Subject: Summer of code: omap5912, C6416, suggestions?
To: None <tech-ports@netbsd.org>
From: Andrew S. Gardner <andrew.scott.gardner@gmail.com>
List: tech-ports
Date: 04/18/2006 12:04:24
Hello tech-ports and port-arm.  I'm Andrew Gardner, and I would like
your opinions on two different projects for Google's Summer of Code.

Project 1 is adding support for the Spectrum Digital OMAP5912 Starter
Kit to evbarm.
http://www.spectrumdigital.com/product_info.php?cPath=3D31_80&products_id=
=3D39

Judging from traffic on the lists, NetBSD has been ported to a number
of OMAP platforms, but as far as I can tell, the code was developed
under a contract that would not permit for the release of the
software.  This is a relatively straight-forward project:  setup cross
compilation tool chain, grab documents from TI, identify where/how to
modify kernel's expected memory map, etc.  Since the OSK comes with a
setup and instructions for running Linux on the board, lots of
logistical problems will be lessened.  A possible follow-on would be
to get the DSP/LINK software running on NetBSD so that C55 DSP
applications could be developed and loaded.

Project 2 is much more ambitious:  porting NetBSD to the TI C64 series
of DSPs, specifically the C6455 DSP Starter Kit.
http://focus.ti.com/docs/toolsw/folders/print/tmdxdsk6455.html

This is a more ambitious project for a number of reasons:  no GNU
toolchain exists; the TI toolchain runs on Windows (and possibly Linux
(and then thanks to binary compatibility NetBSD) but details are hard
to find); C6455 has some memory protection, but it isn't the same sort
of memory protection that you get in a real MMU/virtual memory setup;
and, every time that someone brings up MMU-less systems for NetBSD, a
consensus doesn't seem to develop about the proper manner in which to
approach the various problems that not having an MMU will cause.  I
think the key here is to clearly define the deliverables and make
realistic expectations about how much work is going to be done.

I will make clearer descriptions of project scope and deliverables in
a formal proposal, but I'd like to get your opinions about these two
projects before I submit a formal proposal.

Andrew Gardner

(Quick background on me:  I finished my undergrad studies in
electrical and computer engineering at Carnegie Mellon in May of 2001.
 I'm currently a part time grad student at the University of Arizona
in Tucson.  I work a full time job as a software engineer, splitting
my time between real-time signal processing applications written in C
and assembly for C6201 and C6416 DSPs, and hardware-in-the-loop
simulations in a mix of OpenGL/Cg/C/C++/Perl for SGI IRIX and Linux.)