Milkymist One = http://milkymist.org/wiki/index.php?title=Main_Page (software) http://en.qi-hardware.com/wiki/Milkymist_One (hardware) - Runs GNU/Linux as well as copyleft hardware design and a copyleft CPU on FPGA.
If interested please reply-to only me (and if you'd like one mailing list) to avoid spamming people / mailing list cross-posting issues. If you know of any other people or mailing lists that might have interested parties, please forward.
---------- Forwarded message ----------
From: Wolfgang Spraul <wolfgang@sharism.cc>
I so totally agree with you on the general purpose computer! In
fact, it is already. We are spending tremendous efforts to get all
the little pieces together, and they are coming together, slowly.
We have Linux and OpenWrt booting on M1 today, although it's really
alpha and pioneering stuff. Pretty much nothing works :-)
http://fidelio.qi-hardware.com/~xiangfu/build-milkymist/milkymist-openwrt.minimal-09162011-0423/
(as you see - last build Sept 2011)
What you can do is just email the list that you want to boot Linux
on M1 and use it as a general purpose computer. That will trigger
some support action then and after some time more people become
aware of this possibility.
The biggest blocker right now is the lack of an MMU for the Milkymist
SoC (the free IC design powering the computer).
Because we have no MMU, the Linux kernel effort is really cornered
without hope for upstream inclusion, or an effective Linux rootfs.
https://github.com/milkymist/linux-milkymist
(last update: Nov 23, 2011)
We have made some design planning for the MMU on Milkymist, and probably
it can done in a few hundred or at most 1-2 thousand lines of Verilog,
but... There are few people with experience in that area, and
willingness to publish their work under a free license (my Verilog
book is lying right besides me here :-))
Here was the latest planning status for the MMU
http://lists.milkymist.org/pipermail/devel-milkymist.org/2011-April/001418.html
http://lists.milkymist.org/pipermail/devel-milkymist.org/2011-April/001472.html
Maybe you can hang a big sign at pika "Milkymist needs a MMU" :-)
Once the Milkymist SoC has an MMU, the Linux kernel effort can be
continued, and eventually we can move over the Flickernoise app to
be just another application running on Linux, rather than RTEMS
right now. Can you push out some "please develop a MMU for Milkymist"
action?
After the MMU other roadblocks for the general purpose computer are:
*) the Linux kernel supports very few peripherals of M1 today
*) no C++ compiler (only C), in fact recent gcc versions 4.6 or newer
are broken, and few gcc people care about Milkymist (lm32)
*) no dynamic linking
We are also working hard on the hardware to prepare for a great
future as a general purpose computer. For example we just fixed bugs
that prevented USB full-speed from working (not high-speed or even
super-speed - we are not there). Now we are planning to increase from
2 USB connectors to 6, including 2 internal ones. Then we can slowly
attack the RF area, some things are already moving (GPS), and others in
motion (osmoSDR).
Finally we will overhaul the current expansion header and turn it into
a more refined expansion system, so that multiple people can develop
and use daughterboards in parallel, something quite important to not
be stuck in one particular use case.
Yeah, that's about it :-)
Hope this makes sense, Cheers,
Wolfgang
--
\|/ Daniel JB Clark | Activist; Owner
FREEDOM -+-> INCLUDED ~ http://freedomincluded.com
/|\ Free Software respecting hardware