FreeCalypso firmware update

[Mychaela Falconia]

Hello again everyone,

I have given our new Citrine firmware with L1 from tcs211-l1-reconst a
more thorough round of testing, and here are my findings:

* As far as I can tell, the fw works equally well on all 3 targets I
  have tested it on: C139, GTA02 and Pirelli.

* Deep sleep works and appears to be totally solid - even on the
  Pirelli which I previously thought to be troubled in this area.

* EFR works no worse than plain FR, whereas with L1-2014 it was
  causing the DSP to blow up with errors.

* AMR does not work, neither full rate nor half rate - only noise is
  heard in the earpiece speaker.

Thus relative to L1-2014, we've got the deep sleep bug fixed and we've
gained the EFR codec, but AMR is still broken.  However, in my testing
I also observed some intermittent bad behaviour which I never saw
before, hence it is likely a regression relative to our previous state:

* Sometimes nothing but silence is transmitted in the voice uplink
  path, while the downlink audio sounds just fine in the earpiece
  speaker.  Naturally this behaviour is difficult to catch in a test
  environment when calling non-human test numbers like time of day
  etc: this simple approach tests only the downlink.  Testing uplink
  requires either calling (bothering) another human to see if she can
  hear you, or calling a voice mail number, trying to record something,
  then checking to see if it got recorded - a pita.

  This behaviour is intermittent: I was able to call my S.O. 3 times
  from this Citrine fw running on my Pirelli, once in FR mode and 2
  times in EFR mode, and we were able to hear each other and talk just
  fine.  But a test call from my GTA02 to a voice mail system failed
  to record anything, yet when I later tried it again from the same
  GTA02 with the same fw and the same SIM (not even reflashed in
  between, only power-cycled and the SIM was removed and reinserted),
  I was able to record a test message on the same voice mail system
  just fine.

* Sometimes the L1A task seems to freeze, like it is blocked waiting
  for something and therefore not executing.  The visible behaviour is
  that all L1 trace output suddenly stops, but the firmware is not
  completely dead: RVT System Time messages keep coming every 20 s,
  and ATI still responds to commands - one can send AT and get OK back
  - although of course it can't do anything useful with L1 being
  blocked.

I do not have a concrete plan yet as to how we should proceed forward
- I need to give it some more thought.

M~