Creating a new ARM/AVR platform

Icarus Alive icarus.alive at
Thu Mar 1 03:43:12 UTC 2012

On Thu, Mar 1, 2012 at 3:25 AM, Liam Proven <lproven at> wrote:

> On 29 February 2012 18:48, Jacob Mansfield <cyberjacob at> wrote:
> >
> > I can't thank you all enough for the help you're providing with this!
> You're welcome. :¬)
> > On 29 Feb 2012, at 18:00, Liam Proven wrote:
> >
> > Although if you do want to do something "different", here is a VERY
> > small ARM system:
> >
> >
> >
> > There's going low-powered, and there's going no-powered.
> The CuBox is /dramatically/ higher-specced and significantly more
> powerful than the Rπ. It has a faster CPU, 4× the RAM, on-board
> gigabit Ethernet and SATA. It is probably around the power of a
> low-end Atom system, for perhaps one-tenth of the electrical draw and
> heat dissipation and 1% of the space.
> > Or use Raspberry Pi, of course. ;¬)
> >
> > that's becoming a serious option for us at the moment
> They won't be available for weeks yet - the first run of 10,000 units
> sold out in about three minutes this morning - and are fairly poorly
> specified. No SATA, the Ethernet is over USB, and proprietary blobs
> for the bootloader and the graphics drivers. Not ideal for FOSS
> development, sadly. But then, it's meant as a very cheap educational
> tool for schoolkids.

Indeed. A year back it was hard to gauge the kind of interest RaspberryPi
is going to evoke, but looks like the commercial
put-things-together-to-make community is possibly more excited about it, as
an ultra-economical alternative. I mean, at the price of an Arduino, you
get something that runs Linux, and is possibly almost halfway there, where
Beaglebone is, or has been.

As someone who had been watching out for RasPi announcements at 6:00AM GMT
yesterday, and yet one of the several thousands, who only finally got to
"express interest", it was phenomenal.

 > at the moment, this is looking like a very plausible solution. However
> > none of our team know where to even begin with writing a custom driver
> for
> > linux, is there documentation for such a procedure?
> [...]
> > All of our team have experience with both microcontrollers and
> > microprocessors.
> I think if you're up to that sort of thing, a Linux device driver or
> two won't be a massive challenge for you. :¬)
Actually, depending on what the custom hardware is, and how the x86 box
communicates with it, one could pull it off without having to write a
driver at all. For example, if Serial-over-USB is used, say for instance
with Arduino, or one of the AVR USB uC, or PIC, or TI's MSP430 line, the
silicon manf's or the ecosystem already has the Linux drivers, and one can
make do with purely user-land application.

Writing a well-behaved driver is not very trivial, and personally I think
not all embedded developers, who do not have prior experience in writing
Linux drivers can pull it off so easily. Yeah, following examples of other
drivers, one can write "a" driver, but whether or not, it turns out to be
well-behaved, is really another thing.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <>

More information about the ubuntu-users mailing list