02:13 <levd> "568262bf5492 usb: dwc3: core: Add shutdown callback for dwc3" breaks sdcard reboot on rk3399 board. This commit adds a shutdown handler which calls dwc3_remove(dev). However, during the sdcard reboot, the kernel calls shutdown handler, then remove handler of the usb driver. dwc3_remove is called twice which cause a kernel crash.

02:15 <levd> mmind00: shall this commit be reverted?

08:52 <robmur01> levd: already reverted upstream, see 8f11fe7e4068

08:53 <levd> robmur01: thanks man

10:27 <mmind00> levd robmur01: looks like I'm too late to the party then :-)

11:00 <macc24> leming robmur01: so it works now?

11:00 <macc24> mps: look someone fixed that without me xD

11:01 <mps> macc24: what is fixed?

11:02 <macc24> mps: the poweroff bug on kevin i guess lol

11:02 <mps> ah, nice. never didn't had this problem, iirc

11:03 <mps> some options in kernel config was missing I guess

11:06 <robmur01> or you're not running RC kernels, which unless you're specifically out to find kernel bugs is fairly sensible ;)

11:07 <LinuxHackerman> oh, is the shutdown bug a different one from the bob one then?

11:07 <macc24> i am always running -rc

11:08 <macc24> because it's easier to figure out whats newer with a curl xD

11:08 <macc24> days*

11:10 <mps> I'm testing -rc kernel usually once in a week but running stable for workdays

11:20 <alpernebbi> I'm still at 5.10 because Debian froze there :D

11:26 <alpernebbi> sphalerite: I can try adding some hacks to explicitly probe display and panel stuff, change backlight etc during board initialization

11:28 <alpernebbi> maybe enable debug logs to see what fails how other than those pinctrl messages you got

11:30 <alpernebbi> one thing I'm interested is if it works as RW_LEGACY on the original depthcharge firmware

11:32 <alpernebbi> also, are you trying to boot from emmc? maybe usb/microSD works better

16:38 <sphalerite> Yeah, trying to boot from emmc. Would that really make a difference, between coreboot+u-boot and only u-boot? Especially given that u-boot does manage to load the kernel?

16:38 <sphalerite> Hm, how do I chainload u-boot from depthcharge?

16:39 <sphalerite> and is RW_LEGACY really a thing on ARM? I thought it wasn't

18:02 <alpernebbi> sphalerite: hi

18:04 <alpernebbi> emmc vs external storage probably doesn't make a difference, you're right

18:07 <alpernebbi> I have no idea what's preventing linux from booting

18:10 <alpernebbi> for display I uploaded https://github.com/alpernebbi/u-boot/releases/tag/tmp-gru-bob-hacks-20210701 which has debug messages enabled, and explicitly tries to do some pwm/panel/vidconsole stuff to trigger failures to see logs about them

18:11 <alpernebbi> (at a different branch: rk3399-gru-bob-edp/hacks)

18:12 <alpernebbi> I know that coreboot does some clock stuff different, so if u-boot doesn't know how to initialize some clocks they could stay as coreboot sets them and happen to work in a coreboot+u-boot scenario

18:13 <alpernebbi> does coreboot+u-boot work better in general than just u-boot?

18:17 <alpernebbi> for RW_LEGACY, the mechanism is intact on at least my device, there's just no default payload, trying to use it beeps at you as if it's disabled. I guess people misunderstood that and assumed it's not possible to use it

18:20 <alpernebbi> I tried to write about how to use it a bit at https://github.com/alpernebbi/u-boot/wiki but in short write the .cbfs file I uploaded to the last 2MiB of SPI, run crossystem dev_boot_legacy=1, use CTRL+L to boot with it