[Lenny400] L4Re/Fiasco.OC on the Letux 400

[Paul Boddie]

Hello,

I haven't been able to direct any real attention to the Letux 400 for quite 
some time, but recently I managed to port the L4 Runtime Environment and 
Fiasco.OC microkernel to the Ben NanoNote, most of the work already having 
been done by the L4Re/Fiasco.OC developers and some people who ported these 
things to the MIPS architecture.

Here is a summary of that work, with a collection of articles for anyone 
*really* interested in the details:

https://blogs.fsfe.org/pboddie/?p=2147

Having done this, it did occur to me that I could give the Letux 400 some 
attention in this regard. As everyone reading this will know, the Ben uses the 
JZ4720 (a JZ4740 variant) whereas the Letux uses the mysterious JZ4730. But 
since we gained some experience with "modernising" Linux support, David with 
his "evolved" kernel and myself with some attempts to test functionality 
within the old kernel and then in making a tentative modern kernel (which I 
still haven't tested), it seemed reasonable to see what could be done.

To keep this message short, I can report that I did indeed get Fiasco.OC to 
boot on the Letux, and I have written L4Re support for the CPM (clock and 
power management), GPIO, LCD and PWM peripherals. The LCD support works within 
the driver framework used by the framebuffer driver, and so framebuffer 
programs can apparently just use the display without any special effort.

I have made my patches against the L4Re/Fiasco.OC repository available here:

http://www.boddie.org.uk/paul/L4Re-Fiasco.OC.html

You may wonder what this has to do with these lists, particularly the lenny400 
list, which is largely concerned with Debian root filesystems and suitable 
Linux kernels. And until I send this, I don't even know if the mipsbook-devel 
list is even functioning any more.

I would like to think that this opens another door to modern software 
offerings for the Letux 400, even though work is needed to provide support for 
other aspects of the hardware and to construct a suitable software framework 
for "normal" use of the device. I intend to investigate things like Genode for 
the latter task:

http://genode.org/

Meanwhile, I think that these technologies are also relevant for other Letux 
products. It might be interesting to consider running something like QtMoko on 
top of a software framework rooted in these technologies. Qt is apparently 
supported by Genode, and so it might not be so absurd to consider a relatively 
lightweight environment combining the two.

Anyway, I hope this is of interest to someone out there!

Paul