Re: [PATCH 6/9] arm64: dts: meson: g12a: Add UART A, B & C nodes and pins
From: Neil Armstrong
Date: Sun Mar 17 2019 - 06:43:09 EST
Le 16/03/2019 23:14, Martin Blumenstingl a Ãcrit :
> Hi Neil,
>
> On Sat, Mar 16, 2019 at 3:35 PM Neil Armstrong <narmstrong@xxxxxxxxxxxx> wrote:
> [...]
>>>> + uart_ao_a_c_pins: uart_ao_a_c {
>>>> + mux {
>>>> + groups = "uart_ao_a_rx_c",
>>>> + "uart_ao_a_tx_c";
>>>> + function = "uart_ao_a_c";
>>>> + bias-disable;
>>>> + };
>>>> + };
>>> I'm fine with this part if you mention it in the subject and/or the description
>>> uart_ao_a_c routes two pins from bank C (from the EE domain) to the
>>> uart_AO controller (from the AO domain)
>>
>> Not sure DT is the right place for that, I think I'll remove this until
>> we have it actually used somewhere.
> I'm fine with that as well
>
> [...]
>>>> +
>>>> + uart_C: serial@22000 {
>>>> + compatible = "amlogic,meson-gx-uart";
>>>> + reg = <0x0 0x22000 0x0 0x18>;
>>>> + interrupts = <GIC_SPI 93 IRQ_TYPE_EDGE_RISING>;
>>>> + clocks = <&xtal>, <&clkc CLKID_UART1>, <&xtal>;
>>> does uart_C really use CLKID_UART1? on GX uart_C uses CLKID_UART2
>>
>> It seems so :
>> https://github.com/hardkernel/linux/blob/odroidn2-4.9.y/arch/arm64/boot/dts/amlogic/mesong12a.dtsi#L1020
> it's weird but we can always fix it up later if needed. so let's keep
> it for now to stay consistent with the vendor kernel until we know
> better
>
> Jianxin, can you please check with the hardware team whether the
> uart_C gate clock (pclk) is CLKID_UART1 or CLKID_UART2 on G12A?
I think it's a typo, it should use CLKID_UART2
Neil
>
>
> Regards
> Martin
>