On 22/03/10 21:28, Johann-Christoph Jacob wrote: > Your 2 second timeout sounds like the default timeout of > usb autosuspend. This is configurable via a sysfs file: > /sys/bus/usb/devices/.../power/autosuspend > as described in the kernel documentation at: > http://www.mjmwired.net/kernel/Documentation/usb/power-management.txt
Thanks, I wasn't aware of that. I tried changing the default time-out to some higher value but it doesn't seem to have any effect. I'm not sure whether usbcore should be unloaded and reloaded, though. Now I'm confused. This afternoon the CPU was happily going to C3 after enabling autosuspend with PowerTOP, but now it doesn't like it again. About 84% of the time it stays on C2 even though it doesn't complain about the USB mouse. I get similar results if I unplug it. And in any case, is this an expected behavior for the mouse? Shouldn't it also wake up when it's moved instead of only when it's clicked? Cheers Jose This is what I got just now after pressing the U key in PowerTOP: Cn Avg residency C0 (cpu running) ( 4.3%) polling 0.0ms ( 0.0%) C1 0.7ms ( 0.3%) C2 15.2ms (84.3%) C3 4.8ms (11.1%) P-states (frequencies) 2.00 Ghz 0.1% 1.80 Ghz 0.0% 1.60 Ghz 0.0% 1400 Mhz 0.0% 600 Mhz 99.9% Wakeups-from-idle per second : 83.1 interval: 15.0s no ACPI power usage estimate available Top causes for wakeups: 33.4% ( 37.0) thunderbird-bin 17.8% ( 19.7) [kernel scheduler] Load balancing tick 14.7% ( 16.3) [ide1] <interrupt> 10.9% ( 12.1) medit 5.5% ( 6.1) kblockd/0 4.0% ( 4.5) [kernel core] mod_timer (blk_unplug_timeout) 3.4% ( 3.8) [kernel core] hrtimer_start (tick_sched_timer) 2.6% ( 2.9) acroread 1.7% ( 1.9) xulrunner-stub 1.3% ( 1.5) dirmngr 0.8% ( 0.9) [ide0] <interrupt> 0.7% ( 0.7) devkit-disks-da 0.7% ( 0.7) hald-addon-stor 0.5% ( 0.5) kmix 0.3% ( 0.3) kded4 0.2% ( 0.2) [kernel core] mod_timer (dev_watchdog) 0.2% ( 0.2) kile 0.2% ( 0.2) kded 0.2% ( 0.2) plasma-desktop 0.1% ( 0.1) urxvt 0.1% ( 0.1) Xorg 0.1% ( 0.1) events/0 0.1% ( 0.1) klauncher 0.1% ( 0.1) PS/2 keyboard/mouse/touchpad interrupt 0.1% ( 0.1) flush-3:0 0.1% ( 0.1) kjournald 0.1% ( 0.1) ssh-agent 0.1% ( 0.1) [kernel core] mod_timer (sync_supers_timer_fn) 0.1% ( 0.1) gconfd-2 0.1% ( 0.1) hald 0.1% ( 0.1) cron Recent USB suspend statistics Active Device name 0.0% USB device 2-1 : USB-PS/2 Optical Mouse (Logitech) 0.0% USB device usb4 : UHCI Host Controller (Linux 2.6.32-3-686 uhci_hcd) 0.0% USB device usb3 : UHCI Host Controller (Linux 2.6.32-3-686 uhci_hcd) 0.0% USB device usb2 : UHCI Host Controller (Linux 2.6.32-3-686 uhci_hcd) 0.0% USB device usb1 : EHCI Host Controller (Linux 2.6.32-3-686 ehci_hcd) Recent audio activity statistics Active Device name Recent SATA AHCI link activity statistics Active Partial Slumber Device name > Cheers > > Johann-Christoph > > > Jose Rodriguez wrote: >> On 22/03/10 19:37, Arjan van de Ven wrote: >>> On 3/22/2010 12:22, Jose Rodriguez wrote: >>>> -----BEGIN PGP SIGNED MESSAGE----- >>>> Hash: SHA1 >>>> >>>> On 15/03/10 18:49, Jose Rodriguez wrote: >>>>> Hello. >>>>> >>>>> Pentium M mobile on Debian Sid here. After upgrading the kernel to >>>>> 2.6.32-3 from Debian repositories my CPU doesn't go to the C3 >>>>> state, or >>>>> that is what PowerTOP reports. My previous kernel was 2.6.26. Is there >>>>> anything I can do about this? >>>> Just noting that I observe the same behavior with 2.6.33. Could at >>>> least >>>> somebody confirm whether this is a known issue or more likely something >>>> on my end? >>> >>> how many wakeups do you have? >>> >>> is USB really all in self suspend? (eg if you have usb mouse/keyboard... >>> do things go away if you unplug them?) >> >> Oh, I feel like an idiot now. It actually is the mouse what's holding >> things back. C3 state is reached by both unplugging it or pressing the U >> key in PowerTOP. However, there's a reason why I wasn't following >> PowerTOP's advice regarding autosuspend: the pointer stops moving until >> I click again, and then it will stop responding if I don't move it for >> about 2 seconds. I believe I didn't need to unplug the mouse to get to >> the lowest power state with the previous kernel I used. In any case, the >> mouse behavior with autosuspend is rather annoying, I presume it is not >> intended? >> >> Cheers >> >> Jose >> >> >> This is with the mouse plugged: >> >> >> Cn Avg residency >> C0 (cpu running) ( 0.5%) >> polling 0.0ms ( 0.0%) >> C1 0.2ms ( 0.0%) >> C2 26.1ms (99.5%) >> C3 0.0ms ( 0.0%) >> >> P-states (frequencies) >> 2.00 Ghz 1.1% >> 1.80 Ghz 0.0% >> 1.60 Ghz 0.0% >> 1400 Mhz 1.0% >> 600 Mhz 97.9% >> >> Wakeups-from-idle per second : 38.1 interval: 15.0s >> no ACPI power usage estimate available >> >> Top causes for wakeups: >> 30.1% ( 16.3) [ide1] <interrupt> >> 16.4% ( 8.9) [kernel scheduler] Load balancing tick >> 14.4% ( 7.8) [kernel core] hrtimer_start (tick_sched_timer) >> 8.6% ( 4.7) [kernel core] mod_timer (blk_unplug_timeout) >> 8.6% ( 4.7) kblockd/0 >> 7.4% ( 4.0) [kernel core] mod_timer (rh_timer_func) >> 2.7% ( 1.5) dirmngr >> 2.5% ( 1.3) xulrunner-stub >> 2.0% ( 1.1) Xorg >> 1.6% ( 0.9) hald-addon-stor >> 1.0% ( 0.5) devkit-disks-da >> 0.9% ( 0.5) kmix >> 0.6% ( 0.3) [ide0] <interrupt> >> 0.6% ( 0.3) kded4 >> 0.5% ( 0.3) kile >> 0.4% ( 0.2) kded >> 0.4% ( 0.2) plasma-desktop >> 0.4% ( 0.2) [kernel core] mod_timer (dev_watchdog) >> 0.1% ( 0.1) PS/2 keyboard/mouse/touchpad interrupt >> 0.1% ( 0.1) kwin >> 0.1% ( 0.1) syslogd >> 0.1% ( 0.1) gconfd-2 >> 0.1% ( 0.1) hald >> 0.1% ( 0.1) thunderbird-bin >> 0.1% ( 0.1) kjournald >> 0.1% ( 0.1) ssh-agent >> 0.1% ( 0.1) [kernel core] mod_timer (sync_supers_timer_fn) >> 0.1% ( 0.1) flush-3:0 >> >> A USB device is active 100.0% of the time: >> USB device 2-1 : USB-PS/2 Optical Mouse (Logitech) >> >> Suggestion: Enable USB autosuspend for non-input devices by pressing the >> U key >> >> Recent USB suspend statistics >> Active Device name >> 100.0% USB device 2-1 : USB-PS/2 Optical Mouse (Logitech) >> 0.0% USB device usb4 : UHCI Host Controller (Linux 2.6.32-3-686 >> uhci_hcd) >> 0.0% USB device usb3 : UHCI Host Controller (Linux 2.6.32-3-686 >> uhci_hcd) >> 100.0% USB device usb2 : UHCI Host Controller (Linux 2.6.32-3-686 >> uhci_hcd) >> 0.0% USB device usb1 : EHCI Host Controller (Linux 2.6.32-3-686 >> ehci_hcd) >> >> Recent audio activity statistics >> Active Device name >> >> Recent SATA AHCI link activity statistics >> Active Partial Slumber Device name >> >> >> ############################################################ >> >> >> And this is after I press U and wait a bit: >> >> >> >> Cn Avg residency >> C0 (cpu running) ( 0.2%) >> polling 0.0ms ( 0.0%) >> C1 0.1ms ( 0.0%) >> C2 2.6ms ( 3.1%) >> C3 38.3ms (96.7%) >> >> P-states (frequencies) >> 2.00 Ghz 0.1% >> 1.80 Ghz 0.0% >> 1.60 Ghz 0.0% >> 1200 Mhz 0.9% >> 600 Mhz 98.9% >> >> Wakeups-from-idle per second : 37.3 interval: 15.0s >> no ACPI power usage estimate available >> >> Top causes for wakeups: >> 32.1% ( 16.3) [ide1] <interrupt> >> 21.3% ( 10.9) [kernel scheduler] Load balancing tick >> 13.6% ( 6.9) kblockd/0 >> 9.2% ( 4.7) [kernel core] mod_timer (blk_unplug_timeout) >> 7.6% ( 3.9) [kernel core] hrtimer_start (tick_sched_timer) >> 3.0% ( 1.5) dirmngr >> 2.6% ( 1.3) xulrunner-stub >> 2.1% ( 1.1) Xorg >> 1.7% ( 0.9) hald-addon-stor >> 1.4% ( 0.7) [ide0] <interrupt> >> 1.0% ( 0.5) devkit-disks-da >> 0.9% ( 0.5) kmix >> 0.7% ( 0.3) kded4 >> 0.4% ( 0.2) kile >> 0.4% ( 0.2) plasma-desktop >> 0.4% ( 0.2) kded >> 0.3% ( 0.1) kjournald >> 0.3% ( 0.1) [kernel core] mod_timer (dev_watchdog) >> 0.1% ( 0.1) PS/2 keyboard/mouse/touchpad interrupt >> 0.1% ( 0.1) [kernel core] mod_timer (sync_supers_timer_fn) >> 0.1% ( 0.1) thunderbird-bin >> 0.1% ( 0.1) ssh-agent >> 0.1% ( 0.1) flush-3:0 >> 0.1% ( 0.1) syslogd >> 0.1% ( 0.1) gconfd-2 >> 0.1% ( 0.1) hald >> >> Recent USB suspend statistics >> Active Device name >> 0.0% USB device 2-1 : USB-PS/2 Optical Mouse (Logitech) >> 0.0% USB device usb4 : UHCI Host Controller (Linux 2.6.32-3-686 >> uhci_hcd) >> 0.0% USB device usb3 : UHCI Host Controller (Linux 2.6.32-3-686 >> uhci_hcd) >> 0.0% USB device usb2 : UHCI Host Controller (Linux 2.6.32-3-686 >> uhci_hcd) >> 0.0% USB device usb1 : EHCI Host Controller (Linux 2.6.32-3-686 >> ehci_hcd) >> >> Recent audio activity statistics >> Active Device name >> >> Recent SATA AHCI link activity statistics >> Active Partial Slumber Device name >> >> >> _______________________________________________ >> Power mailing list >> [email protected] >> http://www.bughost.org/mailman/listinfo/power >> _______________________________________________ Power mailing list [email protected] http://www.bughost.org/mailman/listinfo/power
