[Ground-station] Question for ORI:

[Bruce Perens]

On Thu, Apr 1, 2021 at 1:46 PM Leffke, Zachary via Ground-Station
<ground-station at lists.openresearch.institute> wrote:

> *So my question is, given that RTL-SDRs are so prolific (same for funcube
> dongles, though maybe to a lesser extent), does it REALLY matter if the ICs
> are proprietary?  Isn’t what actually matters is that the drivers needed to
> use them are open source?*
>

Well, you sort of answered your own question, since it's problematical for
you to incorporate them in a design.

ICs are a big problem in general. Intel has the infamous system management
processor in their chip which seriously degrades the security of your
platform unless you use reverse-engineered stuff to turn it off (given that
you are not the US government, to whom the method is deliberately
disclosed). Although what is in there is mostly figured out through reverse
engineering, it is not publicly documented. The Raspberry Pi 3 boots from
the GPU which is an entire undocumented CPU on the die, I don't know if 4
is the same. And in general you have no idea what is in your IC, which is a
security issue that could fill books.

So, what do we do in the name of holding up standards for openness? Mainly
a design-in policy: Avoid chips that require BLOBs and proprietary
libraries to work. Prefer to design in ICs where you can download all of
the documentation over the web. Avoid chips where the only possible
configuration data available is generated by a piece of proprietary
software, and it's not otherwise documented. There are a lot of those in
the RF world. Where a chip is available with an Open Source driver but not
public documentation, especially if the driver is the result of
reverse-engineering (which is the case for RTL), we should prefer ones that
are documented.

Then there is the issue of intercommunication. We need higher standards for
design-in there. Intercommunication protocols, modulations, etc. should be
publicly documented and usable without having to execute any sort of patent
licensing. Where standards are designed-in, they should not include
royalty-bearing patents.

    Thanks

    Bruce
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openresearch.institute/pipermail/ground-station-openresearch.institute/attachments/20210401/20ff23fc/attachment.html>