[PATCH v4 0/9] ARM: Initial support for Marvell Berlin SoCs

[Olof Johansson]

On Tue, Dec 10, 2013 at 02:57:05AM +0100, Sebastian Hesselbarth wrote:
> On 12/10/2013 02:40 AM, Olof Johansson wrote:
> >On Sun, Dec 08, 2013 at 03:13:57PM +0100, Sebastian Hesselbarth wrote:
> >>Hopefully last round of initial support patches for Marvell Berlin SoCs
> >>before I can send the final PR for v3.14.
> >>
> >>Compared to last version sent, this patch set has now a Reviewed-by from
> >>Thomas Gleixner for the irqchip driver (Thanks for that!). Also, l2x0
> >>compatibles can now be reordered alphabetically instead of by derivate
> >>thanks to [1].
> >>
> >>Marvell Docs have been updated to not mention Armada 1000 which has been
> >>discontinued by Marvell and vanished from their website. The dtsi/dts file
> >>have been renamed to vendor,name.dts[i], which is the preferred new naming
> >>scheme.
> >>
> >>Open issues are the never ending dw_apb_timers_of story, which I ignore
> >>for now and hope they get in someday. Also, TWD SMP dependency and
> >>early l2x0_of_init will be addressed at a later date. At the current
> >>feature set of Berlin SoC, I don't see why the above issues should further
> >>stall this patches.
> >>
> >>I guess, all patches can go through ARM SoC tree, except Tauros3 patch
> >>which I should submit to Russell's patch tracker?
> >
> >Yep, sounds good.
> >
> >I took a cursory glance at the patchset and it looks sane to me. I didn't
> >review in detail though.
> >
> >One open question: Why don't you just add this to mach-mvebu? I thought
> >all modern Marvell platforms were going to converge on that eventually
> >anyway, and it's easy to add it now that it's early and simple..
> 
> Olof,
> 
> I started with mach-mvebu in the first RFC, but Berlin SoCs are from a
> different business unit at Marvell and are more PXA compatible than
> Orion/Kirkwood/Armada 370/XP. Most notably, they lack the "mbus" and
> IP is either from PXA/MMP or Synopsys DW. Thomas Petazzoni and also
> Jisheng Zhang from Marvell suggested to not put it into mvebu but
> have a different mach folder instead.

Hmm, ok. Well, maybe later on we can look at aggregating them more again. It'd
be nice to get fewer top-level platform directories per vendor.

> >Also, see below:
> >
> >>Sebastian Hesselbarth (9):
> >>   irqchip: add DesignWare APB ICTL interrupt controller
> >>   MAINTAINERS: add ARM Marvell Berlin SoC
> >>   ARM: l2x0: add Marvell Tauros3 support
> >>   ARM: add Marvell Berlin SoC familiy to Marvell doc
> >>   ARM: add Marvell Berlin SoCs to multi_v7_defconfig
> >>   ARM: add Marvell Berlin UART0 lowlevel debug
> >>   ARM: add Armada 1500 and Sony NSZ-GS7 device tree files
> >>   ARM: add Armada 1500-mini and Chromecast device tree files
> >>   ARM: add initial support for Marvell Berlin SoCs
> >>
> >>  Documentation/arm/Marvell/README                   |  24 +++
> >>  Documentation/devicetree/bindings/arm/l2cc.txt     |  23 ++-
> >>  .../devicetree/bindings/arm/marvell,berlin.txt     |  24 +++
> >>  .../interrupt-controller/snps,dw-apb-ictl.txt      |  32 +++
> >>  MAINTAINERS                                        |   6 +
> >>  arch/arm/Kconfig                                   |   2 +
> >>  arch/arm/Kconfig.debug                             |  10 +
> >>  arch/arm/Makefile                                  |   1 +
> >>  arch/arm/boot/dts/Makefile                         |   3 +
> >>  arch/arm/boot/dts/google,chromecast.dts            |  29 +++
> >>  arch/arm/boot/dts/marvell,berlin2.dtsi             | 227 +++++++++++++++++++++
> >>  arch/arm/boot/dts/marvell,berlin2cd.dtsi           | 210 +++++++++++++++++++
> >>  arch/arm/boot/dts/sony,nsz-gs7.dts                 |  29 +++
> >
> >We have had a long-standing standard of naming the dts files
> ><family>-<board>.dts (or <soc_vendor>-<board>.dts).  Let's continue
> >sticking to that since it helps keep the namespace somewhat segmented per
> >platform in arch/arm/boot/dts.
> 
> Also, here: I had the naming that way until v3, then Kumar suggested to
> use prefixed naming scheme. Maybe I got it wrong?
> 
> Are you suggesting to name the above back to:
> berlin2[cd].dtsi,
> berlin2-nsz-gs7.dts, and
> berlin2cd-chromecast.dts?

Personally I prefer something similar to the mach-directory name as prefix, so
berlin-<soc number>.dts and then berlin-chromecast.dts, etc. But I'm not that
picky, if you have something else you prefer then that's fine too.

At some point we should restructure the dts directory into subdirectories,
which will remove some of these requirements, but we're not there quite yet.


-Olof