Status report about suspend/resume issues

Holger Freyther zecke at openmoko.org
Thu Oct 30 12:17:41 CET 2008


On Thursday 30 October 2008 11:17:02 Jeremy Chang wrote:

players:
	ompower and qpe

	qpe opens /dev/apm_bios

	ompower executes apm -s

	qpe gets a suspend request
		=> it executes some commands to prepare the modem for suspend
		=> when this completes it ack's the suspend
		=> if it doesn't ack, the modem did not respond or there is a qtopia bug
			=> the device is messed up anyway.


>     And there is only one way to do resume, press the Pwr button.

Or insert a usb cable.

>
>     The problems about suspend/resume generally are as below:
>     1) Unable to suspend, just screen black, Led is still on and ssh
> through usb is still alive. This will make user think it's already
> suspended and lead to further problems. (Trac ticket 1991)

#1991 it is not clear if the device was not suspended, regina is not very 
clear on that.


>     2) Suspend is ok, but just as resuming, it goes to suspend
> immediately after you see the screen for one second. Sometimes it's too
> quick, so the user thinks it can not be resumed.

file a ticket for that, this comes from multiple button presses?

>
>     3) It can not be resumed. This may due to several scenarios, as ticket
> #1884. Matt is also looking into this.

yes, matt solved one issue, there is something else still broken, see the 
posts on the kernel mailinglist (picture from the scope)



>     4) After serveral tyring to wake up by pressing pwr button with at
> least 1 sec interval, finally it wakes up, but it's going to "shutdown".
>

press and hold action of enlightenment, so we miss a key up in the kernel?


>     I am trying to fix this for scenario 1. That's caused by screensaver
> timeout or the timer timeout in illume as soon as the device resumes.
> I am looking into if I can do something in ompower to avoid this.

raster has this system("xset ...") in ompower. Now there are too ugly things, 
first DISPLAY is set for ompower so the command can work, I don't know if 
sandman's patches where ever merged to make apm -s not return until after the 
suspend. So we might invoke xset before going to sleep. The best fix is to open 
the apm bios to get the notification for resume...

z. 



More information about the devel mailing list