[Buildroot] [PATCH v1 2/2] configs/zynqmp_zcu106_defconfig: transition to tarballs

[Neal Frager]

Hi Luca,

>
>>   >> [    4.384715] zynqmp-display fd4a0000.display: vtc bridge property
>>   >> not present
>>   >> [    4.393156] ------------[ cut here ]------------
>>   >> [    4.397776] More than allowed devices are using the vpll_int, which is forbidden
>>
>>
>>   > This is caused by not proper psu_init which setup PLLs incorrectly.
>>
>> This is using the psu_init_gpl.c in U-Boot:
>>
>> https://github.com/Xilinx/u-boot-xlnx/blob/xlnx_rebase_v2022.01_2022.
>> 1 /board/xilinx/zynqmp/zynqmp-zcu106-revA/psu_init_gpl.c
>>
>> And the pmu configuration object from Neal:
>>
>> https://git.buildroot.net/buildroot/tree/board/zynqmp/zcu106/pm_cfg_o
>> b
>> j.c
>>
>> So what exactly needs to change?
>
>> I don't have this board here that's why likely taken from any old design before this check was added to Linux.
>> U-Boot psu_init_gpl.c should be updated.
>
> Could you please share exactly which register writes in the psu_init_gpl.c need to change?
> Perhaps you can point me to a known good one, so that I can compare those specific register writes with the zynqmp-zcu106-revA file?

> I'm OK with either solution, but I wonder whether it would be a better idea to regenerate psu_init_gpl.c (and pm_cfg_obj.c while there) using Vivado 2022 and mainline them in U-Boot. Would that fix the issue as well? Vivado is assumed to generate > correct files, if it didn't in the past, chances are it has been fixed in the meantime.

In the case of the pm_cfg_obj.c file, we are using the file that is generated by Vitis.  
I have confirmed that Vitis 2022.1 and 2021.2 both generate the exact same pm_cfg_obj.c files for the zcu102 and zcu106 boards.
And these are the files that are currently part of the buildroot sources in the board/zynqmp directory.

But the psu_init_gpl.c file is coming from u-boot, so if this is not properly updated, that could be an issue.
Plus, the pll configurations all happen in the psu_init routine of the psu_init_gpl.c.

I agree with Luca that it would be best to refresh all of the psu_init_gpl.c files in u-boot mainline with Vivado generated versions.  
That should fix any issues.

Best regards,
Neal Frager
AMD