[PATCH v2 4/9] ARM: dts: rockchip: rv1126: Add i2c2 nodes

[Heiko Stübner]

Am Montag, 27. November 2023, 10:58:43 CET schrieb Tim Lunn:
> 
> On 11/27/23 20:45, Heiko Stübner wrote:
> > Am Montag, 27. November 2023, 10:27:41 CET schrieb Krzysztof Kozlowski:
> >> On 27/11/2023 10:23, Tim Lunn wrote:
> >>> Hi
> >>>
> >>> On 11/27/23 17:52, Krzysztof Kozlowski wrote:
> >>>> On 27/11/2023 00:06, Heiko Stübner wrote:
> >>>>> Hi Krzysztof,
> >>>>>
> >>>>> Am Mittwoch, 22. November 2023, 13:29:47 CET schrieb Krzysztof Kozlowski:
> >>>>>> On 22/11/2023 13:22, Tim Lunn wrote:
> >>>>>>> Add i2c2 node and i2c2_xfer pinctrl for Rockchip RV1126
> >>>>>>>
> >>>>>>> Signed-off-by: Tim Lunn <tim at feathertop.org>
> >>>>>>> ---
> >>>>>>>
> >>>>>>> (no changes since v1)
> >>>>>>>
> >>>>>>>    arch/arm/boot/dts/rockchip/rv1126-pinctrl.dtsi | 10 ++++++++++
> >>>>>>>    arch/arm/boot/dts/rockchip/rv1126.dtsi         | 15 +++++++++++++++
> >>>>>>>    2 files changed, 25 insertions(+)
> >>>>>>>
> >>>>>>> diff --git a/arch/arm/boot/dts/rockchip/rv1126-pinctrl.dtsi b/arch/arm/boot/dts/rockchip/rv1126-pinctrl.dtsi
> >>>>>>> index 4f85b7b3fc4c..167a48afa3a4 100644
> >>>>>>> --- a/arch/arm/boot/dts/rockchip/rv1126-pinctrl.dtsi
> >>>>>>> +++ b/arch/arm/boot/dts/rockchip/rv1126-pinctrl.dtsi
> >>>>>>> @@ -87,6 +87,16 @@ i2c0_xfer: i2c0-xfer {
> >>>>>>>    				<0 RK_PB5 1 &pcfg_pull_none_drv_level_0_smt>;
> >>>>>>>    		};
> >>>>>>>    	};
> >>>>>>> +	i2c2 {
> >>>>>>> +		/omit-if-no-ref/
> >>>>>>> +		i2c2_xfer: i2c2-xfer {
> >>>>>>> +			rockchip,pins =
> >>>>>>> +				/* i2c2_scl */
> >>>>>>> +				<0 RK_PC2 1 &pcfg_pull_none_drv_level_0_smt>,
> >>>>>>> +				/* i2c2_sda */
> >>>>>>> +				<0 RK_PC3 1 &pcfg_pull_none_drv_level_0_smt>;
> >>>>>>> +		};
> >>>>>>> +	};
> >>>>>>>    	pwm2 {
> >>>>>>>    		/omit-if-no-ref/
> >>>>>>>    		pwm2m0_pins: pwm2m0-pins {
> >>>>>>> diff --git a/arch/arm/boot/dts/rockchip/rv1126.dtsi b/arch/arm/boot/dts/rockchip/rv1126.dtsi
> >>>>>>> index 6c5c928f06c7..cf1df75df418 100644
> >>>>>>> --- a/arch/arm/boot/dts/rockchip/rv1126.dtsi
> >>>>>>> +++ b/arch/arm/boot/dts/rockchip/rv1126.dtsi
> >>>>>>> @@ -21,6 +21,7 @@ / {
> >>>>>>>    
> >>>>>>>    	aliases {
> >>>>>>>    		i2c0 = &i2c0;
> >>>>>>> +		i2c2 = &i2c2;
> >>>>>> No, this should be per-board to match board labeling/schematics.
> >>>>> At least for i2c, uarts and i.e. spi ... Rockchip manuals, pin namings
> >>>>> and also all board schematics I've seen so far are very consistent for
> >>>>> these ... i2c2 for example is labled i2c2 both in the pins in the socs
> >>>>> and also in the board-schematics using them.
> >>>>>
> >>>>> So while I can agree that things like mmc-aliases might be board-specific,
> >>>>> I do think aliases for the core busses should be able to live in the soc dtsi
> >>>>> as for all Rockchip SoCs so far?
> >>>> If you do not list here all aliases, it is already board-specific, isn't it?
> >>>>
> >>>> https://lore.kernel.org/linux-rockchip/CAK8P3a25iYksubCnQb1-e5yj=crEsK37RB9Hn4ZGZMwcVVrG7g@mail.gmail.com/
> >>> I had only added aliases for nodes that are implemented. RV1126 has 6
> >>> i2c busses but so far not all these exist in the soc.dtsi.
> >>>
> >> OK, that would explain why you have only two. Anyway, it is just generic
> >> guideline, so up to Heiko what to do with it.
> > People see "uart2-tx" on their pin-header description and then of course
> > want to use ttyS2 ... same with reading i2c2-sda on the pin-header and then
> > running i2cdetect on said i2c2 dev node.
> >
> > And even if i2c1 is not populated (would even be very rare), then people
> > would still expect i2c2 to be named that way.
> >
> > As this is the same for _every_ board, it doesn't really make sense to
> > duplicate it every time. Which I guess is similar to what I wrote in reply
> > to the mail you linked above :-) .
> So I will leave the i2c and serial aliases as they are in rv1126.dtsi then?

correct


> > It's different for mmc, where this naming scheme isn't normally used,
> > so we (mostly?) migrated to them being in the board-specific alias section.
> 
> What about the ethernet alias,   the MAC is part of the SoC, should this 
> stay as it is
> in patch 6 from this series, or move to board dtsi?

I think judging from everything above (and the other modern boards
like all rk3588 ones) the ethernet0 alias should move to the board dts(i).

The gmac generally does not have this numbering scheme like i2c or uarts.

Heiko