Re: [PATCH 5/5] serial: sh-sci: Add device tree bindings documentation

[Laurent Pinchart <laurent.pinchart@ideasonboard.com>]

Hi Kumar,

On Thursday 31 October 2013 10:00:36 Kumar Gala wrote:
> On Oct 31, 2013, at 9:55 AM, Laurent Pinchart wrote:
> > On Thursday 31 October 2013 09:42:48 Kumar Gala wrote:
> >> On Oct 29, 2013, at 5:29 AM, Laurent Pinchart wrote:
> >>> Document the device tree bindings for the sci serial port devices.
> >>> 
> >>> Cc: devicetree@vger.kernel.org
> >>> Signed-off-by: Laurent Pinchart
> >>> <laurent.pinchart+renesas@ideasonboard.com>
> >>> ---
> >>> .../bindings/serial/renesas,sci-serial.txt         | 51 ++++++++++++++++
> >>> 1 file changed, 51 insertions(+)
> >>> create mode 100644
> >>> Documentation/devicetree/bindings/serial/renesas,sci-serial.txt
> >>> 
> >>> diff --git
> >>> a/Documentation/devicetree/bindings/serial/renesas,sci-serial.txt
> >>> b/Documentation/devicetree/bindings/serial/renesas,sci-serial.txt new
> >>> file mode 100644
> >>> index 0000000..5658b4f
> >>> --- /dev/null
> >>> +++ b/Documentation/devicetree/bindings/serial/renesas,sci-serial.txt
> >>> @@ -0,0 +1,51 @@
> >>> +* Renesas SH-Mobile Serial Communication Interface
> >>> +
> >>> +Required properties:
> >>> +
> >>> +  - compatible: should be one of the following.
> >> 
> >> Being pedantic, but how about saying 'one of the following types (scif,
> >> scifa, scifb, or hscif)
> > 
> > Sounds good to me.
> > 
> >>> +
> >>> +    - "renesas,scif-r8a7790" for R8A7790 (R-Car H2) SCIF compatible
> >>> UART.
> >>> +    - "renesas,scifa-r8a7790" for R8A7790 (R-Car H2) SCIFA compatible
> >>> UART.
> >>> +    - "renesas,scifb-r8a7790" for R8A7790 (R-Car H2) SCIFB compatible
> >>> UART.
> >>> +    - "renesas,hscif-r8a7790" for R8A7790 (R-Car H2) HSCIF compatible
> >>> UART.
> >>> +    - "renesas,scif-generic" for generic SCIF compatible UART.
> >>> +    - "renesas,scifa-generic" for generic SCIFA compatible UART.
> >>> +    - "renesas,scifb-generic" for generic SCIFB compatible UART.
> >>> +    - "renesas,hscif-generic" for generic HSCIF compatible UART.
> >>> +
> >>> +    When compatible with the generic version, nodes must also list the
> >>> +    SoC-specific version corresponding to the platform.
> >>> +
> >>> +  - reg: Base address and length of the memory resource used by the
> >>> UART.
> >>> +
> >>> +  - interrupt-parent: Reference to the parent interrupt controller.
> >>> +  - interrupts: Interrupt number(s). Depending on the SoC SCIx UARTs
> >>> are
> >>> tied
> >>> +    to one or multiple interrupt lines. When using multiple interrupt
> >>> lines,
> >>> +    specify the interrupt names as described below.
> >>> +
> >>> +  - clocks: Reference to the SCIx UART interface clock.
> >>> +  - clock-names: Should be "sci_ick".
> >>> +
> >>> +Optional properties:
> >>> +
> >>> +  - interrupt-names: When using multiple interrupts, report the
> >>> interrupt
> >>> +    names as "eri" (receive error), "rxi" (receive), "txi" (transmit)
> >>> and
> >>> +    "bri" (break). When using a single interrupt this property should
> >>> not
> >>> be +    present.
> >> 
> >> Hmm, is there a reason not to just have the 4 interrupts always present
> >> and in the case they are all wired to the same interrupt, just have the
> >> interrupts all be the same?
> >> 
> >> example:
> >> 
> >> interrupts = <0 144 4>, <0 144 4>, <0 144 4>, <0 144 4>;
> > 
> > The multiple interrupts case is the exception, I thought it would be
> > easier to specify a single interrupt in the general case and have a more
> > complex binding for the exceptions.
> > 
> >>> +
> >>> +Note: Each enabled SCIx UART should have an alias correctly numbered in
> >>> the +"aliases" node.
> >>> +
> >>> +Example:
> >>> +	aliases {
> >>> +		serial0 = &scifa0;
> >>> +	};
> >>> +
> >>> +	scifa0: serial@e6c40000 {
> >>> +		compatible = "renesas,scifa-r8a7790", "renesas,scifa-generic";
> >>> +		reg = <0 0xe6c40000 0 64>;
> >>> +		interrupt-parent = <&gic>;
> >>> +		interrupts = <0 144 4>;
> >>> +		clocks = <&mstp2_clks 4>;
> >>> +		clock-names = "sci_ick";
> >>> +	};
> >> 
> >> might be nice to have an example with the 4 interrupts.
> > 
> > The example would be pretty theoretical, as the r8a7790 uses a single
> > interrupt in all cases. What about adding such an example when adding an
> > SoC that uses multiple interrupts to the DT bindings documentation ?
> 
> I'm fine with that if we strip the multiple interrupt bits out of the
> binding until a device/compatible exists that needs it.

I'll do that and repost.

-- 
Regards,

Laurent Pinchart