Amd Turion TL-52 X2 on Amilo PA 1510> No way to change Frequency!

So, im sorry for my last mail...
don read your mail correctly.

but nowaday after reading many How-to's and tutorial's about compiling own kernel in ubuntu i managed my own kernel to work =)
so sorry again that i donot understand what you meant!

now my kernel got that module cpufreq debubugging within it! now i got that output from dmesg | grep power
:

[   56.286434] powernow-k8: Found 2 AMD Turion(tm) 64 X2 Mobile
Technology TL-52 processors (version 2.00.00)
[   56.286449] powernow-k8:    0 : fid 0x8, vid 0x13
[   56.286455] powernow-k8:    1 : fid 0x0, vid 0x1e
[   56.286459] powernow-k8:    0 : fid 0x8 (1600 MHz), vid 0x13
[   56.286462] powernow-k8:    1 : fid 0x0 (800 MHz), vid 0x1e
[   56.286467] powernow-k8: cpu0, init lo 0x1308, hi 0x1
[   56.286471] powernow-k8: policy current frequency 1600000 kHz
[   56.286475] powernow-k8: cpu_init done, current fid 0x8, vid 0x13
[   56.286520] powernow-k8: targ: cpu 0, 800000 kHz, min 800000, max
800000, relation 1
[   56.286524] powernow-k8: targ: curr fid 0x8, vid 0x13
[   56.286527] powernow-k8: cpu 0 transition to index 1
[   56.286530] powernow-k8: table matched fid 0x0, giving vid 0x1e
[   56.286533] powernow-k8: cpu 0, changing to fid 0x0, vid 0x1e
[   56.286539] powernow-k8: ph1 (cpu0): start, currfid 0x8, currvid
0x13, reqvid 0x1e, rvo 0x2
[   56.286543] powernow-k8: ph1 maxvid=0x11
[   56.286546] powernow-k8: ph1 complete, currfid 0x8, currvid 0x13
[   56.286549] powernow-k8: ph2 (cpu0): starting, currfid 0x8, currvid
0x13, reqfid 0x0
[   56.286554] powernow-k8: writing fid 0x0, lo 0x11300, hi 0x258
[   56.300255] powernow-k8: ph2 complete, currfid 0x0, currvid 0x13
[   56.300258] powernow-k8: ph3 (cpu0): starting, currfid 0x0, currvid
0x13
[   56.300262] powernow-k8: writing vid 0x1e, lo 0x11e00, hi 0x1
[   56.300265] powernow-k8: ph3 complete, currfid 0x0, currvid 0x1e
[   56.300269] powernow-k8: transitioned (cpu0): new fid 0x0, vid 0x1e
[   56.300275] powernow-k8: targ: cpu 0, 800000 kHz, min 800000, max
800000, relation 1
[   56.300278] powernow-k8: targ: curr fid 0x0, vid 0x1e
[   56.300281] powernow-k8: cpu 0 transition to index 1
[   56.300284] powernow-k8: table matched fid 0x0, giving vid 0x1e
[   56.300287] powernow-k8: target matches current values (fid 0x0, vid
0x1e)
[   59.119526] powernow-k8: targ: cpu 0, 800000 kHz, min 800000, max
800000, relation 1
[   59.119534] powernow-k8: targ: curr fid 0x0, vid 0x1e
[   59.119539] powernow-k8: cpu 0 transition to index 1
[   59.119542] powernow-k8: table matched fid 0x0, giving vid 0x1e
[   59.119545] powernow-k8: target matches current values (fid 0x0, vid

0x1e)

so far... looks like i can only get 800 MHz ...

The processor says it supports 1600 MHz, and even starts in that state. Then it transitions to 800 MHz and I never see another request for a higher frequency.

What are the values of the files in the
/sys/devices/system/cpu/cpu0/cpufreq
directory?

-Mark Langsdorf
Operating System Research Center
AMD

my cpufreq directory says:

cpuinfo_cur_freq
800000

cpuinfo_max_freq
1600000

cpuinfo_min_freq
800000

scaling_cur_freq
800000

scaling_driver
powernow-k8

scaling_max_freq
800000

scaling_min_freq
800000

affected_cpus
0 1

so but i've already tried to change the scaling_max_freq to 1600000 but nothing happens @ su mode.
if i add an line with
echo 1600000 > /sys/devices...../cpufreq/scaling_max_freq
at booting before entering init 1 modus ... nothing happens or it says directory or file not found.

so im not able to change this file.

I'm not even sure how the scaling_max and the cpuinfo_max can be different values. Anyone else have any ideas?

By the way, what governor are you using?

-Mark Langsdorf
Operating System Research Center
AMD

Have you tried manually at some later stage, e.g. in init 3. You may want to use:
cat /dev/zero >/dev/null
to set the processor under load and check:
scaling_cur_freq or /proc/cpuinfo for freq changes.

This is when the frequency got limited. AFAIK for x86 there are only two possibilities:
- thermal (check "watch -n1
cat /proc/acpi/thermal_zone/*/{trip_points,temperature}") If the passive cooling trip point got exceeded passive cooling gets enabled (also stated in /proc/acpi/thermal_zone/*/state)
- _PPC BIOS freq limitation
Some BIOSes limit freq e.g. when working on battery.
In some BIOSes this is configurable (-> check BIOS) There was a bug, that if booted with limited freq (e.g. on some BIOSes when booted on battery), you will never be able to reach max_freq again. This has been fixed in 2.6.22-rc1 or -rc2.

If it's _PPC it could also be some other general ACPI breakage.

Best is you try with ondemand governor, like that you can't forget anything to set up in userspace.

Thomas

So i tried this commands after adding cat /dev/zero >/dev/null to my boot file.

cat /proc/acpi/thermal_zone/*/state
state: ok

cat /proc/acpi/thermal_zone/*/{trip_points,temperature}

critical (S5):           100 C
temperature:             51 C

so i already done an bios update to newest version 7.0c and checked my bios but no "_PPC BIOS freq limitation" anywhere

now i tried using only the ondemand governor.
so i got as default Performance, so i also tried
ondemand .... Nothing happens to my cpu.

Hmm, it can't be the passive trip point, your BIOS doesn't declare one. Only thing I could imagine is _PPC or some weird cpufreq bug.

Could you open a bug entry on bugzilla.kernel.org, put me into CC and attach acpidump and dmesg output.

If you know how to compile the latest kernel, it would be great if you could add CPU_FREQ_DEBUG=y and ACPI_DEBUG=y in the compile options and boot with cpufreq.debug=7 and again post dmesg output after powernow-k8 got loaded and ondemand got activated for a short while.

Thomas