On 2/22/24 08:47, Andy Shevchenko wrote:
> On Thu, Feb 22, 2024 at 06:43:08PM +0200, Andy Shevchenko wrote:
>> On Thu, Feb 22, 2024 at 03:23:24PM +0200, Andy Shevchenko wrote:
>>> On Thu, Feb 22, 2024 at 11:07:05AM +1030, Andrew Jeffery wrote:
>>>> On Wed, 2024-02-21 at 20:31 +0200, Andy Shevchenko wrote:
>>>>> Since we have now a common helper to read port properties
>>>>> use it instead of sparse home grown solution.
>>>>
>>>> I did some brief testing of the series for the Aspeed machines under
>>>> qemu, building them on top of v6.8-rc5:
>>>>
>>>> export ARCH=arm
>>>> export CROSS_COMPILE=arm-linux-gnueabihf-
>>>> make aspeed_g5_defconfig
>>>> make -j$(nproc)
>>>> qemu-system-arm -M rainier-bmc -nographic -no-reboot -kernel arch/arm/boot/zImage -dtb arch/arm/boot/dts/aspeed/aspeed-bmc-ibm-rainier.dtb -initrd ...
>>>>
>>>> I got an oops during boot, which bisected to this change:
>>>
>>> Thank you for prompt testing! I will look at it.
>>
>> I found the issue, will be fixed in next version.
> 
> Whoever is going to test this series, the
> 
> -		port->iotype = use_defaults ? UPIO_MEM : port->iotype;
> +		port->iotype = UPIO_MEM;
> 
> should be applied to uart_read_port_properties() implementation.
> 

Thanks, on 8250_bcm7271.c with the above hunk applied, I did not spot 
any differences between the values returned by stty or a cat 
/sys/class/tty/ttyS0/* before or after, the console remained fully 
functional. I will see if I can run an additional test where I removed 
the DT's "clocks" property and confirm that the fall back to 
"clock-frequency" works.

Thanks Andy!
-- 
Florian