(GTA02) resume after 30 minutes, or absence thereof

[Werner Almesberger]

[ Background: QA found that GTA02v5 sometimes refuses to resume if left
  suspended for more than 30 minuted. ]

Seems that we'll have some fun with this one :) I've tried a number of
times and out of at least 8 "good" runs (*), none failed to resume.

I think we should simply relocate Openmoko to Buenos Aires. The good air
clearly makes most of the bugs people find elsewhere just vanish ;-)

(*) My GTA02v5 often wakes up from a GSM interrupt. Not quite sure why.
    Could be network activity. We had some unexpected GSM wakeup bug in
    late February/early March. I think this was the (in)famous LL-debug
    serial code, correct ? I'll try later without SIM card.

I rebuilt my kernel with the GSM interrupt disabled, and this made the
unexpected wakeups go away (and allowed me to resume normally), except
once, when it still woke up on its own after about 50 minutes of
peaceful sleep.

This wakeup turned out to come from the PMU and was the battery full
indication, i.e., a perfectly harmless if boring condition.

This got me thinking that failure to resume could also be caused by
the system unexpectedly waking up and then draining its battery.
(This couldn't have happened in the unexpected wakeup I observed, but
there are other reasons why the PMU may think it appropriate to disturb
its master.)

Allen, can you please give us a bit more information on the test
scenario ? In particular:

- was the system powered from battery or lab supply ?

- in case of a battery, do you remember in what state it roughly was
  at the time you suspended ?

- is USB connected to a host ?

- is there a SIM card ?

- was this with a freshly booted system or has it been used before
  suspending ?

To make the test more reproducible and to eliminate possible unrelated
sources of errors, could you please:

- begin the test with a fully charged battery ? (Don't use a lab power
  supply.)

- do the suspend on a freshly booted system

If you observe the problem again, could you please report this:

- what is the battery voltage after you find the system unresponsive ?

- if you had a debug board connected, all of the serial output
  (If you use a recent version of neocon, you can log to a file with
  neocon -l file)

Thanks !

- Werner