headers
Benjamin Herrenschmidt | 1 Apr 2005 01:07

Re: [PATCH] Add suspend routine to core

On Wed, 2005-03-30 at 10:48 +1000, Benjamin Herrenschmidt wrote:
> Hi !
> 
> In order to properly fix some issues with cpufreq vs. sleep on
> PowerBooks, I had to add a suspend callback to the pmac_cpufreq driver.
> I must force a switch to full speed before sleep and I switch back to
> previous speed on resume.
> 
> I also added a driver flag to disable the warnings in suspend/resume
> since it is expected in this case to have different speed (and I want it
> to fixup the jiffies properly).
>
> .../...

No feedback at all ? Without any feedback next week, I'll submit myself
to andrew ;)

Ben.
Permalink | Reply | 1 comment |
headers
Sean Young | 3 Apr 2005 18:27

Re: [PATCH] AMD Elan SC520 cpufreq driver

On Wed, Mar 30, 2005 at 12:04:32AM +0200, Dominik Brodowski wrote:
> > +#include <linux/slab.h>
> 
> What do you need slab.h for?

-snip-

Thanks for the input. Here is a patch against the latest snapshot
(cpufreq-2005-04-03.diff) which fixes the issues.

Sean

diff -urpN linux-2.6.9/arch/i386/kernel/cpu/cpufreq/Kconfig /tmp/linux-2.6.9/arch/i386/kernel/cpu/cpufreq/Kconfig
--- linux-2.6.9/arch/i386/kernel/cpu/cpufreq/Kconfig	2005-04-03 18:21:00.652590000 +0200
+++ /tmp/linux-2.6.9/arch/i386/kernel/cpu/cpufreq/Kconfig	2005-04-03 17:52:41.949832072 +0200
 <at>  <at>  -23,7 +23,7  <at>  <at>  config X86_ACPI_CPUFREQ
 	  If in doubt, say N.

 config ELAN_CPUFREQ
-	tristate "AMD Elan SC500 and SC410"
+	tristate "AMD Elan SC400 and SC410"
 	select CPU_FREQ_TABLE
 	depends on X86_ELAN
 	---help---
diff -urpN linux-2.6.9/arch/i386/kernel/cpu/cpufreq/sc520_freq.c /tmp/linux-2.6.9/arch/i386/kernel/cpu/cpufreq/sc520_freq.c
--- linux-2.6.9/arch/i386/kernel/cpu/cpufreq/sc520_freq.c	2005-04-03 18:21:00.709581336 +0200
+++ /tmp/linux-2.6.9/arch/i386/kernel/cpu/cpufreq/sc520_freq.c	2005-04-03 17:52:58.789272088 +0200
 <at>  <at>  -17,7 +17,6  <at>  <at> 
 #include <linux/module.h>
 #include <linux/init.h>
(Continue reading)

Permalink | Reply | 3 comments |
headers
Eric Piel | 4 Apr 2005 01:53
Favicon

Re: [PATCH] AMD Elan SC520 cpufreq driver

Sean Young a écrit :
> -	policy->cpuinfo.transition_latency = CPUFREQ_ETERNAL;
> +	policy->cpuinfo.transition_latency = 1000; /* 1ms */

transition_latency is expressed in nanoseconds, so it should be 1000000.
I guess the confusion comes from cpufreq.h, where there are two comments 
about it, one saying it's nanosecond, another one (older) saying it's 
microseconds. Dominik, maybe this typo should be fixed, do you want me 
to send a patch?

Eric
Permalink | Reply | 2 comments |
headers
Ivor Hewitt | 4 Apr 2005 12:42
Favicon

starting governor ondemand failed

Hi,
I was trying to have a play with the ondemand governor but I just get 
the following logged when I try to switch to it:-

cpufreq-core: governor switch
cpufreq-core: __cpufreq_governor for CPU 0, event 2
userspace: managing cpu 0 stopped
cpufreq-core: __cpufreq_governor for CPU 0, event 1
cpufreq-core: starting governor ondemand failed
cpufreq-core: __cpufreq_governor for CPU 0, event 1

This is on a PIIIM using the speedstep-smi interface.
The cpu is only capable of switching between two frequencies, would this 
prevent ondemand starting up?

Regards,

--

-- 
Ivor
http://www.ivor.it
Permalink | Reply | 3 comments |
headers
Dominik Brodowski | 4 Apr 2005 14:43

Re: starting governor ondemand failed

On Mon, Apr 04, 2005 at 11:42:19AM +0100, Ivor Hewitt wrote:
> Hi,
> I was trying to have a play with the ondemand governor but I just get 
> the following logged when I try to switch to it:-
> 
> cpufreq-core: governor switch
> cpufreq-core: __cpufreq_governor for CPU 0, event 2
> userspace: managing cpu 0 stopped
> cpufreq-core: __cpufreq_governor for CPU 0, event 1
> cpufreq-core: starting governor ondemand failed
> cpufreq-core: __cpufreq_governor for CPU 0, event 1
> 
> This is on a PIIIM using the speedstep-smi interface.
> The cpu is only capable of switching between two frequencies, would this 
> prevent ondemand starting up?

This isn't the problem, but the following: we do not know how long 
transitions using speedstep-smi take, therefore we still need to assume
that it takes very long, thus making it unable to be used for ondemand
frequency switching.

	Dominik
Permalink | Reply | 2 comments |
headers
Ivor Hewitt | 4 Apr 2005 15:27
Favicon

Re: starting governor ondemand failed

Dominik Brodowski wrote:
> On Mon, Apr 04, 2005 at 11:42:19AM +0100, Ivor Hewitt wrote:
> 
>>Hi,
>>I was trying to have a play with the ondemand governor but I just get 
>>the following logged when I try to switch to it:-
>>
>>cpufreq-core: governor switch
>>cpufreq-core: __cpufreq_governor for CPU 0, event 2
>>userspace: managing cpu 0 stopped
>>cpufreq-core: __cpufreq_governor for CPU 0, event 1
>>cpufreq-core: starting governor ondemand failed
>>cpufreq-core: __cpufreq_governor for CPU 0, event 1
>>
>>This is on a PIIIM using the speedstep-smi interface.
>>The cpu is only capable of switching between two frequencies, would this 
>>prevent ondemand starting up?
> 
> This isn't the problem, but the following: we do not know how long 
> transitions using speedstep-smi take, therefore we still need to assume
> that it takes very long, thus making it unable to be used for ondemand
> frequency switching.
> 
Hi,
I'll stick to to cpufreqd then. :-)

Ah I see "CPUFREQ_ETERNAL".

What should be an acceptable transition time? Is it possible for me to 
determine my speedstep-smi transition time and enable the switching?
(Continue reading)

Permalink | Reply | 1 comment |
headers
Pallipadi, Venkatesh | 4 Apr 2005 15:39
Picon
Favicon

RE: starting governor ondemand failed


Dominik,

Can we do some calibration of the time taken to switch frequencies
at the time of initialization and use it here instead of hardcoding 
it to CPUFREQ_ETERNAL? 

Thanks,
Venki

>-----Original Message-----
>From: cpufreq-bounces <at> ZenII.linux.org.uk 
>[mailto:cpufreq-bounces <at> ZenII.linux.org.uk] On Behalf Of Ivor Hewitt
>Sent: Monday, April 04, 2005 6:27 AM
>To: Dominik Brodowski
>Cc: cpufreq <at> ZenII.linux.org.uk
>Subject: Re: starting governor ondemand failed
>
>Dominik Brodowski wrote:
>> On Mon, Apr 04, 2005 at 11:42:19AM +0100, Ivor Hewitt wrote:
>> 
>>>Hi,
>>>I was trying to have a play with the ondemand governor but I 
>just get 
>>>the following logged when I try to switch to it:-
>>>
>>>cpufreq-core: governor switch
>>>cpufreq-core: __cpufreq_governor for CPU 0, event 2
>>>userspace: managing cpu 0 stopped
>>>cpufreq-core: __cpufreq_governor for CPU 0, event 1
(Continue reading)

Permalink | Reply | 503 comments |
headers
Ivor Hewitt | 4 Apr 2005 15:50
Favicon

Re: starting governor ondemand failed

Ivor Hewitt wrote:
> Dominik Brodowski wrote:
> 
>> On Mon, Apr 04, 2005 at 11:42:19AM +0100, Ivor Hewitt wrote:
>>
>>> Hi,
>>> I was trying to have a play with the ondemand governor but I just get 
>>> the following logged when I try to switch to it:-
>>>
>>> cpufreq-core: governor switch
>>> cpufreq-core: __cpufreq_governor for CPU 0, event 2
>>> userspace: managing cpu 0 stopped
>>> cpufreq-core: __cpufreq_governor for CPU 0, event 1
>>> cpufreq-core: starting governor ondemand failed
>>> cpufreq-core: __cpufreq_governor for CPU 0, event 1
>>>
>>> This is on a PIIIM using the speedstep-smi interface.
>>> The cpu is only capable of switching between two frequencies, would 
>>> this prevent ondemand starting up?
>>
>>
>> This isn't the problem, but the following: we do not know how long 
>> transitions using speedstep-smi take, therefore we still need to assume
>> that it takes very long, thus making it unable to be used for ondemand
>> frequency switching.
>>
> Hi,
> I'll stick to to cpufreqd then. :-)
> 
> Ah I see "CPUFREQ_ETERNAL".
(Continue reading)

Permalink | Reply | 0 comments |
headers
Bruno Ducrot | 4 Apr 2005 18:58

Re: starting governor ondemand failed

On Mon, Apr 04, 2005 at 06:39:52AM -0700, Pallipadi, Venkatesh wrote:
> 
> 
> Dominik,
> 
> Can we do some calibration of the time taken to switch frequencies
> at the time of initialization and use it here instead of hardcoding 
> it to CPUFREQ_ETERNAL? 
> 

What we want is worst case...

--

-- 
Bruno Ducrot

--  Which is worse:  ignorance or apathy?
--  Don't know.  Don't care.
Permalink | Reply | 501 comments |
headers
Michał Pytasz | 5 Apr 2005 11:28
Picon

powernow-k8 on Athlon64 3200+ (s939) stability problem...

Hi,

A while ago I bought the mentioned Athlon, with MSI K8T Neo2 mainboard. It 
works fine (without powernow/cool&quiet or with "performance" govenor), but 
encouraged by a perfectly working laptop (Mobile Sempron 2800) I tried to use 
it with desktop (tried ondemand govenor as well as powernowd deamon). 
Well, it does work, but after switching to lowest (1000) frequency (not 
immediately) system hangs hard or reboots.
I have 2.6.11.6 kernel, gcc 3.4.3.20050110. System is running in native 64 bit 
mode:
Linux 2.6.11-gentoo-r5 #1 Sat Apr 2 21:52:05 CEST 2005 x86_64 AMD Athlon(tm) 
64 Processor 3200+ AuthenticAMD GNU/Linux

I was wondering how to check if it is a hardware problem (for example too low 
voltage from mainboard, deffective CPU,... ) or something in the system.

All ideas are welcome,
Michal
Permalink | Reply | 1 comment |

Gmane