[U-Boot] [PATCH 0/2] Add Rock960 board support

[Manivannan Sadhasivam]

On Fri, Aug 31, 2018 at 09:27:25AM -0300, Ezequiel Garcia wrote:
> On 31 August 2018 at 09:24, Manivannan Sadhasivam
> <manivannan.sadhasivam at linaro.org> wrote:
> > On Fri, Aug 31, 2018 at 09:08:08AM -0300, Ezequiel Garcia wrote:
> >> On 31 August 2018 at 07:56, Peter Robinson <pbrobinson at gmail.com> wrote:
> >> >> On Thu, Aug 30, 2018 at 12:44:00AM -0300, Ezequiel Garcia wrote:
> >> >> > On 30 August 2018 at 00:00, Manivannan Sadhasivam
> >> >> > <manivannan.sadhasivam at linaro.org> wrote:
> >> >> > > Hi Ezequiel,
> >> >> > >
> >> >> > > On Wed, Aug 29, 2018 at 03:11:17AM -0300, Ezequiel Garcia wrote:
> >> >> > >> Hi Manivannan,
> >> >> > >>
> >> >> > >> On 21 August 2018 at 14:09, Manivannan Sadhasivam
> >> >> > >> <manivannan.sadhasivam at linaro.org> wrote:
> >> >> > >> > This patchset adds board support for Vamrs Limited Rock960,
> >> >> > >> > which is one of the 96Boards Consumer Edition platform based
> >> >> > >> > on Rockchip RK3399 SoC.
> >> >> > >> >
> >> >> > >>
> >> >> > >> What are the differences between this consumer edition board,
> >> >> > >> and the enterprise edition (aka Ficus) Vamrs board?
> >> >> > >>
> >> >> > >
> >> >> > > I asked Vamrs about this and they said the difference is very minimal.
> >> >> > >
> >> >> >
> >> >> > In that case, you should try to leverage the Linux ficus.dts:
> >> >> >
> >> >> > https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/arch/arm64/boot/dts/rockchip/rk3399-ficus.dts
> >> >> >
> >> >> > If no differences, using the ficus dts should do. If there are differences,
> >> >> > we can create a common rock960.dtsi and then enterprise and consumer
> >> >> > edition dts.
> >> >> >
> >> >>
> >> >> Okay. Here are the differences between Ficus and Rock960 CE:
> >> >>
> >> >> 1. Different host enable GPIO for USB (vcc5v0_host)
> >> >> 2. Different power and reset for PCI-E (vcc3v3_pcie, pcie0)
> >> >> 3. No Ethernet port on Rock960 (gmac)
> >> >>
> >> >> So, I would suggest keeping USB, PCI-E and GMAC related nodes on the board
> >> >> specific devicetree and rest on the rk3399-rock960.dtsi. What do you think?
> >> >>
> >> >> Same applies to Linux also!
> >> >
> >>
> >> Sounds good. If you have some cycles to work on the dts/dtsi split,
> >> that would be great.
> >>
> >
> > Sure, will do it for Linux now. Once your u-boot patches gets in,
> > will tackle it also.
> >
> 
> Well, I think we can tackle u-boot from scratch. No need to
> merge my patches if we think they are already wrong :-)
>

Makes sense!

> >> > Yes, I think a rk3399-rock960.dtsi with the differences in two .dts
> >> > would be great, and even better a single U-Boot which can detect the
> >> > board and load the right DT, but I do think it should be a separate
> >> > config to evb as Mani mentioned.
> >> >
> >
> > Thanks Peter for your thoughts!
> >
> >>
> >> Sounds good too. That'd make board-specific hooks easier.
> >>
> >> On the other side, the documentation should be
> >> merged somewhere.
> >>
> >> Seems nonsense to have a README per board,
> >> with more or less the same instructions each time.
> >>
> >
> > This has other side as well. If we continue to merge board specific
> > instructions onto evb-rk3399, it will become messy. So, IMO it's better
> > to have it separate. If you have other ideas, please let me know.
> >
> 
> Bootloader wise, there is no such thing as board specific instructions, is it?

Yeah, but I'm thinking about the overhead of having a common doc for
boards which might come in future (like Rock960c have an optional eMMC
module).

Even in Rock960 CE, we need to specify the load address for u-boot while
preparing the uboot.img (not sure why it is required)...

I'm kind of skeptical here.

Peter, any thoughts?

> -- 
> Ezequiel García, VanguardiaSur
> www.vanguardiasur.com.ar