[Resending this as I don't see it on the archives]
On Fri, Oct 27, 2023 at 11:47:25AM +0200, Johan Jonker wrote:
Hi Tom, Simon,
The test for my patch ends with an error that is reported on various locations on the internet. My knowledge is limited here. Could you have a look?
Thanks,
Johan
====
rockchip: include: asm: fix entering download mode rk3066 https://source.denx.de/u-boot/custodians/u-boot-rockchip/-/pipelines/18283
Error log: https://source.denx.de/u-boot/custodians/u-boot-rockchip/-/jobs/720456
Error message: Cleaning up project directory and file based variables 00:03 ERROR: Job failed: exit code 100
So, this is just the last line of the job saying the build failed. Over at https://source.denx.de/u-boot/custodians/u-boot-rockchip/-/jobs/720456#L1294 we see the build summary, and one platform failed. Looking at the commit you note itself, I suspect the problem is that some platform (a 32bit one since that's the job that failed without a useful error) is hitting the ROCKCHIP_BOOT_LR_REG and ROCKCHIP_BOOT_RETURN_REG questions and not having a value and so config fails, and buildman isn't catching and reporting the error. Looking further, the values in arch/arm/mach-rockchip/rk3066/Kconfig for those should be with the question as: default 0x00001058 if RK3066 and so forth.
Doing ./tools/qconfig.py -s I see that it eventually hangs on one platform. Doing lsof | grep kconf points me at the build directory and it is one with: CONFIG_SYS_SOC="rk3188" CONFIG_SYS_VENDOR="radxa" CONFIG_SYS_BOARD="rock" CONFIG_SYS_CONFIG_NAME="rock" in the .config that is stuck. That points me at rock_defconfig as being the one which is tripping up the question. If those new options are only valid on say RK3066 they should depend on it specifically too.