00:06 <naoki> Kwiboo: same for new patch... only cpu core (not cluster) is offline and gpu is enabled even if it's marked as broken

00:32 <naoki> oops

00:33 <naoki> # CONFIG_OF_SYSTEM_SETUP is not set...

00:46 <naoki> cpu-code: 35 82

00:46 <naoki> ip-state: 10 00 00

00:46 <naoki> fail gpu

00:46 <naoki> fail cpu cpu@400

00:46 <naoki> fail cpu cpu@500

00:46 <naoki> looks good to me, but it's done before loading devicetree via grub/efi...

00:51 <naoki> supporting devicetree in grub userland(grub.cfg generator) is almost broken, I should use dtb in u-boot

11:04 <Kwiboo> naoki: yeah, latest uboot rk3582 patch should work, tested it on my rock 5c lite with a broken cpu core, it will mark cpu and gpu with status=fail, for Linux v5.17-rc1+ the cpu core will be ignored, however there will still be 2 WARN_ON splat in dmesg during boot, will send a patch to Linux later

11:05 <Kwiboo> and yeas, u-boot only have control over the devicetree that is passed to linux or efi

11:15 <Kwiboo> qschulz: I found another workaround for my u-boot-spl binman entry issue, adding a no-write-symbols prop, should prevent binman adding offset/size symbols for u-boot proper into SPL, could possible impact non-FIT ramboot

12:00 <naoki> it's not 100% sure but it seems this broke USB2.0 HOST ports on ROCK 5A :( https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/arch/arm64/boot/dts/rockchip/rk3588s-rock-5a.dts?h=v6.12&id=b8ac0cf405c2092decc5cc102ed3a2d5d3200ab4

12:03 <naoki> power is supplied but unresponsive after 1 minutes of booting

12:08 <naoki> maybe on-board usb hub stopped

12:34 <naoki> 50% not sure...

12:34 <naoki> I may misunderstand something...