On Wed, Dec 15, 2021 at 05:52:24PM +0300, Dmitry Osipenko wrote:

15.12.2021 17:01, Thierry Reding пишет:

quoted

On Sun, Dec 12, 2021 at 12:13:55AM +0300, Dmitry Osipenko wrote:
[...]

quoted

+		i2c@1 {
+			reg = <1>;
+			#address-cells = <1>;
+			#size-cells = <0>;
+
+			dsi-bridge@7 {
+				compatible = "toshiba,tc358768";
+				reg = <0x7>;
+
+				#address-cells = <1>;
+				#size-cells = <0>;
+
+				clocks = <&tc358768_osc>;
+				clock-names = "refclk";
+
+				reset-gpios = <&gpio TEGRA_GPIO(N, 6) GPIO_ACTIVE_LOW>;
+
+				vddc-supply = <&vdd_1v2_mipi>;
+				vddio-supply = <&vdd_1v8_vio>;
+				vddmipi-supply = <&vdd_1v2_mipi>;
+
+				ports {
+					#address-cells = <1>;
+					#size-cells = <0>;
+
+					port@0 {
+						reg = <0>;
+
+						bridge_input: endpoint {
+							remote-endpoint = <&dpi_output>;
+							data-lines = <24>;
+						};
+					};
+
+					port@1 {
+						reg = <1>;
+
+						bridge_output: endpoint {
+							remote-endpoint = <&panel_input>;
+						};
+					};
+				};
+
+				/*
+				 * Panasonic VVX10F004B00 or HYDIS HV101WU1-1E1
+				 * LCD SuperIPS+ Full HD panel.
+				 */
+				panel@1 {
+					compatible = "panasonic,vvx10f004b00";
+					reg = <1>;
+
+					power-supply = <&vdd_pnl>;
+					backlight = <&backlight>;
+
+					port {
+						panel_input: endpoint {
+							remote-endpoint = <&bridge_output>;
+						};
+					};
+				};

make dtbs_check complains about this and says that panel@1 (as well as
#address-cells and #size-cells) are not allowed here. And indeed the
binding for the Toshiba bridge doesn't mention them here.

Do we need this here or should this be moved to the top level to fix
those warnings? I guess what you're doing above is describe a DSI bus
created by the DSI bridge, which also makes sense, so another
alternative would be to fix up the binding and let it accept those
properties.

Toshiba bridge binding is incomplete. David has patch for that [1], I
don't think that it was sent out yet.

[1]
https://github.com/okias/linux/commit/0875230062294b6db17f395ced0a8384a4c1cfc7

Okay, please make sure this finds its way upstream eventually. That
patch looks quite similar to what I tried to do to fix this up locally.

Thierry