Re: [PATCH v2 2/3] phy: zynqmp: Add dt bindings for ZynqMP PHY.

[SÃren Brinkmann]

On Wed, 2016-01-13 at 07:43PM +0530, Subbaraya Sundeep Bhatta wrote:
> This patch adds the document describing dt bindings for ZynqMP
> PHY. ZynqMP SOC has a High Speed Processing System Gigabit
> Transceiver which provides PHY capabilties to USB, SATA,
> PCIE, Display Port and Ehernet SGMII controllers.
> 
> Signed-off-by: Subbaraya Sundeep Bhatta <sbhatta@xxxxxxxxxx>

I missed the v2 hence again.

> ---
>  v2:
> 	modified to use phy cells as 2.
> 
>  .../devicetree/bindings/phy/phy-zynqmp.txt         | 103 +++++++++++++++++++++
>  1 file changed, 103 insertions(+)
>  create mode 100644 Documentation/devicetree/bindings/phy/phy-zynqmp.txt
> 
> diff --git a/Documentation/devicetree/bindings/phy/phy-zynqmp.txt b/Documentation/devicetree/bindings/phy/phy-zynqmp.txt
> new file mode 100644
> index 0000000..975cf21
> --- /dev/null
> +++ b/Documentation/devicetree/bindings/phy/phy-zynqmp.txt
> @@ -0,0 +1,103 @@
> +Xilinx ZynqMP PHY binding
> +
> +This binding describes a ZynqMP PHY device that is used to control ZynqMP
> +High Speed Gigabit Transceiver(GT). ZynqMP PS GTR provides four lanes
> +and are used by USB, SATA, PCIE, Display port and Ethernet SGMMI controllers.
> +
> +Required properties (controller (parent) node):
> +- compatible    : Should be "xlnx,zynqmp-psgtr"
> +
> +- reg		: Address and length of register sets for each device in
> +		  "reg-names"
> +- reg-names     : The names of the register addresses corresponding to the
> +		  registers filled in "reg":
> +			- serdes: SERDES block register set
> +			- siou: SIOU block register set
> +			- lpd: Low power domain peripherals reset control
> +			- fpd: Full power domain peripherals reset control

The reset registers should not be directly modified by Linux. Any access
to resets is likely requiring a reset controller that uses platform FW
to modify the resets.
> +
> +-xlnx,tx_termination_fix: Include fix for a functional issue in the GT. The TX
> +			  termination resistance can be out of spec due to a
> +			  bug in the calibration logic. This issue will be fixed
> +			  in silicon in future versions.

The silicon version is run-time detectable. There should be a way to get
away without this property.

	SÃren