[Gta04-owner] Status GTA04 EA
Dr. H. Nikolaus Schaller
hns at goldelico.com
Mon Oct 3 10:24:18 CEST 2011
thank you fro bringing up this topic to our all attention.
Am 02.10.2011 um 09:32 schrieb NeilBrown:
> On Mon, 19 Sep 2011 11:06:40 +0200 "Dr. H. Nikolaus Schaller"
> <hns at goldelico.com> wrote:
>> Am 19.09.2011 um 09:52 schrieb Butrus Damaskus:
>>> On Mon, Sep 19, 2011 at 9:13 AM, Dr. H. Nikolaus Schaller
>>> <hns at goldelico.com> wrote:
>>>> 4. FSF is interested in removing the non-free libertas firmware from the
>>>> rootfs. So they suggested to put it into a small MCU. If you are interested,
>>>> please read the project proposal and apply:
>>> Hm, how big is the firmware? Cannot this be solved in another way then
>> approx. 128 kByte.
>>> by adding components -
>>> (e.g. reserving a special RO partition in the flash reserved for the
>>> firmware and distributing the boards already with it)?
>> I have discussed that with RMS - and it is not permitted to be user-changeable
>> by not so easy means. The idea is that is must be "like hardware".
> As you have probably noticed, this issues has generated quite some discussion:
> A lot of it is just people arguing with each other, but there is the
> occasional comment that asks "Why not use an Atheros chip with GPL firmware"
> I don't doubt that you considered this option (and others) and had good
> reason to reject it. If you can recall what the key issues in the choice of
I have not found that they offer a real combo chip like the w2CBW003 (although it is
just two silicon dies in a single package with internal wire bonding).
What I have seen is e.g.:
> hardware was, that might be useful to inform those who want to be informed.
> The particular devices mentioned doesn't support bluetooth, though it seems
> other Atheros chips do.
There have been several factors leading to the current choice of the W2CBW003:
* availability: the W2CBW is available through standard distributors like Mouser,
Avnet etc. in quantities starting at 1. And, the price is (still) reasonable.
* dimensions: it comes in a single 12x12 mm package an requires just external
level shifters, decoupling capacitors and two SMD antennas. So it is smaller and
less complex/risky than anything else we know.
* proven to interface with OMAP3: exactly the same chip is used in the Gumstix Overo,
so we were 100% sure that it can be interfaced to a OMAP CPU and working Linux
drivers are available.
* pure technical decision: the decision was made more than 2 years ago without
knowing or considering that it could become a "political" issue.
We prefer to provide a working system as soon as possible, even if it has
some non-free components instead of discussing for years that a 100% free
system should be achieved, but never comes to existence :) Or crippling down
the feature set. Therefore my personal preference over the FSF-Injector idea
would be that some team is founded to write a free and open replacement for
the Marvell firmware BLOB.
Our goal is to provide a platform that is really well and openly documented,
as good as we can - to support running free and open software on it. But we
are not religious about 100% freedom, if the solution works (i.e. by downloading
some Marvell originated BLOB from a Debian mirror).
And I think we should do something today even if it is only 99% free. Openmoko
hasn't done differently. There is no open information about the display controller
inside the LCM, there is no open information about the GSM module used
in the GTA02 etc. There are no Gerber files. And, the public 3D CAD model is
not the final production data. Nevertheless, we all are proud of their work!
> If there was a reasonably comparable chip (same interface, supports Wifi and
> BT, acceptable price/availability) would it be practical to swap the chip for
> the next board version, or would that push the time line back too much?
Yes. I think there won't be big problems. Most systems run from 3.3V power
and use 3.3V interface (SDIO for WLAN and UART for Bluetooth).
The key issue is that although we strive for high quantities, we have to start
with small batches. And therefore we must be able to buy through distributors.
If there is a working alternative, we would simply switch it into a new design.
A similar thing is the idea to swap the OMAP3 with OMAP4 as soon as a
OMAP4 can be bought in batches much smaller than currently 100k units.
But this may happen next year or in two years or never. I don't know the plans
of TI in this area.
> Personally I have no strong objections to a non-free firmware, but I am
> curious to understand the background to the decision.
I hope I have given some background...
More information about the Gta04-owner