University project - recommendations?

[Gedare Bloom]

On Fri, Nov 28, 2014 at 4:17 PM, Dominik Taborsky <bremby at seznam.cz> wrote:
> Hello,
>
> I am a university student and am looking for a school project. I am
> interested in RTEMS and so I thought I could help out with development.
> The project should take several "school months". It should also involve
> the kernel space, not user space.
>
Welcome!

What areas are you most interested in working on? Are you interested
in something that might continue beyond your project? Do you want to
be more into core kernel development, hardware abstraction,
"middleware" (networks, drivers, etc)? Based on the three projects you
posted, I'm guessing the latter, something that operates between the
kernel and the hardware.

What is your expected hours/wk to spend on this? That will help in
gauging the level of effort you should undertake. GSoC-size projects
aim at full-time (40 hours) effort for 2 months AFTER getting through
the "getting started" and defining project requirements and
milestones. Keep in mind that many of the open projects we defined are
aimed at GSoC size or greater. You'll probably need to carve off a
smaller-size project from a larger one to fit your time constraints.
You should also do the "Getting Started" with the RTEMS Source Builder
to start getting comfortable with the tools. Ask if you need help, or
especially if you run into broken links. Our website has recently
migrated and some things broke.

> I have browsed through the wiki/trac and I found these 3:
> 1) CFI-standard flash device interface,
> 2) CEXP integration,
> 3) TCP stack rewrite.
>
I don't know that CEXP integration is that interesting anymore. One of
the key features of CEXP is dynamic loading, which is not supported
through the RTEMS linker and loader projects (RTL). You might ask
Chris Johns if there are projects available for RTL.

The TCP stack has mostly been refreshed from BSD by now. I'm not sure
about its status.

Improving the support for devices and especially frameworks for
drivers is a timely project with good potential. See the recent
commits that added cpukit/dev/i2c. You might try writing some drivers
for i2c devices to fit the new framework. Sebastian Huber may have
more info for you. The difficult part in this is testing.

We had a GSoC student (Andre Marques) look at some I/O device
frameworks as part of the raspberrypi effort too, so you might look
for those conversations or contact him directly.

> I don't have any experience with implementing TCP, but I have some
> experience with data structures on block devices (manipulating MBR and GPT
> labels). But I don't know how much of help my experience is.
>
> I have hard time guessing time requirements for these, so can anyone give
> me a hint how complex these are? Or can you give me any other project
> assignments?
>
Estimating time is hard! Especially without much context for your
availability or skills. :) In general, I'd suggest scoping out a
project with many smaller milestones so that you can make good
progress, even if you don't get to the end, you should have a nice set
of accomplishments.

Gedare

> Any input is welcome.
>
> Best regards,
> Dominik Taborsky
> _______________________________________________
> devel mailing list
> devel at rtems.org
> http://lists.rtems.org/mailman/listinfo/devel