Nathan J. Conrad   (XXX)-687-7449    https://bungled.net/
312 Moore Hall // UNC-C // Charlotte, NC 28223
GPG: F4FC 7E25 9308 ECE1 735C  0798 CE86 DA45 9170 3112

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]