GSM flowcontrol problem does not appear to affect GTA02

[Mike (mwester)]

Regarding Bug #1376 
(http://bugzilla.openmoko.org/cgi-bin/bugzilla/show_bug.cgi?id=1376), I've had 
an opportunity over the past several days to put the GTA02 through the same 
battery of serial comms tests I've run on the GTA01.

I'm happy to note that the GTA02 does not seem to be affected by this 
troublesome problem.  In side-by-side testing, in all cases where the GTA01 
under test would fail with a buffer overrun from the GSM, the GTA02 read all 
data without data loss.

The astute observer might note that the GSM is supposed to be exactly the same 
on the two units, and therefore if the problem is in the GSM, then it is not 
really solved on the GTA02, rather my tests failed to trigger it.  This is a 
fair critisism, however I do think that the resume-from-suspend sequence 
represents a rather more extreme situation regarding IRQ latency and load than 
we are likely to hit in normal usage.  (I defer to the experts to comment on the 
possible impact of the Glamo on IRQ latency -- is there a test I can/should run 
to see if there is some other situation that might trigger this on the GTA02?)

Regards,
Mike (mwester)