Query on pinctrl usage for DT nodes

[Tony Lindgren]

* Stephen Warren <swarren at wwwdotorg.org> [130403 09:49]:
> On 04/03/2013 03:16 AM, Prabhakar Lad wrote:
> > Hi Linus/Stephen,
> > 
> > I am working adding  DT nodes for DA850.
> > 
> > Following is the pin control snippet of da850.dtsi:-
> > 
> > 		pmx_core: pinmux at 1c14120 {
> ...
> >                         i2c0_pins: pinmux_i2c0_pins {
> >                                 pinctrl-single,bits = <
> >                                         /* I2C0_SDA,I2C0_SCL */
> >                                         0x10 0x00002200 0x0000ff00
> >                                 >;
> >                         };
> >                         mdio_pins: pinmux_mdio_pins {
> >                                 pinctrl-single,bits = <
> >                                         /* MDIO_CLK, MDIO_D */
> >                                         0x10 0x00000088 0x000000ff
> >                                 >;
> >                         };
> >                 }
> > 
> > And following is the code snippet for dts:-
> > 
> > 		i2c0: i2c at 1c22000 {
> > 			status = "okay";
> > 			clock-frequency = <100000>;
> > 			pinctrl-names = "default";
> > 			pinctrl-0 = <&i2c0_pins>;
> > 		};
> > 		mdio: mdio at 1e24000 {
> > 			status = "okay";
> > 			bus_freq = <2200000>;
> > 			pinctrl-names = "default";
> 
> I assume there's also the following there:
> 
> 			pinctrl-0 = <&mdio_pins>;
> 
> >               };
> > 		
> > But while booting I see the following boot log:-
> > ...
> > cpuidle: using governor menu
> > TCP: cubic registered
> > NET: Registered protocol family 17
> > pinctrl-single 1c14120.pinmux: pin 1c14130 already requested by
> > davinci_mdio.0; cannot claim for i2c_davinci.1
> > pinctrl-single 1c14120.pinmux: pin-4 (i2c_davinci.1) status -22
> > pinctrl-single 1c14120.pinmux: could not request pin 4 on device pinctrl-single
> > console [netcon0] enabled
> > ....
> > 
> > This is because the mdio and i2c are using same pin 0x10,
> 
> How can two devices use the same pin? I mean physically, in hardware?
> 
> Is this because pinctrl-single uses the register address as the pin
> number, whereas you have registers which configure multiple pins at
> once? If so, your hardware isn't something that can be represented by
> pinctrl-single.
> 
> > Is there any
> > alternative way to handle if the two node's are using same pins any
> > pointers could be very much helpful ?

It could also that the mux register(s) follow the one-mux-per-bit
mapping.

In that case pinctrl-single,bits option as documented in the
Documentation/devicetree/bindings/pinctrl/pinctrl-single.txt.

Regards,

Tony