[lm-sensors] [PATCH] hwmon: (coretemp) Document Atom E660 support

[lm-sensors] [PATCH] hwmon: (coretemp) Document Atom E660 support

[Jean Delvare]

Document the new Atom series (Tunnel Creek) as being supported, and
list TjMax for the Atom E660 and E660T.

Signed-off-by: Jean Delvare <khali@linux-fr.org>
Cc: Alexander Stein <alexander.stein@systec-electronic.com>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: Guenter Roeck <guenter.roeck@ericsson.com>
---
As reported by Alexander weeks ago.

 Documentation/hwmon/coretemp |    5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

--- linux-3.1-rc4.orig/Documentation/hwmon/coretemp	2011-09-28 12:45:58.000000000 +0200
+++ linux-3.1-rc4/Documentation/hwmon/coretemp	2011-09-28 14:28:24.000000000 +0200
@@ -6,7 +6,8 @@ Supported chips:
     Prefix: 'coretemp'
     CPUID: family 0x6, models 0xe (Pentium M DC), 0xf (Core 2 DC 65nm),
                               0x16 (Core 2 SC 65nm), 0x17 (Penryn 45nm),
-                              0x1a (Nehalem), 0x1c (Atom), 0x1e (Lynnfield)
+                              0x1a (Nehalem), 0x1c (Atom), 0x1e (Lynnfield),
+                              0x26 (Tunnel Creek Atom)
     Datasheet: Intel 64 and IA-32 Architectures Software Developer's Manual
                Volume 3A: System Programming Guide
                http://softwarecommunity.intel.com/Wiki/Mobility/720.htm
@@ -85,6 +86,8 @@ Process		Processor					TjMax(C)
 		N475/470/455/450				100
 		N280/270					90
 		330/230						125
+		E660						90
+		E660T						110
 
 45nm		Core2 Processors
 		Solo ULV SU3500/3300				100


-- 
Jean Delvare

_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors

Re: [lm-sensors] [PATCH] hwmon: (coretemp) Document Atom E660

[Jean Delvare]

Hi Durga,

On Wed, 28 Sep 2011 18:36:50 +0530, R, Durgadoss wrote:
> I have the CPU model of the next generation Atom based Core.
> Can we add this also in the list here ?
> If you don't find time, I can add the one liner, after this gets in.
> 
> )
> > +                              0x1a (Nehalem), 0x1c (Atom), 0x1e (Lynnfield),
> > +                              0x26 (Tunnel Creek Atom)
> 
> 0x27 (Medfield Atom) and TjMax 90

I can add it, no problem. Is TjMax the same for all Medfield-based Atom
CPUs? If so, and if it can't be read from MSR_TEMPERATURE_TARGET, then
this value shouldn't go in the documentation file but in the driver,
unless the fragile heuristic we have already comes to that conclusion
(it's so complex that honestly I can't tell.)

> 
> >      Datasheet: Intel 64 and IA-32 Architectures Software Developer's Manual
> >                 Volume 3A: System Programming Guide
> >                 http://softwarecommunity.intel.com/Wiki/Mobility/720.htm
> > @@ -85,6 +86,8 @@ Process		Processor
> > 	TjMax(C)
> >  		N475/470/455/450				100
> >  		N280/270					90
> >  		330/230						125
> > +		E660						90
> > +		E660T						110
> > 
> >  45nm		Core2 Processors

-- 
Jean Delvare

_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors

Re: [lm-sensors] [PATCH] hwmon: (coretemp) Document Atom E660

[R, Durgadoss]

Hi Jean,

I have the CPU model of the next generation Atom based Core.
Can we add this also in the list here ?
If you don't find time, I can add the one liner, after this gets in.

)
> +                              0x1a (Nehalem), 0x1c (Atom), 0x1e (Lynnfield),
> +                              0x26 (Tunnel Creek Atom)

0x27 (Medfield Atom) and TjMax 90

>      Datasheet: Intel 64 and IA-32 Architectures Software Developer's Manual
>                 Volume 3A: System Programming Guide
>                 http://softwarecommunity.intel.com/Wiki/Mobility/720.htm
> @@ -85,6 +86,8 @@ Process		Processor
> 	TjMax(C)
>  		N475/470/455/450				100
>  		N280/270					90
>  		330/230						125
> +		E660						90
> +		E660T						110
> 
>  45nm		Core2 Processors

Thanks,
Durga

_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors

Re: [lm-sensors] [PATCH] hwmon: (coretemp) Document Atom E660

[R, Durgadoss]

Hi Jean,

> -----Original Message-----
> From: Jean Delvare [mailto:khali@linux-fr.org]
> Sent: Wednesday, September 28, 2011 6:48 PM
> To: R, Durgadoss
> Cc: LM Sensors; Yu, Fenghua; Alexander Stein
> Subject: Re: [lm-sensors] [PATCH] hwmon: (coretemp) Document Atom E660 support
> 
> Hi Durga,
> 
> On Wed, 28 Sep 2011 18:36:50 +0530, R, Durgadoss wrote:
> > I have the CPU model of the next generation Atom based Core.
> > Can we add this also in the list here ?
> > If you don't find time, I can add the one liner, after this gets in.
> >
> > )
> > > +                              0x1a (Nehalem), 0x1c (Atom), 0x1e
> (Lynnfield),
> > > +                              0x26 (Tunnel Creek Atom)
> >
> > 0x27 (Medfield Atom) and TjMax 90
> 
> I can add it, no problem. Is TjMax the same for all Medfield-based Atom
> CPUs? If so, and if it can't be read from MSR_TEMPERATURE_TARGET, then
> this value shouldn't go in the documentation file but in the driver,
> unless the fragile heuristic we have already comes to that conclusion
> (it's so complex that honestly I can't tell.)

Thanks Jean.
Yes. It's the same for all Medfield based Atom products.
And it cannot be read from MSR_TEMPERATURE_TARGET.
So, we have to add the model name in the driver.
Currently I have added a OR inside the 'if' in adjust_tjmax Function.

if (c->x86_model = 0x1c || c->x86_model = 0x27) {

This fetches me 90 and I have confirmed it with the Hardware folks here.

Thanks,
Durga

_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors

Re: [lm-sensors] [PATCH] hwmon: (coretemp) Document Atom E660

[Jean Delvare]

On Wed, 28 Sep 2011 18:54:58 +0530, R, Durgadoss wrote:
> > > 0x27 (Medfield Atom) and TjMax 90
> > 
> > I can add it, no problem. Is TjMax the same for all Medfield-based Atom
> > CPUs? If so, and if it can't be read from MSR_TEMPERATURE_TARGET, then
> > this value shouldn't go in the documentation file but in the driver,
> > unless the fragile heuristic we have already comes to that conclusion
> > (it's so complex that honestly I can't tell.)
> 
> Thanks Jean.
> Yes. It's the same for all Medfield based Atom products.
> And it cannot be read from MSR_TEMPERATURE_TARGET.
> So, we have to add the model name in the driver.
> Currently I have added a OR inside the 'if' in adjust_tjmax Function.
> 
> if (c->x86_model = 0x1c || c->x86_model = 0x27) {
> 
> This fetches me 90 and I have confirmed it with the Hardware folks here.

OK, I'll do this. What about model 0x26? Wouldn't 90°C be a better
default for them too?

And please tell your hardware friends to implement
MSR_TEMPERATURE_TARGET in future Atom product lines, it makes our life
so much easier...

-- 
Jean Delvare

_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors

Re: [lm-sensors] [PATCH] hwmon: (coretemp) Document Atom E660

[Jean Delvare]

On Wed, 28 Sep 2011 18:54:58 +0530, R, Durgadoss wrote:
> Yes. It's the same for all Medfield based Atom products.
> And it cannot be read from MSR_TEMPERATURE_TARGET.
> So, we have to add the model name in the driver.
> Currently I have added a OR inside the 'if' in adjust_tjmax Function.
> 
> if (c->x86_model = 0x1c || c->x86_model = 0x27) {
> 
> This fetches me 90 and I have confirmed it with the Hardware folks here.

Oh, BTW, there is magic in this block to adjust tjmax based on the
north bridge. Is there any such magic possible for Atom model 0x26?

-- 
Jean Delvare

_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors

Re: [lm-sensors] [PATCH] hwmon: (coretemp) Document Atom E660

[R, Durgadoss]

Hi Jean,

> >
> > if (c->x86_model = 0x1c || c->x86_model = 0x27) {
> >
> > This fetches me 90 and I have confirmed it with the Hardware folks here.
> 
> OK, I'll do this. What about model 0x26? Wouldn't 90°C be a better
> default for them too?

The TjMax is 90 for 0x26 also. I have tested this on my old 0x26 core.
But, we have to add one more || inside the 'if' :-(
Also, we have to add these cpu models inside the switch-case inside
get_tjmax, in order for the flow to reach adjust_tjmax.

> 
> And please tell your hardware friends to implement
> MSR_TEMPERATURE_TARGET in future Atom product lines, it makes our life
> so much easier...

I will try my best :-)

Thanks,
Durga

_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors

Re: [lm-sensors] [PATCH] hwmon: (coretemp) Document Atom E660

[R, Durgadoss]

Hi Jean,

> -----Original Message-----
> From: Jean Delvare [mailto:khali@linux-fr.org]
> Sent: Wednesday, September 28, 2011 7:20 PM
> To: R, Durgadoss
> Cc: LM Sensors; Yu, Fenghua; Alexander Stein
> Subject: Re: [lm-sensors] [PATCH] hwmon: (coretemp) Document Atom E660 support
> 
> On Wed, 28 Sep 2011 18:54:58 +0530, R, Durgadoss wrote:
> > Yes. It's the same for all Medfield based Atom products.
> > And it cannot be read from MSR_TEMPERATURE_TARGET.
> > So, we have to add the model name in the driver.
> > Currently I have added a OR inside the 'if' in adjust_tjmax Function.
> >
> > if (c->x86_model = 0x1c || c->x86_model = 0x27) {
> >
> > This fetches me 90 and I have confirmed it with the Hardware folks here.
> 
> Oh, BTW, there is magic in this block to adjust tjmax based on the
> north bridge. Is there any such magic possible for Atom model 0x26?

I guess this is not needed now, since we know the tjmax is 90 for 0x26.

Thanks,
Durga

_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors

Re: [lm-sensors] [PATCH] hwmon: (coretemp) Document Atom E660

[Jean Delvare]

On Wed, 28 Sep 2011 19:27:37 +0530, R, Durgadoss wrote:
> Hi Jean,
> 
> > -----Original Message-----
> > From: Jean Delvare [mailto:khali@linux-fr.org]
> > Sent: Wednesday, September 28, 2011 7:20 PM
> > To: R, Durgadoss
> > Cc: LM Sensors; Yu, Fenghua; Alexander Stein
> > Subject: Re: [lm-sensors] [PATCH] hwmon: (coretemp) Document Atom E660 support
> > 
> > On Wed, 28 Sep 2011 18:54:58 +0530, R, Durgadoss wrote:
> > > Yes. It's the same for all Medfield based Atom products.
> > > And it cannot be read from MSR_TEMPERATURE_TARGET.
> > > So, we have to add the model name in the driver.
> > > Currently I have added a OR inside the 'if' in adjust_tjmax Function.
> > >
> > > if (c->x86_model = 0x1c || c->x86_model = 0x27) {
> > >
> > > This fetches me 90 and I have confirmed it with the Hardware folks here..
> > 
> > Oh, BTW, there is magic in this block to adjust tjmax based on the
> > north bridge. Is there any such magic possible for Atom model 0x26?
> 
> I guess this is not needed now, since we know the tjmax is 90 for 0x26.

Not always. As discussed earlier, the E660T has TjMax = 110°C. If this
was the only exception, it's documented now so we'd me mostly fine. But
apparently all "T" variants of the Atom E6x0 have TjMax = 110°C. It
would be really nice to have a way to set tjmax = 110000 in the driver
for them, even if this is only a heuristic.

-- 
Jean Delvare

_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors

Re: [lm-sensors] [PATCH] hwmon: (coretemp) Document Atom E660

[Guenter Roeck]

Hi Jean,

On Wed, 2011-09-28 at 08:30 -0400, Jean Delvare wrote:
> Document the new Atom series (Tunnel Creek) as being supported, and
> list TjMax for the Atom E660 and E660T.
> 
> Signed-off-by: Jean Delvare <khali@linux-fr.org>
> Cc: Alexander Stein <alexander.stein@systec-electronic.com>
> Cc: Fenghua Yu <fenghua.yu@intel.com>
> Cc: Guenter Roeck <guenter.roeck@ericsson.com>
> ---
> As reported by Alexander weeks ago.
> 
>  Documentation/hwmon/coretemp |    5 ++++-
>  1 file changed, 4 insertions(+), 1 deletion(-)
> 
> --- linux-3.1-rc4.orig/Documentation/hwmon/coretemp	2011-09-28 12:45:58.000000000 +0200
> +++ linux-3.1-rc4/Documentation/hwmon/coretemp	2011-09-28 14:28:24.000000000 +0200
> @@ -6,7 +6,8 @@ Supported chips:
>      Prefix: 'coretemp'
>      CPUID: family 0x6, models 0xe (Pentium M DC), 0xf (Core 2 DC 65nm),
>                                0x16 (Core 2 SC 65nm), 0x17 (Penryn 45nm),
> -                              0x1a (Nehalem), 0x1c (Atom), 0x1e (Lynnfield)
> +                              0x1a (Nehalem), 0x1c (Atom), 0x1e (Lynnfield),
> +                              0x26 (Tunnel Creek Atom)
>      Datasheet: Intel 64 and IA-32 Architectures Software Developer's Manual
>                 Volume 3A: System Programming Guide
>                 http://softwarecommunity.intel.com/Wiki/Mobility/720.htm
> @@ -85,6 +86,8 @@ Process		Processor					TjMax(C)
>  		N475/470/455/450				100
>  		N280/270					90
>  		330/230						125
> +		E660						90
> +		E660T						110
>  
>  45nm		Core2 Processors
>  		Solo ULV SU3500/3300				100
> 

I got lost somewhere in the exchange. Should I apply this one, or are
you going to send out a new version ?

Thanks,
Guenter



_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors

Re: [lm-sensors] [PATCH] hwmon: (coretemp) Document Atom E660

[Jean Delvare]

On Wed, 28 Sep 2011 08:11:29 -0700, Guenter Roeck wrote:
> Hi Jean,
> 
> On Wed, 2011-09-28 at 08:30 -0400, Jean Delvare wrote:
> > Document the new Atom series (Tunnel Creek) as being supported, and
> > list TjMax for the Atom E660 and E660T.
> > 
> > Signed-off-by: Jean Delvare <khali@linux-fr.org>
> > Cc: Alexander Stein <alexander.stein@systec-electronic.com>
> > Cc: Fenghua Yu <fenghua.yu@intel.com>
> > Cc: Guenter Roeck <guenter.roeck@ericsson.com>
> > ---
> > As reported by Alexander weeks ago.
> > 
> >  Documentation/hwmon/coretemp |    5 ++++-
> >  1 file changed, 4 insertions(+), 1 deletion(-)
> > 
> > --- linux-3.1-rc4.orig/Documentation/hwmon/coretemp	2011-09-28 12:45:58.000000000 +0200
> > +++ linux-3.1-rc4/Documentation/hwmon/coretemp	2011-09-28 14:28:24.000000000 +0200
> > @@ -6,7 +6,8 @@ Supported chips:
> >      Prefix: 'coretemp'
> >      CPUID: family 0x6, models 0xe (Pentium M DC), 0xf (Core 2 DC 65nm),
> >                                0x16 (Core 2 SC 65nm), 0x17 (Penryn 45nm),
> > -                              0x1a (Nehalem), 0x1c (Atom), 0x1e (Lynnfield)
> > +                              0x1a (Nehalem), 0x1c (Atom), 0x1e (Lynnfield),
> > +                              0x26 (Tunnel Creek Atom)
> >      Datasheet: Intel 64 and IA-32 Architectures Software Developer's Manual
> >                 Volume 3A: System Programming Guide
> >                 http://softwarecommunity.intel.com/Wiki/Mobility/720.htm
> > @@ -85,6 +86,8 @@ Process		Processor					TjMax(C)
> >  		N475/470/455/450				100
> >  		N280/270					90
> >  		330/230						125
> > +		E660						90
> > +		E660T						110
> >  
> >  45nm		Core2 Processors
> >  		Solo ULV SU3500/3300				100
> > 
> 
> I got lost somewhere in the exchange. Should I apply this one, or are
> you going to send out a new version ?

Please wait, I'll resubmit once the dust settles down.

Thanks,
-- 
Jean Delvare

_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors