FreeCalypso on the D-Sample

Mychaela Falconia mychaela.falconia at gmail.com
Sat Jan 20 07:40:07 UTC 2018


Hello FreeCalypso community,

I've got some exciting news: after almost two and a half years of
gathering dust as a fancy souvenir, the D-Sample board from TI which I
scored back in 2015 is finally running our own FC Magnetite firmware!
Well, almost - I still haven't fully reconstructed the tpudrv10 driver
code for Clara RF (I am reverse-engineering from the 20020917 binary-
only, no symbols fw image this board came flashed with), so the radio
does not work at all, but I've got a lot of other things to work:

* I can compile images for target dsample from our FC Magnetite source
tree and flash them into the board, and they boot successfully.  This
particular D-Sample board (the only one I have, and may well be the
world's last remaining one, period) is an early board with an older
version of the Calypso chip on it - C preprocessor symbol CHIPSET
needs to be set to 8 instead of 10, the clocks run slower, the DSP ROM
is an older version - but TI's mainline fw obviously supported this
config once upon a time, and I have resurrected this support in
Magnetite.

* The DSP is alive, despite being the older version.  I got a DSP
patch code that matches this ROM version from the TSM30 source, and
the TST_C trace from L1 reports the expected ROM version, patch
version and checksum.  This DSP plays beeps successfully; I haven't
tried E1 melodies yet, but they will probably work fine too.

* Now that we are running our own fw build instead of the ancient one
from 20020917 this board came flashed with, we have the full freedom
to poke at the audio subsystem through ETM commands; I have used this
method to confirm that the speaker in this kit works the way I
expected: the speaker amplifier is connected to both EAR and AUX
outputs from the Iota ABB, but through different resistor values, so
that the speaker is quiet like a typical earpiece when driven through
EAR, but loud as a handsfree/ringing loudspeaker when driven through
AUX.  The firmware's default on boot is to have both EAR and AUX
enabled; this default makes the speaker loud on the D-Sample, but it
also plays nice with the other targets (C1xx, Openmoko devices and our
own FCDEV3B) on which only the EAR interface is connected.

* The SIM interface is alive, and the fw successfully retrieves my own
phone number, phonebook entries and saved messages from the SIM,
despite having no working radio yet.

* The fancy 176x220 pixel color LCD in the handset part of the kit
also works for the most part with the UI-enabled build of the fw,
although I see some strangeness which I need to debug some more.
The keypad buttons in the handset part work as well.

Going forward, I will continue the already-started work to reverse-
engineer and reconstruct the tpudrv10 driver code for Clara RF - if we
can get the radio to work with FreeCalypso fw on this ancient board
(and don't worry, I'll be testing it extensively on the CMU200 before
hitting any real airwaves), we will gain a platform for exercising the
demo / prototype / proof-of-concept UI code we got from TI (i.e., the
UI code for which this D-Sample is its native target) and shaking the
bugs out of it, and that in turn may enable us to develop usable end
user fw for the Mot C139 ahead of our own FreeCalypso Libre Dumbphone
hardware.  The latter is still my ultimate goal and I am not giving it
up, but it would surely be nice to have FreeCalypso on the C139 as an
intermediate step.

Hasta la Victoria, Siempre,
Mychaela aka The Mother


More information about the Community mailing list