On Mon, Jan 30, 2017 at 01:42:40AM +0000, Andr? Przywara wrote:

quoted

+&ccu {
+	compatible = "allwinner,sun8i-h3-ccu";
+};

I believe this kind of sharing nodes is a bit frowned upon in connection
with sharing .dtsi's. If the compatible name differs, I think it
deserves to be a separate node spelt out in each SoC's .dtsi.
This also makes the DT more readable, since a reader doesn't have to
refer to two files to see what's in that node.

quoted

 
-		codec_analog: codec-analog at 01f015c0 {
-			compatible = "allwinner,sun8i-h3-codec-analog";
-			reg = <0x01f015c0 0x4>;
-		};
+&mmc0 {
+	compatible = "allwinner,sun7i-a20-mmc";
+	clocks = <&ccu CLK_BUS_MMC0>,
+		 <&ccu CLK_MMC0>,
+		 <&ccu CLK_MMC0_OUTPUT>,
+		 <&ccu CLK_MMC0_SAMPLE>;
+	clock-names = "ahb",
+		      "mmc",
+		      "output",
+		      "sample";

This applies even more here, since the MMC controllers also have
different clock requirements.

So why can't we just leave the CCU, MMC and possibly the pinctrl nodes
completely out of the shared h3-h5.dtsi and introduce them from scratch
in the SoC specific .dtsi?

I think we still have enough identical nodes to justify this kind of
.dtsi sharing.

We did it that way in the past in order to reduce the unneeded
duplication, but I can definitely understand your point. We'll wait
for the DT maintainers answer on this one.

Thanks,
Maxime

-- 
Maxime Ripard, Free Electrons
Embedded Linux and Kernel engineering
http://free-electrons.com
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 801 bytes
Desc: not available
URL: <http://lists.infradead.org/pipermail/linux-arm-kernel/attachments/20170130/cafa1483/attachment.sig>