Month: May 2007

powertop: misc causes for wakeups

Posted by – May 26, 2007

Cn Avg residency (10s) Long term residency avg
C0 (cpu running) ( 1.9%)
C1 0.0ms ( 0.0%) 0.0ms
C2 13.3ms (67.5%) 13.4ms
C3 10.9ms (26.9%) 10.5ms
C4 8.2ms ( 3.6%) 8.4ms

Wakeups-from-idle per second : 79.9

Top causes for wakeups:
19.0% (10.3) firefox-bin : schedule_timeout (process_timeout)
18.7% (10.1) <interrupt> : Intel ICH6 Modem, ipw2200, lan0
18.5% (10.0) cpufreq-set : cpufreq_governor_dbs (delayed_work_timer_f
16.6% ( 9.0) <interrupt> : ide0
5.9% ( 3.2) <interrupt> : acpi
5.0% ( 2.7) gkrellm2 : schedule_timeout (process_timeout)
4.6% ( 2.5) : do_setitimer (it_real_fn)
2.8% ( 1.5) wpa_supplicant : schedule_timeout (process_timeout)
0.9% ( 0.5) hald-addon-stor : do_nanosleep (hrtimer_wakeup)
0.9% ( 0.5) <kernel core> : queue_delayed_work_on (delayed_work_timer_

ipw2200 is really bad. ipw2200 by itself means the difference between whether the CPU spends it’s time in C4 or not. Though it seems the bug is not in the laptop but in the AP I’m connected to causing wakeups. Which really stinks since my router doesn’t give any obvious options to change it.

ide0 is waking up because of constant harrassment from hald. hal-disable-polling is included in v0.5.9. Which according to bug #176380 will be stablilized with Gnome 2.18.

ACPI is waking up due to queries from gkrellm2. Changing configs in gkrellm helped this.

Quote: credit to mat|gentoo in #gentoo@irc.freenode.net

Posted by – May 26, 2007

mat|gentoo left the room (quit: “The day Microsoft makes something that doesn’t suck is the day they start making vacuum cleaners”).

It made me giggle. heh

powerTOP: CPU governors

Posted by – May 17, 2007

“Conservative” Governor

2.1% (10.0) cpufreq-set : cpufreq_governor_dbs (delayed_work_timer_f

“Ondemand” Governor

32.2% (50.0) cpufreq-set : queue_delayed_work_on (delayed_work_timer_

The userspace, performance, powersave governors do not poll automatically therefore powerTOP doesn’t list them. The Ondemand governor is suitable while the laptop is powered by AC. While Conservative or powersave is logical to use on battery mode.

powerTOP

Posted by – May 15, 2007

Intel has released a fascinating new utility that reveals what is preventing the CPU from going to sleep/low power state. A few of the bad actors I found on my system so far are ipw2200, hald, and audacious. hald appears to poll the disk constantly for something so it’s ide0 that appears active but not hald directly. Audacious causes 160 wakeups-from-idle all by it’s lonesome. Finally, when everything else is inactive my wireless driver (ipw2200) keeps the CPU awake constantly. If I’m plugged into a hardwire connection the CPU can sit in C3/C4 for greater than 90% of the time.