Re: GSOC - Interested in making u-boot compilable on NetBSD

[Joe Davis <joe.davis512%gmail.com@localhost>]

On 12 March 2016 at 16:43, Christos Zoulas <christos%astron.com@localhost> wrote:

> This is the right place. U-boot can be used to boot many different
> machines, so you should choose a machine/architecture you have access to
> in order to be able to test. The most useful to us I think would be arm.
> As I mentioned to others before, you should write a proposal and split
> down the project into manageable chunks with intermediate deliverables
> so that you can measure your progress and judge if you are ahead or behind.
>
> christos
>

I've appended a quick draft of the proposal in markdown format, and
I've made an attempt to split the project up into intermediate stages.
I'd appreciate any comments, especially with regard to the timeframe.
Having ported a C library (newlib) to a homebrew kernel in the past,
the initial 4-weeks for getting the *preliminaries* up and running
seems feasible to me (i.e. having it build without issue), but to be
quite honest I haven't before worked on a project of this scale, so I
could very well be underestimating.

Thanks in advance,

Joe

Draft:

> # U-Boot on NetBSD
>
> ## Goal
>
> Currently using U-Boot on NetBSD requires Linux as an intermediate step,
> in order to build it. My goal is to fully port U-Boot, allowing it to be
> built and run solely using NetBSD.
>
> ## Deliverables
>
>  1. Enable U-Boot to successfully build on NetBSD, at first using GNU
>     make.
>
>  2. Reproducibly and reliably boot a full NetBSD system using U-Boot
>     from an emulated x86 system under QEMU.
>     - This is faster to test than actual hardware, and may allow for
>       easier compatibility of NetBSD with coreboot in the future. It
>       also has the advantage of sidestepping any issues with cross
>       compilation that may occur.
>
>  3. Reproducibly and reliably boot a full NetBSD system from a natively
>     built U-Boot on a Beaglebone Black (ARMv7 system)
>      - U-Boot is far more useful on architectures other than x86, and
>
>  4. If possible, allow U-Boot to build using BSD Make
>     - The U-Boot makefile and build system is based heavily around
>       features and syntax specific to GNU make. While building with
>       BSD-make would be far more convenient, it may not be feasible from
>       a maintenance perspective.
>     - It will likely be necessary to create a wrapper around the
>       existing build system, in order to integrate U-Boot with pkgsrc.
>
> ## Timeline
>
> During the initial week of GSOC, I will be completing my end-of-year
> exams, and won't be available for work on this project. This leaves me
> with 13 weeks during which I can work full time.
>
> ### Week 1
>
> Exam week. Unavailable.
>
> ### Weeks 2-5
>
>  * Have U-Boot successfully build on x86, and begin preliminary work on
>    ARM.
>  * The majority of this work should deliverable by the mid-term
>    evaluation period, 20^th^--27^th^ June.
>
> ### Weeks 6-10
>
>  * Complete any remaining work on x86, and begin testing U-Boot on
>    physical ARM hardware, in this case the Beaglebone Black.
>
> ### Weeks 11-13
>
>  * The final two weeks of this project will involve creating a pkgsrc
>    entry for U-Boot.
>
> ## About me
>
> I'm an Irish student, currently in my first year of Theoretical Physics
> at Trinity College Dublin. I'm a self taught programmer, and quite
> comfortable with C and assembly, having a fair amount of experience
> developing kernels and bootloaders as a hobby.
>
> I'm experienced with \*nix systems in general, and have previously
> worked a summer as a maintenance programmer for a small domain registrar
> and hosting service. As well as that, at the age of 16 I completed a
> week long work experience placement at Symantec, most of which was spent
> learning the basics of reverse engineering.
>