[U-Boot] Porting u-boot to a new soc, unsure of standards to follow.

[Marty Plummer]

No comments?

On Sat, Jan 7, 2017 at 12:04 AM, Marty Plummer <netz.kernel at gmail.com> wrote:
> Greetings,
>
> So, out of a desire to learn (or simply pure masochism), I've taken it
> upon myself to work on mainlining support for a new arm SoC and board,
> which the vendor SDK only provides source for release 2010.06 (which I
> had to find via taobao since the vendor in question is not being nice
> in regards to providing gpl source code from their sdks, but this is
> besides the point).
>
> Having already cobbled together devicetree files for the SoC which has
> allowed me to boot mainline linux from Linus' tree using the u-boot on
> the board, I figured I'd take a go at using devicetree in u-boot itself
> to describe the hardware. Asking about this on irc, I was pointed at the
> firefly-rk3288 as an example of how to do this. Referencing the source
> code and the datasheet in question, I started to wrap my head around
> bootrom, spl, u-boot, and their interactions, and began the task of
> defining the various system controller registers, using the rockhip code
> as a guide.
>
> However, upon mentioning the structs on irc, I was asked by Marex to not
> use this method, instead using macros and hex offsets. At this point
> I've become a bit confused as to how I should proceed. The coding style
> guide[1] says one should use "a C structure to map out the registers in
> an I/O region", which appears to be what the rk3288 code does, but is at
> odds with Marex's request and instruction to use macros.
>
> So, I'm asking the ml and whatever powers that be how I should proceed,
> as I'd really hate to write a bunch of code that has to be changed to
> such a degree that it may as well be discarded and started from scratch.
>
> Regards,
>         Marty Plummer
>
> [1] http://www.denx.de/wiki/U-Boot/CodingStyle