headers
Lutz Vieweg | 1 Mar 2004 11:12
Picon
Favicon

Re: [patches] 2.6.2 to 2.6.3 regression: powernow-k8 doesn't work anymore

Pavel Machek wrote:

>>A factor 1000 lost somewhere?
> 
> Yes.
> 
> Try this one. [Warning: extremely ugly hand-edited patch follows;
> taking akpm off Cc list, I do not want him to even see it]

This second patch (applied addtionally to the first one) fixes
the cpufreq behaviour for the K8V board - thanks!

Regards,

Lutz Vieweg

> 
> --- clean/arch/i386/kernel/cpu/cpufreq/powernow-k8.c	2004-02-20 12:29:10.000000000 +0100
> +++ linux/arch/i386/kernel/cpu/cpufreq/powernow-k8.c	2004-02-23 21:06:01.000000000 +0100
>  <at>  <at>  -8,6 +8,8  <at>  <at> 
>   *
>   *  Based on the powernow-k7.c module written by Dave Jones.
>   *  (C) 2003 Dave Jones <davej <at> codemonkey.ork.uk> on behalf of SuSE Labs
> + *  (C) 2004 Dominik Brodowski <linux <at> brodo.de>
> + *  (C) 2004 Pavel Machek <pavel <at> suse.cz>
>   *  Licensed under the terms of the GNU GPL License version 2.
>   *  Based upon datasheets & sample CPUs kindly provided by AMD.
>   *
>  <at>  <at>  -34,10 +36,6  <at>  <at> 
>  #define VERSION "version 1.00.08a"
(Continue reading)

Permalink | Reply | 0 comments |
headers
Lutz Vieweg | 1 Mar 2004 11:15
Picon
Favicon

Re: [patches] 2.6.2 to 2.6.3 regression: powernow-k8 doesn't work anymore

Pavel Machek wrote:

>>>A factor 1000 lost somewhere?
>>
>>Yes.
> 
> 
> Better patch follows.

Ah, I just realized I replied to the wrong mail - of course I just
tried this "better patch" :-) and confirmed that it works with the K8V board.

Regards,

Lutz Vieweg

> Index: arch/i386/kernel/cpu/cpufreq/powernow-k8.c
> ===================================================================
> RCS file: /home/pavel/sf/bitbucket/bkcvs/linux-2.5/arch/i386/kernel/cpu/cpufreq/powernow-k8.c,v
> retrieving revision 1.6
> diff -u -u -r1.6 powernow-k8.c
> --- arch/i386/kernel/cpu/cpufreq/powernow-k8.c	23 Feb 2004 07:07:45 -0000	1.6
> +++ arch/i386/kernel/cpu/cpufreq/powernow-k8.c	28 Feb 2004 13:02:27 -0000
>  <at>  <at>  -8,6 +8,8  <at>  <at> 
>   *
>   *  Based on the powernow-k7.c module written by Dave Jones.
>   *  (C) 2003 Dave Jones <davej <at> codemonkey.ork.uk> on behalf of SuSE Labs
> + *  (C) 2004 Dominik Brodowski <linux <at> brodo.de>
> + *  (C) 2004 Pavel Machek <pavel <at> suse.cz>
>   *  Licensed under the terms of the GNU GPL License version 2.
(Continue reading)

Permalink | Reply | 0 comments |
headers
Bruno Ducrot | 1 Mar 2004 14:00

Re: VIA C3 M10k Nehemiah longhaul support in 2.6.x ?

On Fri, Feb 27, 2004 at 08:49:46PM -0500, Doug Sutherland wrote:
> François wrote:
> 
> > [Nehemiah longhaul] Not seems to be supported still in actual
> > 2.6.3-bk9 version, although it was working on 2.4.x
> 
> Are you sure Nehemiah was supported for 2.4 kernels?
> I just patched a 2.4.25 kernel with cpufreq.
> I have a Nehemiah model 9 stepping 4 processor.

longhaul support under 2.4 should be the exact same than the 2.6 version
from the cpufreq bk tree made by Dave.  No more, no less.  There may be
a miss sometimes though, depending when the backport is made, and I must
confess I don't have looked recently on the most recent bits from Dave
for the longhaul driver.

> http://www3.sympatico.ca/howlettfamily/epia/longhaul-1.2.tar.gz
> See this in the viaarena forums
> http://forums.viaarena.com/messageview.cfm?catid=4&threadid=52209

I will look on this.

--

-- 
Bruno Ducrot

--  Which is worse:  ignorance or apathy?
--  Don't know.  Don't care.
Permalink | Reply | 2 comments |
headers
Bruno Ducrot | 1 Mar 2004 15:07

Re: VIA C3 M10k Nehemiah longhaul support in 2.6.x ?

On Mon, Mar 01, 2004 at 02:00:19PM +0100, Bruno Ducrot wrote:
> On Fri, Feb 27, 2004 at 08:49:46PM -0500, Doug Sutherland wrote:
> > François wrote:
> > 
> > > [Nehemiah longhaul] Not seems to be supported still in actual
> > > 2.6.3-bk9 version, although it was working on 2.4.x
> > 
> > Are you sure Nehemiah was supported for 2.4 kernels?
> > I just patched a 2.4.25 kernel with cpufreq.
> > I have a Nehemiah model 9 stepping 4 processor.
> 
> longhaul support under 2.4 should be the exact same than the 2.6 version
> from the cpufreq bk tree made by Dave.  No more, no less.  There may be
> a miss sometimes though, depending when the backport is made, and I must
> confess I don't have looked recently on the most recent bits from Dave
> for the longhaul driver.
> 
> > http://www3.sympatico.ca/howlettfamily/epia/longhaul-1.2.tar.gz
> > See this in the viaarena forums
> > http://forums.viaarena.com/messageview.cfm?catid=4&threadid=52209
> 

Ok, I see some changes on some tables.  I'm not sure though what is
correct (I need to get a least specs, if there are publicy available).

The 2.4 version should be OK today or in worst case tomorrow.

Thanks,

--

--
(Continue reading)

Permalink | Reply | 1 comment |
headers
Ducrot Bruno | 1 Mar 2004 16:11
Picon

cpufreq/linux/arch/i386/kernel/cpufreq longhaul.c, 1.93.2.4, 1.93.2.5

Update of /mnt/src/cvsroot/cpufreq/linux/arch/i386/kernel/cpufreq
In directory flint:/tmp/cvs-serv30352

Modified Files:
      Tag: LINUX_2_4
	longhaul.c 
Log Message:
update to latest longhaul driver from cpufreq bk tree.

Index: longhaul.c
===================================================================
RCS file: /mnt/src/cvsroot/cpufreq/linux/arch/i386/kernel/cpufreq/longhaul.c,v
retrieving revision 1.93.2.4
retrieving revision 1.93.2.5
diff -u -r1.93.2.4 -r1.93.2.5
--- longhaul.c	9 Dec 2003 09:07:30 -0000	1.93.2.4
+++ longhaul.c	1 Mar 2004 15:11:25 -0000	1.93.2.5
 <at>  <at>  -1,5 +1,5  <at>  <at> 
 /*
- *  (C) 2001-2003  Dave Jones. <davej <at> codemonkey.org.uk>
+ *  (C) 2001-2004  Dave Jones. <davej <at> codemonkey.org.uk>
  *  (C) 2002  Padraig Brady. <padraig <at> antefacto.com>
  *
  *  Licensed under the terms of the GNU GPL License version 2.
 <at>  <at>  -63,11 +63,12  <at>  <at> 

 static unsigned int calc_speed (int mult, int fsb)
 {
-	int mhz;
-	mhz = (mult/10)*fsb;
(Continue reading)

Permalink | Reply | 0 comments |
headers
Ducrot Bruno | 1 Mar 2004 16:13
Picon

cpufreq/linux/arch/i386/kernel/cpufreq longhaul.c, 1.93.2.5, 1.93.2.6

Update of /mnt/src/cvsroot/cpufreq/linux/arch/i386/kernel/cpufreq
In directory flint:/tmp/cvs-serv30382

Modified Files:
      Tag: LINUX_2_4
	longhaul.c 
Log Message:
Try to enable Nehemiah.

Index: longhaul.c
===================================================================
RCS file: /mnt/src/cvsroot/cpufreq/linux/arch/i386/kernel/cpufreq/longhaul.c,v
retrieving revision 1.93.2.5
retrieving revision 1.93.2.6
diff -u -r1.93.2.5 -r1.93.2.6
--- longhaul.c	1 Mar 2004 15:11:25 -0000	1.93.2.5
+++ longhaul.c	1 Mar 2004 15:13:14 -0000	1.93.2.6
 <at>  <at>  -467,12 +467,8  <at>  <at> 
 		return -ENODEV;

 	switch (c->x86_model) {
-	case 6 ... 8:
+	case 6 ... 9:
 		return cpufreq_register_driver(&longhaul_driver);
-	case 9:
-		printk (KERN_INFO PFX "Nehemiah unsupported: Waiting on working silicon "
-						"from VIA before this is usable.\n");
-		break;
 	default:
 		printk (KERN_INFO PFX "Unknown VIA CPU. Contact davej <at> codemonkey.org.uk\n");
(Continue reading)

Permalink | Reply | 0 comments |
headers
Bruno Ducrot | 1 Mar 2004 16:16

Re: VIA C3 M10k Nehemiah longhaul support in 2.6.x ?

> The 2.4 version should be OK today or in worst case tomorrow.
> 

I just commited that change, but only for 2.4.  Could you report if that
is ok, please ?

Index: longhaul.c
===================================================================
RCS file: /mnt/src/cvsroot/cpufreq/linux/arch/i386/kernel/cpufreq/longhaul.c,v
retrieving revision 1.93.2.3
retrieving revision 1.93.2.6
diff -u -r1.93.2.3 -r1.93.2.6
--- longhaul.c	9 Dec 2003 08:58:33 -0000	1.93.2.3
+++ longhaul.c	1 Mar 2004 15:13:14 -0000	1.93.2.6
 <at>  <at>  -1,5 +1,5  <at>  <at> 
 /*
- *  (C) 2001-2003  Dave Jones. <davej <at> codemonkey.org.uk>
+ *  (C) 2001-2004  Dave Jones. <davej <at> codemonkey.org.uk>
  *  (C) 2002  Padraig Brady. <padraig <at> antefacto.com>
  *
  *  Licensed under the terms of the GNU GPL License version 2.
 <at>  <at>  -63,11 +63,12  <at>  <at> 

 static unsigned int calc_speed (int mult, int fsb)
 {
-	int mhz;
-	mhz = (mult/10)*fsb;
+	int khz;
+	khz = (mult/10)*fsb;
 	if (mult%10)
(Continue reading)

Permalink | Reply | 0 comments |
headers
Bruno Ducrot | 1 Mar 2004 16:31

Re: Pentium III Coppermine and SiS 630 southbridge

On Fri, Feb 27, 2004 at 12:20:04PM +0100, Reynaldo Cordero wrote:
> Hi!
> I have a laptop with Pentium III Coppermine and SiS 630
> southbridge (clevo 2700s [1], kernel 2.6.3 on debian
> unstable).
> 
> I want to recompile it with CPU Frequency scaling,
> and I wondered if I can use:
> 
> - 'powersave' governor
> or
> - 'userspace' governor for userspace frequency scaling
> 
> And which of the drivers are the correct:
> 
> - Intel Speedstep on ICH-M chipsets (ioport interface) or
> - Intel SpeedStep on 440BX/ZX/MX chipsets (SMI interface)
> or anithing else.
> 
> [1] http://hemmerling.free.fr/html/en/linux01.html
> 

AFAIK, there is not even a start to get speedstep working on this
southbridge.  I'm sorry, but nobody don't have even a start of a clue to
get it.

Cheers,

--

-- 
Bruno Ducrot
(Continue reading)

Permalink | Reply | 0 comments |
headers
Bruno Ducrot | 1 Mar 2004 16:58

Re: supported x86 processors?

On Sun, Feb 29, 2004 at 09:03:30PM +0100, Franco wrote:
> Hello,
> 
> could you tell me which x86 processors are well
> supported by cpufreq, or are likely to be in a 
> near future?
> 
> I've already read Documentation/cpu-freq/user-guide.txt
> of my 2.6.3 kernel, but it doesn't say which are
> supported well and which aren't.
> 
> Sorry if this message is somehow a repetition,
> but I fear my previous message was a bit unclear,
> as I got no answers...
> 
> Thank You in advance,
> Franco
> 

For those I know the most, powernow are well supported (k6 and k8.
There are trouble to get for everyone documentations for the k7, at
least for me, but that one is now ok).
Speedstep are well supported also, but not on all hardware configurations.
If you have some VIA or SiS southbridges, this won't unlikely work.
On those systems, though, it may be possible to get it working with
the generic acpi io driver (not on all old PIII processors though, but
perhaps on recent PIIIs and P4m, no warranties, though).

For the others, I don't know but believe that all VIA C3 should be
OK (at least on near future).
(Continue reading)

Permalink | Reply | 4 comments |
headers
Franco | 1 Mar 2004 18:50
Picon

Re: supported x86 processors?

Thanks a lot for the answer.

Then, the candidates for my application are K7 and non-VIA, non-SiS
Speeddstep. (K6, K8, VIA C3 are too slow or too expensive).

> Note that some driver do not support voltage scaling
> You may take a look to arch/i386/kernel/cpu/cpufreq/ 

I took a look at the code.
As to K7, the only mention to voltage are a couple of printk().
As to P4, I'm not sure I've understood correctly, but seems me
that voltage scaling is only for mobile chips, is it right?

Then, as my application is a desktop PC, are K7 and P4 equivalent?
(relatively to voltage/frequency scaling)

Or, is there a chance that one of the two 
will also get voltage scaling in the future?

Thank you a lot in advance,
Franco
Permalink | Reply | 2 comments |

Gmane