[PATCH mvebu-dt64] arm64: dts: marvell: armada-37xx: Add DTS file for Turris Mox

[marek.behun@nic.cz (Marek Behun)]

On Thu, 30 Aug 2018 17:57:01 +0200
Andrew Lunn <andrew@lunn.ch> wrote:

> > +	switch0 at 10 {
> > +		compatible = "marvell,mv88e6190";
> > +		reg = <0x10 0>;  
> 
> MDO reg should have a single value, i think.
> 
> > +	switch0 at 2 {
> > +		status = "disabled";  
> ...
> > +	};
> > +
> > +	switch1 at 11 {
> > +		status = "disabled";  
> ...
> > +	};
> > +
> > +	switch1 at 2 {
> > +		status = "disabled";  
> ...
> > +	};
> > +
> > +	switch2 at 12 {
> > +		status = "disabled";  
> ...
> > +	};
> > +
> > +	switch2 at 2 {
> > +		status = "disabled";  
> ...
> > +	};
> > +};  
> 
> I'm assuming the boatloader is probing the hardware, and then enabling
> what it finds? This is a poor mans DT fragments. Lets see that Rob has
> to say about this.

The bootloader can read which hardware is connected via a SPI shift
register. Three switch modules can be connected in DSA, in these
possibilities:
  6190
  6190-6190
  6190-6190-6190
  6141
  6190-6141
  6190-6190-6141
At first I had the Linux's device tree completely without the switch
nodes, and I generated the nodes via libfdt in u-boot, but the code was
terrible to read. This way u-boot's code is nice and readable.

> 
> > +	moxtet at 1 {
> > +		#address-cells = <1>;
> > +		#size-cells = <0>;
> > +		compatible = "cznic,moxtet";
> > +		reg = <1>;
> > +		devrst-gpio = <&gpiosb 2 GPIO_ACTIVE_LOW>;
> > +		spi-max-frequency = <1000000>;
> > +		spi-cpol;
> > +		spi-cpha;
> > +
> > +		moxtet_sfp: moxtet-sfp at 0 {
> > +			compatible = "cznic,moxtet-gpio";
> > +			gpio-controller;
> > +			#gpio-cells = <2>;
> > +			reg = <0>;
> > +			moxtet,id = <1>;
> > +			moxtet,input-mask = <0x7>;
> > +			moxtet,output-mask = <0x3>;
> > +			status = "disabled";
> > +		};
> > +	};
> > +};  
> 
> Have the drivers for this been merged yet?

No, I sent the patches today as this one.

> 
>      Andrew