Re: [PATCH v13 2/2] arm64: dts: imx: Add jpeg encoder/decoder nodes

On Fri, 2021-06-11 at 21:33 +0800, Dong Aisheng wrote:

[...]

quoted

quoted

quoted

+img_subsys: bus@58000000 {
+   compatible = "simple-bus";
+   #address-cells = <1>;
+   #size-cells = <1>;
+   ranges = <0x58000000 0x0 0x58000000 0x1000000>;
+
+   img_ipg_clk: clock-img-ipg {
+           compatible = "fixed-clock";
+           #clock-cells = <0>;
+           clock-frequency = <200000000>;
+           clock-output-names = "img_ipg_clk";
+   };
+
+   jpegdec: jpegdec@58400000 {

Node should be disabled by default.
And enable it in board dts including LPCG.

At version v5 of this patch, the node was disabled by default, and I
received this feedback from Ezequiel Garcia:

"Pure memory-to-memory are typically not enabled per-board, but just
per-platform.
So you can drop the disabled status here."

So, in v6 I made it enabled by default.

Any strong reasons for enabled/disabled per platform?

AFAIK we usually only enable system basic features and let other
user selectable features disabled by default in dts.
Even for device LPCG clocks, if it's enabled by default and later
enter runtime suspend if no users, it still consumes power.

Well-written drivers shouldn't draw any power if not used.

And DT is about hardware-description, not about usage-description.
Which means, at the soc.dtsi level you disable devices that need
some board-level hardware thing to be enabled (e.g. a physical
connected, a regulator, etc.).

A pure memory-to-memory should be enabled by default, because
in practice you can't predict what the users a board will want
to use, nor the DT is the place for that.

Sticking to hardware description is the best way to get DT right :-)

Cheers,
Ezequiel


_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel