From: "Kok, Auke" <[email protected]>
Subject: Re: USB activity in powertop 1.11
Date: Tue, 13 Jan 2009 15:00:29 -0800
> Peter F. Patel-Schneider wrote:
> > I just installed powertop 1.11 on a Thinkpad T60p under Fedora 10.
> >
> > I'm now getting the following amusing message from powertop:
> >
> > A USB device is active 0.0% of the time:
> > USB device 5-1 : BCM2045B (Broadcom Corp)
> >
> > Q - Quit R - Refresh U - Enable USB suspend
> >
> > I have the bluetooth device turned off using the IBM ACPI interface.
>
> hmm
>
> this particular tip got added in 1.10, and it signifies that the device is
> actually in use a significant amount, except that it's actually less than 0.1%
> (which is still a lot of course).
>
> you can verify this by looking at the output of this command:
>
> watch -d 'grep . /sys/devices/*/*/usb*/power/*_duration'
>
> you should be able to see how much it's busy by changing this:
>
> - sprintf(usb_hint, _("A USB device is active %4.1f%% of the time:\n%s"
> + sprintf(usb_hint, _("A USB device is active %4.4f%% of the time:\n%s"
>
> in urbnum.c (for instance).
>
> Can you see how much (for the device mentioned) "connected_duration" grows in
> relation to "active_duration" ? I think there might be something weird
> happening, such as the device being connected less than active, which
> would screw up the calculation, as well as being bogus in the first place.
>
> Auke
I've finally got around to looking more at the problem.
Note that I'm running on an IBM Thinkpad T60p.
The device appears to be a Bluetooth controller.
The first problem is that there is no hci_usb module on the system.
Instead the relevant modules appear to be bluetooth and btusb.
Following the powertop advice doesn't appear to do anything (except
complain about the non-existing kernel module).
Powertop then complains about the device being active 100% of the time,
which is confirmed via the "watch" you suggested.
Thinkpads have an ACPI interface to the bluetooth device,
/proc/acpi/ibm/bluetooth. Disabling the device this way does turn it
off, as verified by the "watch".
However, I'm no longer getting the weird message. Strange.
So the takeaway, for now, appears to be that Thinkpads (maybe only in
some distributions) need a different way to suspend bluetooth. Any
suggestions on how to go about figuring out just how to proceed?
Peter F. Patel-Schneider
_______________________________________________
Power mailing list
[email protected]
http://www.bughost.org/mailman/listinfo/power
