[U-Boot] [RFC PATCH] arm: zynqmp: Add ZynqMP minimal R5 support

[Michal Simek]

>>>> diff --git a/configs/xilinx_zynqmp_r5_defconfig
>>>> b/configs/xilinx_zynqmp_r5_defconfig
>>>> new file mode 100644
>>>> index 000000000000..46715242e703
>>>> --- /dev/null
>>>> +++ b/configs/xilinx_zynqmp_r5_defconfig
>>>> @@ -0,0 +1,16 @@
>>>> +CONFIG_ARM=y
>>>> +CONFIG_ARCH_ZYNQMP_R5=y
>>>> +CONFIG_SYS_TEXT_BASE=0x10000000
>>>
>>> You may want to document somewhere what this address is. Is this OCM? Is
>>> this DDR? Is this the super-fast local RAM of the R5?
>>
>> In commit message it is written about DDR but maybe it can be more
>> explicitly said.
> 
> The history is almost invisible when looking at the code itself once things are committed to the tree. So yes, a comment or readme in-tree would be good.

I have extended commit message. Normally you run blame for that line
which setup that address and then looking at commit message.

>>>> +CONFIG_DEFAULT_DEVICE_TREE="zynqmp-r5"
>>>> +CONFIG_DEBUG_UART=y
>>>> +# CONFIG_DISPLAY_CPUINFO is not set
>>>> +CONFIG_SYS_PROMPT="ZynqMP r5> "
>>>> +# CONFIG_CMD_FLASH is not set
>>>> +# CONFIG_CMD_SETEXPR is not set
>>>> +CONFIG_OF_EMBED=y
>>>> +CONFIG_DEBUG_UART_ZYNQ=y
>>>> +CONFIG_DEBUG_UART_BASE=0xff010000
>>>> +CONFIG_DEBUG_UART_CLOCK=100000000
>>>> +CONFIG_ZYNQ_SERIAL=y
>>>> +CONFIG_TIMER=y
>>>> +CONFIG_CADENCE_TTC_TIMER=y
>>>> diff --git a/drivers/serial/Kconfig b/drivers/serial/Kconfig
>>>> index 3d5b2bf15f08..3292edbaf240 100644
>>>> --- a/drivers/serial/Kconfig
>>>> +++ b/drivers/serial/Kconfig
>>>> @@ -617,7 +617,7 @@ config STM32_SERIAL
>>>>     config ZYNQ_SERIAL
>>>>       bool "Cadence (Xilinx Zynq) UART support"
>>>> -    depends on DM_SERIAL && (MICROBLAZE || ARCH_ZYNQ || ARCH_ZYNQMP)
>>>> +    depends on DM_SERIAL && (MICROBLAZE || ARCH_ZYNQ || ARCH_ZYNQMP
>>>> || ARCH_ZYNQMP_R5)
>>>
>>> I'm sure there are more drivers that could be accessed from the R5
>>> instead, right?
>>
>> I would expect that but it needs to be tested which is not at this
>> stage. This is minimal support with DDR/timer/serial which is enough for
>> now.
> 
> I think a quick git grep -e on depends.*ARCH_ZYNQMP would make sense at this stage already. Otherwise people will just fall into traps later on.

When that driver is tested it can be enabled. It is also a question if
make sense to protect these drivers with DM in place.

> 
>>
>>
>>>
>>> Either way, I'm happy to see this port - looks like great work :)
>>
>> Happy to see any your tag to write it to the history.
> 
> I don‘t tag RFCs :). For the next (non-RFC) version, I‘ll be happy to do so.

v1 out.

Thanks,
Michal