FreeCalypso > hg > fc-magnetite
view doc/Calypso-version-override @ 635:baa0a02bc676
niq32.c DTR handling restored for targets that have it
TI's original TCS211 fw treated GPIO 3 as the DTR input (wired so on C-Sample
and D-Sample boards, also compatible with Leonardo and FCDEV3B which have a
fixed pull-down resistor on this GPIO line), and the code in niq32.c called
UAF_DTRInterruptHandler() (implemented in uartfax.c) from the
IQ_KeypadGPIOHandler() function. But on Openmoko's GTA02 with their official
fw this GPIO is a floating input, all of the DTR handling code in uartfax.c
including the interrupt logic is still there, but the hobbled TCS211-20070608
semi-src delivery which OM got from TI contained a change in niq32.c (which
had been kept in FC until now) that removed the call to
UAF_DTRInterruptHandler() as part of those not-quite-understood "CC test"
hacks.
The present change fixes this bug at a long last: if we are building fw for a
target that has TI's "classic" DTR & DCD GPIO arrangement (dsample, fcmodem and
gtm900), we bring back all of TI's original code in both uartfax.c and niq32.c,
whereas if we are building fw for a target that does not use this classic GPIO
arrangement, the code in niq32.c goes back to what we got from OM and all
DTR & DCD code in uartfax.c is conditioned out. This change also removes the
very last remaining bit of "CC test" bogosity from our FreeCalypso code base.
| author | Mychaela Falconia <falcon@freecalypso.org> |
|---|---|
| date | Sun, 19 Jan 2020 01:41:35 +0000 |
| parents | 9f19cc5e46e8 |
| children |
line wrap: on
line source
TI's TCS211 program supported 3 different Calypso silicon versions on their D-Sample and Leonardo boards: C05B (CHIPSET 8, DSP 33), early C035 (CHIPSET 10, DSP 34) and final C035 (CHIPSET 10, DSP 36). In FreeCalypso we generally work only with the last chipset (final C035), but we also support the two earlier ones to a good extent. C05B has been partially exercised on the Mother's D-Sample board (non-functional Clara RF, no genuine tpudrv10.c source and too many unknowns, but at least some of the DSP-based audio services work), whereas Calypso C035 with DSP 34 has been tested and found to work on a specially modified FCDEV3B board - see the FCDEV3B-751774 article. All 3 Calypso chip versions in question are footprint-compatible, i.e., they can all be populated onto the same PCB. Changing between C05 and C035 requires changing the VLRTC strapping on the Iota companion chip, thus not all boards can support Calypso C05 (the newer ones are C035 only), but one can always populate either early C035 (DSP 34) or final C035 (DSP 36) on the same PCB. TI's own D-Sample and Leonardo development boards have most certainly been made with all 3 Calypso silicon versions: D-Sample appears to support both C05 and C035 on the same PCB; the schematics we have for Leonardo show fixed VLRTC strapping for C035, but there also existed an earlier Leonardo version with Calypso C05B. Our own FCDEV3B also has fixed C035-only VLRTC strapping, our regular boards have final C035 chips on them, but we have put a D751774AGHH chip on one experimental board for DSP 34 testing. In the case of target boards that can have different Calypso chip versions populated on them, we want to be able to target all of them without creating a configuration explosion. The implemented solution is that you can pass a target argument to the ./configure.sh script of the following form: dsample-c05b dsample-dsp34 dsample-dsp36 leonardo-c05b leonardo-dsp34 leonardo-dsp36 fcdev3b-dsp34 Our configure.sh script recognizes the hyphen as indicating this construct; the word before the hyphen must be the conventional target name, and the word after the hyphen needs to be one of c05b, dsp34 or dsp36, corresponding to the three supported Calypso silicon versions.