[squeak-dev] Re: squeak image sometimes takes very long to be responsive after XO suspend

[Yoshiki Ohshima]

  Tomeu and all,

At Fri, 16 May 2008 13:53:16 +0200,
Tomeu Vizoso wrote:
> 
> On Fri, May 16, 2008 at 1:50 PM, Ties Stuij <cjstuij at gmail.com> wrote:
> >
> > Basically when coming out of suspend, the Squeak process takes up lots
> > of cpu power and can be unresponsive for about a minute on build 703
> > (other builds not yet tested).
> 
> What about:
> 
> - launch etoys
> - check its pid
> - attach to it with strace -p PID and log the output to a file
> - suspend
> - resume
> - check what etoys is doing in that file
> 
> Good luck,

  I'm now looking into this, but this may not be an issue with Etoys.
I did strace but don't really see any anomaly.  (Saw some interesting
things, but.)

  Suppose I start Pippy and run the "Lines" example, and press the
power button to suspend.  The laptop suspends and the power LED
blinks.  But, here is an interesting happens; if I rub the touch pad
quickly, or put my four fingers together on the keyboard (at shift,
ctrl, tab and `) and slide them over the keyboard to cause a lot of
key input (while the laptop is suspended), the the pattern of power
LED blinking changes.  The LED stays on for a while and turns to off,
and come back to on, etc.  If I press the power button while the
abnormal LED pattern is going, the button press is often ignored.  Or,
it wakes up one second or such but goes back to sleep.  If these
things happen, waking up the unit takes time.  Etoys doesn't have to
be running.

  When Etoys is running, but you don't touch the laptop while it is
suspended, it doesn't happen (that often).

  This is on a G1G1 machine, update.1-708 with the firmware that comes
with it.  I created a track ticket (#7196).

-- Yoshiki