Date:   Mon, 28 Aug 2023 10:23:45 +0200
From:   Neil Armstrong <neil.armstrong@...aro.org>
To:     Dmitry Rokosov <ddrokosov@...rdevices.ru>,
        Lucas Tanure <tanure@...ux.com>
Cc:     Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        Jiri Slaby <jirislaby@...nel.org>,
        Kevin Hilman <khilman@...libre.com>,
        linux-kernel@...r.kernel.org, linux-serial@...r.kernel.org,
        linux-arm-kernel@...ts.infradead.org,
        linux-amlogic@...ts.infradead.org
Subject: Re: [PATCH v2] Revert "tty: serial: meson: Add a earlycon for the T7
 SoC"

Hi,

On 28/08/2023 10:20, Dmitry Rokosov wrote:
> On Sun, Aug 27, 2023 at 11:31:35AM +0100, Lucas Tanure wrote:
>> On 27-08-2023 11:23, Neil Armstrong wrote:
>>> Hi Lucas,
>>>
>>> Le 27/08/2023 à 12:04, Lucas Tanure a écrit :
>>>> On 27-08-2023 10:48, Neil Armstrong wrote:
>>>>> Hi Lucas,
>>>>>
>>>>> Le 27/08/2023 à 10:29, Lucas Tanure a écrit :
>>>>>> This reverts commit 6a4197f9763325043abf7690a21124a9facbf52e.
>>>>>> New SoC will use ttyS0 instead of ttyAML, so T7 SoC doesn't
>>>>>> need a OF_EARLYCON_DECLARE.
>>>>>>
>>>>>> Signed-off-by: Lucas Tanure <tanure@...ux.com>
>>>>>> ---
>>>>>> Since V1:
>>>>>> - add Signed-off-by:
>>>>>>
>>>>>>    drivers/tty/serial/meson_uart.c | 2 --
>>>>>>    1 file changed, 2 deletions(-)
>>>>>>
>>>>>> diff --git a/drivers/tty/serial/meson_uart.c
>>>>>> b/drivers/tty/serial/meson_uart.c
>>>>>> index c4f61d82fb727..790d910dafa5d 100644
>>>>>> --- a/drivers/tty/serial/meson_uart.c
>>>>>> +++ b/drivers/tty/serial/meson_uart.c
>>>>>> @@ -648,8 +648,6 @@ meson_serial_early_console_setup(struct
>>>>>> earlycon_device *device, const char *opt
>>>>>>    OF_EARLYCON_DECLARE(meson, "amlogic,meson-ao-uart",
>>>>>>                meson_serial_early_console_setup);
>>>>>> -OF_EARLYCON_DECLARE(meson, "amlogic,t7-uart",
>>>>>> -            meson_serial_early_console_setup);
>>>>>>    #define MESON_SERIAL_CONSOLE_PTR(_devname)
>>>>>> (&meson_serial_console_##_devname)
>>>>>>    #else
>>>>>
>>>>> How do you use earlycon with this removed ?
>>>>>
>>>>> Neil
>>>> This is my kernel cmdline:
>>>> console=ttyS0,921600 no_console_suspend earlycon=ttyS0,0xfe078000
>>>>
>>>> And I can see my log:
>>>> boot 64bit kernel
>>>> [    0.000000] Booting Linux on physical CPU 0x0000000000 [0x410fd092]
>>>> [    0.000000] Linux version 6.5.0-rc7-next-20230825+
>>>> (tanureal@...en) (aarch64-none-linux-gnu-gcc (GNU Toolchain for the
>>>> A-profile Architecture 10.3-2021.07 (arm-10.29)) 10.3.1 20210621,
>>>> GNU ld (GNU Toolchain fo3
>>>> [    0.000000] KASLR disabled due to lack of seed
>>>> [    0.000000] Machine model: Khadas vim4
>>>> [    0.000000] OF: reserved mem:
>>>> 0x0000000005000000..0x00000000052fffff (3072 KiB) nomap non-reusable
>>>> secmon@...0000
>>>> [    0.000000] OF: reserved mem:
>>>> 0x0000000005300000..0x00000000072fffff (32768 KiB) nomap
>>>> non-reusable secmon@...0000
>>>> ...
>>>> [    0.079368] io scheduler mq-deadline registered
>>>> [    0.079374] io scheduler kyber registered
>>>> [    0.079549] io scheduler bfq registered
>>>> [    0.083373] fe078000.serial: ttyS0 at MMIO 0xfe078000 (irq = 14,
>>>> base_baud = 1500000) is a meson_uart
>>>> [    0.083403] printk: console [ttyS0] enabled
>>>>
>>>> As the log is OK, and T7 for now is binding against S4 code we
>>>> should drop this patch and add a S4 one if needed.
>>>>
>>>> But just having this log is not enough for testing earlycon?
>>>> I am assuming that by just having the log since 0.0000 is a good
>>>> sign that earlycon is working.
>>>> Could you give further guidance?
>>>
>>> First the kernel argument is simply "earlycon" with no other options,
>>> the earlycon code will fetch the first uart using the DT
>>> /chosen/stdout-path property.
>>>
>>> Then you should see something like right after "Machine model":
>>> [    0.000000] earlycon: meson0 at MMIO 0x00000000ff803000 (options
>>> '115200n8')
>>>
>>> The boot log you share doesn't use earlycon, it waits until all
>>> depedency of the UART
>>> has been probed before probing the UART driver and prints the whole boot
>>> log.
>>>
>>> Earlycon is a mechanism to use the UART HW as configured by the
>>> bootloader in
>>> a minimal way very early in the boot in order to detect very early lockups.
>>>
>>> Keeping OF_EARLYCON_DECLARE for t7 would enable this.
>>>
>>> Neil
>>>
>>>>
>>>> Lucas Tanure
>>>>
>>>
>> But then we would need to add Earlycon for S4 too.
>> How can we add one Earlycon for all future SoCs, like S4 and T7?
>>
> 
> Per my understanding, you just need to declare UART dts node with the
> following compatible string: "amlogic,meson-ao-uart". It will enabled
> earlycon. There is no neccessary to declare separate
> OF_EARLYCON_DECLARE(). It's working okay for A1 dts:

This works on A1 because uart_AO_B is an Always-On UART, and using amlogic,meson-ao-uart
is corrent.
But for S4 & T7, Amlogic dropped the Always-On power domain, so there's no more AO uarts,
so using amlogic,meson-ao-uart on those is abusing the DT to enable a driver functionality.

So the solution is to add those OF_EARLYCON_DECLARE and only use the correct
/chosen/stdout-path property.

Neil


> 
> meson-a1.dtsi uart declaration:
> 	uart_AO_B: serial@...0 {
> 		compatible = "amlogic,meson-a1-uart",
> 			     "amlogic,meson-ao-uart";
> 		reg = <0x0 0x2000 0x0 0x18>;
> 		interrupts = <GIC_SPI 26 IRQ_TYPE_EDGE_RISING>;
> 		clocks = <&xtal>, <&xtal>, <&xtal>;
> 		clock-names = "xtal", "pclk", "baud";
> 		status = "disabled";
> 	};
> 
> earlycon declaration from dts:
> 	aliases {
> 		serial0 = &uart_AO_B;
> 	};
> 
> 	chosen {
> 		stdout-path = "serial0:115200n8";
> 	};
> 
>>
>> _______________________________________________
>> linux-amlogic mailing list
>> linux-amlogic@...ts.infradead.org
>> http://lists.infradead.org/mailman/listinfo/linux-amlogic
> 

Powered by blists - more mailing lists