[PATCH v2 5/7] arm64: dts: Add initial device tree support for EXYNOS7

Arnd Bergmann wrote:

On Thursday 04 September 2014 01:31:21 Kukjin Kim wrote:

quoted

quoted

This is the board specific file, so it seems ok.

I mean there are many espresso boards are having different exynos7
SoC. I mean exynos7-espresso cannot represent all of espresso boards.

Ah, I see, that makes sense.

We definitely need a top-level .dts file for each board that is different
in a nondiscoverable way then.

Agreed.

And one more, at this moment just using exynos7 would be nice to us then if
required, we could add any specific properties later.

If the mmc settings and possibly some other nodes (to be added later)
are common across them, those can be in a board specific .dtsi file.

Yes, right.

For the memory node, I would actually expect that to be filled by
the boot loader, so we could leave it out entirely. The same applies
to the command line: the parts that are in there at the moment (
"console=ttySAC2,115200n8 root=/dev/ram0 ramdisk=16384 initrd=0x42000000,16M")
clearly don't belong into a generic dts file and none of them should be
set that way.

Hmm...yeah we need to use proper boot-loader but I think kernel needs having
default memory property if boot loader doesn't have anything?

For the initial version, that would mean that the file comes down to having
as its only contents

/dts-v1/;
#include <dt-bindings/gpio/gpio.h>
#include "exynos7123.dtsi"
#include "exynos7-espresso.dtsi" // for the mmc nodes

/ {
       model = "Samsung ESPRESSO board based on EXYNOS7123";
       compatible = "samsung,espresso", "samsung,exynos7123", "samsung,exynos7";
};

One more thing I just noticed is the GPL statement in the dts files.
Can we please change that to a permissive license in order to allow
including it in non-GPL boot loaders and operating systems?

Dual GPL+MIT or GPL+BSD would make most sense here.

I need to check about that with license guy in my team ;)

Naveen, I'll have a look at the patch again.

Thanks,
Kukjin