DABUSB firmware source code has been released this week

Felipe Sanches juca at members.fsf.org
Mon Jan 3 16:32:10 UTC 2011

On Sun, Jan 2, 2011 at 7:04 AM, Alexandre Oliva <lxoliva at fsfla.org> wrote:
> On Jan  1, 2011, Felipe Sanches <juca at members.fsf.org> wrote:
>> No. this is just the 8051 assembly code (actually the original source
>> code developed by the manufacturer, not the result of disassembly of
>> the nibary). The FPGA bitstream still does not have the VHDL sources
>> available. We can try to convince the manufacturer to release those
>> too.
> Just to make sure: this means nothing needs to change in Linux-libre
> because of this development, right?  I don't know enough about the
> firmware files in question to relate what you're saying with them.

Well... right now nothing changes for LinuxLibre because the device
remains non-free (it depends on 2 firmware images: one - the 8051 code
- now has source code available but the other - the fpga configuration
bitstream - doen't yet). On the next kernel release this driver will
be in the staging directory because they are planning to remove it at
all in the following release if nobody gets back to work on it (it is
considered to be an old, deprecated driver).

We may remove just the bitstream in the next linuxlibre release, but
the device can't work without the fpga configuration, so it is not
really important to keep the free firmware while we do not have access
to the fpga source code. There is also an additional issue: the
current driver does not work with the real device at all because it
does not list the correct usb deviceId of the commercial product. This
driver source code is old and supports only the product prototype. I
guess that the company got the code into the mainline tree while
developing the product but then did not get further versions updated
after licensing the technology for commercial use by Terratec. They
mantain their current driver source code distribution out-of-tree.
This out-of-tree code is from 2010 and is said to work with the device
while the mainline tree code is 10 years old.

The only practical thing to do is to get their new driver code into
mainline and convince them to release the fpga source code. Otherwise
we should simply let it die (if nobody moves than it will be removed
from Linux within 2 releases). I guess that this is an example of the
importance of having statistics of hardware usage/popularity in order
to define our hacking priorities.

Felipe Sanches

More information about the linux-libre mailing list