* speedstep-ich issue
@ 2006-09-29 11:11 Marco cusmai
2006-09-29 17:40 ` Mattia Dongili
0 siblings, 1 reply; 6+ messages in thread
From: Marco cusmai @ 2006-09-29 11:11 UTC (permalink / raw)
To: cpufreq
Upgrading kernel form version 2.6.17.2 to 2.6.18 the cpufreq system makes
strange things.. looking at dmesg (with debugging enabled for cpufreq) it
seems that after a while the system recognizes only one step of frequency
(the slowest) .. The CPU is a P3 that can change the frequency from 730 MHz
to 1.1 GHz, using the speedstep-ich driver. Tell me if you need more
informations..
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: speedstep-ich issue
2006-09-29 11:11 speedstep-ich issue Marco cusmai
@ 2006-09-29 17:40 ` Mattia Dongili
2006-09-29 21:49 ` Marco cusmai
0 siblings, 1 reply; 6+ messages in thread
From: Mattia Dongili @ 2006-09-29 17:40 UTC (permalink / raw)
To: Marco cusmai; +Cc: cpufreq
On Fri, Sep 29, 2006 at 11:11:48AM +0000, Marco cusmai wrote:
> Upgrading kernel form version 2.6.17.2 to 2.6.18 the cpufreq system makes
> strange things.. looking at dmesg (with debugging enabled for cpufreq) it
> seems that after a while the system recognizes only one step of frequency
> (the slowest) .. The CPU is a P3 that can change the frequency from 730 MHz
> to 1.1 GHz, using the speedstep-ich driver. Tell me if you need more
> informations..
probably dmesg output and some kind of evidence of the problem you're
reporting would be appreciated.
Regards
--
mattia
:wq!
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: speedstep-ich issue
2006-09-29 17:40 ` Mattia Dongili
@ 2006-09-29 21:49 ` Marco cusmai
2006-10-02 2:13 ` Dominik Brodowski
0 siblings, 1 reply; 6+ messages in thread
From: Marco cusmai @ 2006-09-29 21:49 UTC (permalink / raw)
To: cpufreq
On Friday 29 September 2006 17:40, Mattia Dongili wrote:
> On Fri, Sep 29, 2006 at 11:11:48AM +0000, Marco cusmai wrote:
> > Upgrading kernel form version 2.6.17.2 to 2.6.18 the cpufreq system makes
> > strange things.. looking at dmesg (with debugging enabled for cpufreq)
> > it seems that after a while the system recognizes only one step of
> > frequency (the slowest) .. The CPU is a P3 that can change the frequency
> > from 730 MHz to 1.1 GHz, using the speedstep-ich driver. Tell me if you
> > need more informations..
>
> probably dmesg output and some kind of evidence of the problem you're
> reporting would be appreciated.
>
> Regards
Here it is some dmesg output with cpufreq.debug=2
speedstep-lib: x86: 6, model: b
speedstep-lib: ebx is 6
speedstep-ich: pmbase is 0x1000
speedstep-lib: trying to determine both speeds
speedstep-lib: P3 - MSR_IA32_EBL_CR_POWERON: 0xcd840020 0x0
speedstep-lib: speed is 1130500
speedstep-lib: previous speed is 1130500
speedstep-ich: read at pmbase 0x1000 + 0x50 returned 0x0
speedstep-ich: writing 0x1 to pmbase 0x1000 + 0x50
speedstep-ich: read at pmbase 0x1000 + 0x50 returned 0x1
speedstep-lib: P3 - MSR_IA32_EBL_CR_POWERON: 0xc5040020 0x0
speedstep-lib: speed is 731500
speedstep-ich: change to 731 MHz succeeded
speedstep-lib: P3 - MSR_IA32_EBL_CR_POWERON: 0xc5040020 0x0
speedstep-lib: speed is 731500
speedstep-lib: low speed is 731500
speedstep-ich: read at pmbase 0x1000 + 0x50 returned 0x1
speedstep-ich: writing 0x0 to pmbase 0x1000 + 0x50
speedstep-ich: read at pmbase 0x1000 + 0x50 returned 0x0
speedstep-lib: P3 - MSR_IA32_EBL_CR_POWERON: 0xcd840020 0x0
speedstep-lib: speed is 1130500
speedstep-ich: change to 1130 MHz succeeded
speedstep-lib: P3 - MSR_IA32_EBL_CR_POWERON: 0xcd840020 0x0
speedstep-lib: speed is 1130500
speedstep-lib: high speed is 1130500
speedstep-lib: transition latency is 0 uSec
speedstep: frequency transition measured seems out of range (0 nSec), falling
back to a safe one of 500000 nSec.
speedstep-lib: P3 - MSR_IA32_EBL_CR_POWERON: 0xcd840020 0x0
speedstep-lib: speed is 1130500
speedstep-ich: detected 1130500 kHz as current frequency
speedstep-ich: currently at high speed setting - 1130 MHz
speedstep-lib: P3 - MSR_IA32_EBL_CR_POWERON: 0xcd840020 0x0
speedstep-lib: speed is 1130500
speedstep-ich: detected 1130500 kHz as current frequency
speedstep-ich: transiting from 1130500 to 1130500 kHz
Here starts the userspace program (cpufreqd):
speedstep-lib: P3 - MSR_IA32_EBL_CR_POWERON: 0xcd840020 0x0
speedstep-lib: speed is 1130500
speedstep-ich: detected 1130500 kHz as current frequency
speedstep-ich: transiting from 1130500 to 1130500 kHz
speedstep-lib: P3 - MSR_IA32_EBL_CR_POWERON: 0xcd840020 0x0
speedstep-lib: speed is 1130500
speedstep-ich: detected 1130500 kHz as current frequency
speedstep-ich: transiting from 1130500 to 731500 kHz
speedstep-ich: read at pmbase 0x1000 + 0x50 returned 0x0
speedstep-ich: writing 0x1 to pmbase 0x1000 + 0x50
speedstep-ich: read at pmbase 0x1000 + 0x50 returned 0x1
speedstep-lib: P3 - MSR_IA32_EBL_CR_POWERON: 0xc5040020 0x0
speedstep-lib: speed is 731500
speedstep-ich: change to 731 MHz succeeded
speedstep-lib: P3 - MSR_IA32_EBL_CR_POWERON: 0xc5040020 0x0
speedstep-lib: speed is 731500
speedstep-ich: detected 731500 kHz as current frequency
speedstep-ich: transiting from 731500 to 731500 kHz
speedstep-lib: P3 - MSR_IA32_EBL_CR_POWERON: 0xc5040020 0x0
speedstep-lib: speed is 731500
speedstep-ich: detected 731500 kHz as current frequency
speedstep-ich: transiting from 731500 to 731500 kHz
from now on the frequency stays on 731500 kHz..
I've tried to read informations from the sys interface:
cat scaling_available_frequencies : 1130500 731500
cat scaling_min_freq : 731500
cat scaling_max_freq : 731500
trying to change frequency through scaling_setspeed does nothing.
I've also noticed these errors during ACPI startup
acpiphp: ACPI Hot Plug PCI Controller Driver version: 0.5
decode_hpp: Could not get hotplug parameters. Use defaults
acpiphp: Slot [1] registered
acpiphp_ibm: ibm_find_acpi_device: Failed to get device
information<3>acpiphp_ibm: ibm_find_acpi_device: Failed to get device
information<3>acpiphp_ibm: ibm_acpiphp_init: acpi_walk_namespace failed
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: speedstep-ich issue
2006-09-29 21:49 ` Marco cusmai
@ 2006-10-02 2:13 ` Dominik Brodowski
2006-10-02 8:14 ` Marco Cusmai
0 siblings, 1 reply; 6+ messages in thread
From: Dominik Brodowski @ 2006-10-02 2:13 UTC (permalink / raw)
To: Marco cusmai; +Cc: cpufreq
Hi,
On Fri, Sep 29, 2006 at 09:49:42PM +0000, Marco cusmai wrote:
> cat scaling_available_frequencies : 1130500 731500
> cat scaling_min_freq : 731500
> cat scaling_max_freq : 731500
Yes, then it won't scale any more -- the CPU will only be set to a value
between scaling_min_freq and scaling_max_freq. Therefore, please check the
configuration file for cpufreqd (or any other cpufreq daemon you're
running).
Thanks,
Dominik
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: speedstep-ich issue
2006-10-02 2:13 ` Dominik Brodowski
@ 2006-10-02 8:14 ` Marco Cusmai
2006-10-02 8:21 ` Mattia Dongili
0 siblings, 1 reply; 6+ messages in thread
From: Marco Cusmai @ 2006-10-02 8:14 UTC (permalink / raw)
To: Marco cusmai, cpufreq
On 10/2/06, Dominik Brodowski <linux at dominikbrodowski.net> wrote:
>
> Hi,
>
> On Fri, Sep 29, 2006 at 09:49:42PM +0000, Marco cusmai wrote:
> > cat scaling_available_frequencies : 1130500 731500
> > cat scaling_min_freq : 731500
> > cat scaling_max_freq : 731500
>
> Yes, then it won't scale any more -- the CPU will only be set to a value
> between scaling_min_freq and scaling_max_freq. Therefore, please check the
> configuration file for cpufreqd (or any other cpufreq daemon you're
> running).
>
> Thanks,
> Dominik
>
For my tests I've already disabled the cpufreqd deamon at startup.. The
results are the same.
I've also tried the following:
echo 1130500 > scaling_setspeed
cat scaling_cur_freq : 731500
The governor I actualy use is "userspace". Even with "performance" the freq
stays low
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: speedstep-ich issue
2006-10-02 8:14 ` Marco Cusmai
@ 2006-10-02 8:21 ` Mattia Dongili
0 siblings, 0 replies; 6+ messages in thread
From: Mattia Dongili @ 2006-10-02 8:21 UTC (permalink / raw)
To: Marco Cusmai; +Cc: cpufreq
On Mon, October 2, 2006 10:14 am, Marco Cusmai said:
> On 10/2/06, Dominik Brodowski <linux at dominikbrodowski.net> wrote:
>>
>> Hi,
>>
>> On Fri, Sep 29, 2006 at 09:49:42PM +0000, Marco cusmai wrote:
>> > cat scaling_available_frequencies : 1130500 731500
>> > cat scaling_min_freq : 731500
>> > cat scaling_max_freq : 731500
>>
>> Yes, then it won't scale any more -- the CPU will only be set to a value
>> between scaling_min_freq and scaling_max_freq. Therefore, please check
>> the
>> configuration file for cpufreqd (or any other cpufreq daemon you're
>> running).
>>
>> Thanks,
>> Dominik
>>
> For my tests I've already disabled the cpufreqd deamon at startup.. The
> results are the same.
> I've also tried the following:
>
> echo 1130500 > scaling_setspeed
> cat scaling_cur_freq : 731500
try this before:
$ echo 1130500 > scaling_max_freq
As explained, scaling_max_freq and scaling_min_freq define the upper and
lower limit to which the processor is bound (this is a software enforced
limit), so if this limit is MAX=MIN=731500 then the kernel won't allow
1130500.
--
mattia
:wq!
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2006-10-02 8:21 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2006-09-29 11:11 speedstep-ich issue Marco cusmai
2006-09-29 17:40 ` Mattia Dongili
2006-09-29 21:49 ` Marco cusmai
2006-10-02 2:13 ` Dominik Brodowski
2006-10-02 8:14 ` Marco Cusmai
2006-10-02 8:21 ` Mattia Dongili
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.