[PATCH RFC 0/5] ARM: dts: zynq: pinctrl, LED, USB for Parallella

[Sören Brinkmann]

On Thu, 2015-02-12 at 01:55AM +0100, Andreas Färber wrote:
> Hello,
> 
> This series is based on linux-next plus Michal's aliases patch.
> https://patchwork.kernel.org/patch/5812541/
> 
> Patch 2 adds pinctrl for the Parallella board. No such information was 
> visible downstream, so this is based on Sören's patch and the schematic 
> plus some testing. In particular the Ethernet voltage level seems to be 
> different from the other Zynq boards upstream (io-standard 1 vs. 4).
> 
> Patch 3 adds an LED. This is a feature not present downstream. No 
> default trigger is thus used and I chose the label scheme used by sunxi 
> and omap (zynq-zc702.dts uses just "ds23").
> Tested as follows:
> echo mmc0 > /sys/class/leds/parallella:cr10:usr/trigger #or s/mmc0/heartbeat/
> (on openSUSE Tumbleweed I appended this to /etc/init.d/boot.local script)
> 
> Patch 5 goes on to add USB support. As the Microserver edition does not 
> have USB, the bulk of nodes is moved to a new .dtsi file in patch 1.
> 
> Note: Adapteva's downstream tree has in the meantime split zynq-parallella.dts 
> into zynq-parallella1-{hdmi,headless}.dts (without discussing this here first).
> Since zynq-parallella.dts has already been released, I have chosen to 
> stay with that name and have therefore named the Microserver edition 
> zynq-parallella-microserver.dts, while naming the shared file 
> zynq-parallella1.dtsi.
> Anyone prefer to name the new file zynq-parallella1-microserver.dts?
> 
> On my work branch https://github.com/afaerber/linux/commits/parallella-next
> I have further prepared a zynq-parallella-headless.dts variant 
> following this scheme. Depending on how the Epiphany driver evolves, we 
> may further need to split the non-Microserver variant(s) into e16 vs. e64
> variants to cope with some Kickstarter boards.
> 
> Is there some rule for device tree file name stability when splitting a 
> file due to additional nodes?
> 
> Example of historic growth:
>  zynq-parallella.dtb (today works on all boards) -- now with USB/HDMI
>  zynq-parallella-microserver.dtb (this series) -- without USB/HDMI
>  zynq-parallella-headless.dtb -- with USB/HDMI, but no FPGA HDMI support
>  zynq-parallella-e64.dtb -- with Epiphany-IV instead of Epiphany-III
>  zynq-parallella-e64-headless.dtb -- ditto, but different FPGA bitstream
>  zynq-parallella2-*.dtb
> vs. extreme distinction:
>  zynq-parallella1-e16-hdmi.dtb
>  zynq-parallella1-e16-headless.dtb
>  zynq-parallella1-e16-microserver.dtb (doesn't fully fit the scheme...)
>  zynq-parallella1-e64-hdmi.dtb
>  zynq-parallella1-e64-headless.dtb
>  zynq-parallella2-*.dtb
> In the former case, the name remains stable for capable boards, but using 
> it on lesser capable boards may break. In the latter case, users of all 
> boards need to choose a more specific file whenever incompatible changes 
> are made. Since no effort has been made to provide users with a smooth
> downstream-to-upstream path, I guess we can at least neglect that scenario,
> already having diverged by now.

I'm just thinking out loud here. I think it's reasonable to have a dts
for each platform/board. On dtsi files the opinions seem to diverge,
but as long as it isn't a completely confusing include scheme,
I'm always for minimizing duplication.
But what I'm not convinced of is, additional dts files for different
bitstreams. The amount of variation you can achieve by programming
different bitstreams is virtually unlimited. Having a dts file for all
of that upstream doesn't seem that reasonable to me.

So, I'd say it needs parallella-{userver, e16, e64} (conceptually, the
names I don't care). I think it should be possible for all of those to
inherit from zynq-7000.dtsi or is there a reason to have a parallella
dtsi?

	Sören