[PATCH] fix-pm-gsm-disable-serial-drive-when-off.patch

[Werner Almesberger]

Dieter Spaar wrote:
> Is tweaking a few 100 ms really that important, considering the
> problems it might cause ?

No, to the contrary. My original question was why we had such a
long delay in the first place, but that has been answered.

Now I'm interested in whether the delay we have is safe. To know
this, we need to know a) what the typical time-to-ready is and
b) how much it varies.

Engineers like round numbers so the total time may well be 1 s
from the moment the button press is debounced, so the current
1.5 ms would be safe and so would the 500+750 = 1250 ms Andy has
suggested. I would hope that it's fairly constant as well, but
without measuring, this is just a wild guess.

> A reliable way is to wait till the AT
> command interpreter tells it is ready, but this requires a more
> complicated way to check.

Hmm, this would require user space (i.e., the middleware) to
take care of the GSM setup business. If Mickey wants this task,
I won't stand in his way :-)

- Werner