[indiana-discuss] b111a: Power Management and nwam-issue ?

[Mark Haywood]

Detlef Drewanz at Sun.COM wrote:
> Mark,
> I watched the discussion thread from the week end.
> I changed in /etc/power.conf cpupm to
> "cpupm enabled poll-mode"
> and indeed after doing that the system switched down to 800 Mhz 
> clocking. I was able to see this with kstat and powertop. Good !
>
> Then I suspended my Toshiba M9 Laptop and resumed this morning.
> Interestingly this starts in "prstat" with load average of 154 (!) I 
> don't know why this load is so high in suspend mode ? But what' more 
> happening is that now again that powertop (and kstat) shows always 
> 2001 Mhz used. So it looks like that after a suspend/resume the power 
> management is no longer working correctly. I was then not able to 
> bring this power management to work again. I had to reboot to come 
> again down to 800 Mhz (and a low noise system).

What does mpstat tell you? Possibly a manifestation of:

 http://defect.opensolaris.org/bz/show_bug.cgi?id=5693

Mark

>
> Detlef
>
>
> On 04/24/09 17:39, Mark Haywood wrote:
>> Detlef Drewanz at Sun.COM wrote:
>>> Hi,
>>> with b111a now I can boot again my Toshiba M9.
>>>
>>> 1.I discoverred that Power Management seems no longer to work.
>>> (The fan is blowing like hell and "kstat -m cpu_info" shows in
>>> currentClock_Hz 2001000000
>>> but
>>> supported_frequencies_Hz 
>>> 800000000:1200000000:1600000000:2000000000:2001000000
>>>
>>> New Bug with b111a ?
>>
>> The cpu_info kstat might be a little deceiving now that the power 
>> aware dispatcher has integrated. Try running /usr/bin/powertop. It 
>> might be more informative.
>>
>> Mark
>>
>>>
>>> 2. I also have the impression that the boot up is much more disk-io 
>>> intensiv and takes much more time (can't identify yet directly)
>>>
>>> 3. Also: After the boot I see dhcpagent/1 consuming 50 % CPU (looks 
>>> like 1 core) and the nwam applet will not show up in the gnome 
>>> panel. I first have to "kill -9" the dhcpagent, then the nwam applet 
>>> comes up and the cpu consuming goes away.
>>>
>>> Any comments ? Anyone else seeing this too.
>>>
>>> Thanks
>>> Detlef
>