@ Jan,
I hope you are not running with idle=poll, because thats meant to kill the cpu
;). It indeed keeps the cpu more or less 100% in C0, and I just wanted to give
an example what it means to be almost all the time in C0 ;). (In my case, the
fans won't stop whining if I give idle=poll and temperature is always above
50C.)
I had fiddled with clocksources some time back with no use. First of all, pit
apparently only works in single processor. Try see if your kernel has smp (if
so, try with smp disabled). I have observed that the system starts with tsc as
my clocksource. The moment I enable C2 and C3 states (by inserting a cpuidle
governor), I get a message that my tsc has been marked unstable and the
clocksource is changed to hpet. And 'jiffies' and 'acpi_pm' screw up my time
(the system clock goes very slow or fast).
At the moment, I would really suggest you to try enabling the cpuidle option in
the kernel, but without any governors. Atleast in my case, the problem doesn't
appear if my cpu doesn't go to C2/C3 at all. With this, powertop will show as
if cpu is 100% in C0, but actually, it will be using both C0 and C1 as
mentioned earlier. So, you can try this and see if it solves the issue and also
whats the power consumption.
Btw, some time back at gentoo forums, someone (who also had this high wakeups
issue) had suggested that the userspace cpufreq governor isn't that good (btw,
his problem also mysteriously disappeared for good, after some arbirtary kernel
config changes). So try if changing the cpufreq governor in kernel to
performance or ondemand makes those 'tsc unstable' messages to disappear.
Jan Willies <[EMAIL PROTECTED]> wrote: Andreas Mohr wrote:
> On Tue, Oct 02, 2007 at 01:16:11PM +0200, Jan Willies wrote:
>> hirakendu das wrote:
>>> Btw, Is your cpu dual core? If so, does disabling smp work ? Do you also
>>> get the "tsc marked unstable" message in kernel log ?
>> It's a single core, AMD Turion(tm) 64 Mobile Technology ML-30.
>>
>> Time: tsc clocksource has been installed.
>> Marking TSC unstable due to: cpufreq changes.
>> Clocksource tsc unstable (delta = -108470834 ns)
>
> Oh, right, just a random shot:
>
> Doing
> cat /sys/devices/system/clocksource/clocksource*/*
> (and fiddling with those settings) might perhaps influence ACPI idle state
> behaviour? This would hint at more kernel work to be done then...
available_clocksource: acpi_pm pit jiffies tsc
current_clocksource: acpi_pm
I tried echo pit > current_clocksource but that completely freezes the
laptop. Sorry for being a noob but can you give me some hints what to
try specifically?
- Jan
_______________________________________________
Power mailing list
[email protected]
http://www.bughost.org/mailman/listinfo/power
---------------------------------
Be a better Globetrotter. Get better travel answers from someone who knows.
Yahoo! Answers - Check it out._______________________________________________
Power mailing list
[email protected]
http://www.bughost.org/mailman/listinfo/power
