FreeCalypso website update

Mychaela Falconia falcon at ivan.Harhan.ORG
Sun Nov 22 05:08:10 CET 2015


Hello everyone,

I just made an update to the www.freecalypso.org home page reflecting
the new reality we are in thanks to the tcs211-c139 port.  You may
need to hit refresh in your browser.

Going forward, the first logical thing I will need to do is dig into
the UI code in tcs211-c139 to find and fix its crashing and other
show-stopping bugs, and also implement battery charging and turning
off of the display when idle, so that a C139 running this fw will
become a practically usable (albeit ultra-minimalistic) libre phone.

As I wrote for the website, my own libre phone goal is something more
like the Pirelli DP-L10.  The 3 major features sacrificed when
downgrading from the Pirelli to Mot C139 are as follows:

* Hands-free loudspeaker;
* Ability to use a wired headset - C139 has the jack, but we need it
  for the debug UART, hence the wired headset function is lost;
* The ability to both charge the battery and PC-sync your data (SMS
  and call history etc) through the same cable - Pirelli's USB-serial
  arrangement is so nice in this regard.

I could live with the loss of the USB charge+data interface, and I may
even try making a special charging cable that would go from a USB A
plug to Mot C1xx: as far as I understand, Mot's chargers put out
regulated +5V, so I don't see why taking the same +5V from a USB port
won't work.  Then maybe make an even hackier cable that would plug
into a USB A port on one end, and split into two plugs on the other
end: one going into the charging jack (carrying +5V) and the other
going into the headset jack, carrying data through a USB-serial chip.

But the loss of both the loudspeaker *and* the wired headset capability
for hands-free operation is going to be a real killer for me.  Thus
*my* personal libre phone goal can only be satisfied by building our
own hardware: the prospects of turning the existing Pirelli DP-L10
into a libre phone through a firmware change are rather bleak: the
mysterious undocumented hardware is too much of an obstacle.

But I am still willing to produce a minimalistic libre phone using Mot
C139 hardware, for two reasons:

1. Perhaps others' needs are more basic than mine, and they would not
   mind a libre phone which you have to hold up to your ear, with no
   loudspeaker or wired headset options.  In any case the tcs211-c139
   subproject is now so close to a practically usable phone that it
   seems morally impermissible to leave it unfinished at this point,
   i.e., we need to finish it.

2. Even thinking selfishly, my own dream phone that would require
   building our own hw will still need UI firmware, and the issues we
   are currently seeing in tcs211-c139 are in need of solving no
   matter what the hw platform is, so why not fix them now within
   tcs211-c139, and then later reuse that work in our gcc-built fw on
   our own hw.

After the milestone of a practically usable albeit minimalistic
tcs211-c139, the next logical step on the software track would be to
revisit our not-quite-working gcc-built gsm-fw.  However, the problem
with this lovely idea is that I could not think of any way to attack
our two L1 bugs (deep sleep breakage and the show-stopper of voice
calls not working on Operator 310260's network even with codec
restrictions) except the brutally painful one: retrofitting LoCosto L1
C modules back into TCS211 one module at a time and trying to match
the disassembly of the original.

The task of reconstructing a working L1 for our version of the Calypso
chipset from the LoCosto source and TCS211 binary objects would be
brutally painful, but not impossible.  If we do not find any better
way, I will have to bite the bullet and go down that path: I do want a
libre phone without being beholden forever to TI's proprietary compiler
toolchain (complete with a not-fully-understood assembler and linker
and some very annoying bugs) and to the byzantine build system of
TCS211.  However, just to set some realistic expectations, let me
clarify exactly what I mean by this task being brutally painful: it is
painful to the point that an entire year or two may be needed just for
this task alone.

Given the bleak prospect of just how much work we'll have to do if we
have to recover the necessary L1 code in the painful way, I am thinking
that we should first make an all-out effort to try to find and obtain
a surviving copy of some original source for the code in question.  In
other words, we should make an all-out effort to find and obtain
another surviving copy of TI's TCS211 fw (different from the one we
got from that particular obscure manufacturer of GSM/GPRS modems) that
would have its L1 component in source form, rather than a blob - or
alternatively, some other L1 source that targets the Calypso and
supports DSP version 36.

To be sure, we have no way of knowing for certain whether or not there
is another surviving copy of TCS211 anywhere in the world.  It is
entirely possible that the copy of TCS211 we got from Sotovik, with
the specific source-binary mix featured therein, is the world's only
surviving copy of any subversion of TCS211, and that the source we are
after does not exist anywhere at all, completely lost to Humanity.  And
it's equally possible that a copy of such source does exist somewhere:
maybe in the junk pile of some Chinese no-name phone manufacturer that
made a Calypso phone 10 years ago, or in the private stash of some
former TI employee who got canned when TI shut that business unit down,
or maybe even TI as a company kept some archives - who knows.

I know that such copies of TI's fw *used* to float around in the past:

http://bbs.52rd.com/forum.php?mod=viewthread&tid=17358

It's dated 2006-03-22; I emailed the address listed in the ad (in
English, as I don't speak Chinese), but got no response.  The BBS post
is from almost 10 years ago, so it's hard to say if that person is
even still around...

I need to put some more thought into where and how we should look for
a possible surviving copy of TCS211 with L1 in source form, but I just
wanted to put the thought out there.  Once again, making an all-out
search effort does not guarantee that we'll get anything, so we still
need to be prepared to reconstruct the needed L1 code in the brutally
painful way - but if that brutally painful way will take a year or two
onto itself, there is no reason why other members of our community
can't be searching the world high and low for a surviving source at
the same time.

By the way, would any of our European members happen to know if this
place still exists:

http://www.ti.com/europe/docs/sites/france.htm

I get the impression that the office in question has been fully closed,
and that the page on TI's website (found via a Google search for
"TI Nice France") is a forgot-to-delete:

http://www.riviera-buzz.com/features/news/item/94-texas-instruments-in-villeneuve-loubet-to-close-with-major-job-losses.htm

For those who don't know, the office in question is the one where they
developed the chips themselves (Calypso and friends), the DSP code, L1
and the SSA (System Software & Applications) parts of the firmware -
i.e., the juiciest parts.  In contrast, TI-Berlin (originally Condat
Gmbh) did the mostly hardware-independent upper layers, which we
already got well taken care of thanks to the LoCosto source find.
TI Aalborg (Denmark) did the development boards.

M~


More information about the Community mailing list