Ls1028a: Current image broken?

MNT Reform
1

I downloaded the latest image for ls1028a from job 5421 (2024-08-14).

Unfortunately, output to the display appears broken in this iteration. The panel back light turns on during boot, but the display stays black all the while.

I connected to UART to see what’s going on. The system boots up and I end up with a login prompt on UART, but the display is still black.

Here is the UART boot log: ls1028a UART boot log (image from job 5421, 2024-08-14) ($9) · Snippets · GitLab

I suspect that this might have been an issue with earlier images, too. With my custom kernel, I haven’t been able to get screen output since 6.9. But I’ve never taken the time to investigate. This is the first time in some time that I booted an official image though…

2

Would you mind if I used your boot log in context of Collecting known-good u-boot output for debugging purposes

That is entirely possible. My LS1028A is in repair and even if it were not, I’m running a headless system. Would you mind trying out the images from reform.debian.net as well? There you can try out the bookworm kernel (6.1) and the bookworm-backports kernel (6.9): reform.debian.net

In the worst case, somebody would have to bisect the kernel to find out which commit broke it…

3

No, not at all. Use it in which ever way you please.

I’m on it. Will get back to you asap.

Thanks for the help!

4

I have now booted both images, as suggested. Interestingly enough, I have found that neither of them managed to print to the screen. Both of them booted fully, with me able to login via UART and the screen lit up both times, but never got anything printed to it.

That had me a bit suspicious. So I reverted back to my 6.8.12 kernel and booted that again. That still works just fine, no issue. So I don’t think it’s a hardware issue :thinking:

I’ve checked and unfortunately, I do not have any of the older system images around any more, so I can’t test with any of those, either :frowning:

Anyway, here are the UART boot logs:

5

So the only thing you changed there is the kernel version but kept everything else as it was? Because there were also a couple of changes we made in reform-tools related to ls1028a, like the modules that get loaded in the initramfs or the kernel console level. But if the only thing you changed was to boot 6.8 instead of 6.10, then this suggests that the issue is on the kernel side.

6

Yes, that is correct. Note however, that this is on Gentoo with my custom kernel.

Unfortunately, I don’t have access to an older official image to test with and see what happens there.

7

OK, scrap that. I’ve found this older image that is working. It’s kernel version 6.8.12.

8

So, after booting the working image with the 6.8.12 kernel, I connected to Wifi and then upgraded the kernel and the kernel only. After that, I rebooted the system. Now the screen is not working any more. So it really is an issue with the kernel it would seem.

Here are the commands I ran to upgrade the kernel:

apt update
apt install linux-image-mnt-reform-arm64

It upgraded to 6.10.4. apt said:

  • Upgrading:
    • libapparmor1
    • linux-headers-mnt-reform-arm64
    • linux-image-mnt-reform-arm64
  • Installing dependencies:
    • apparmor
    • linux-headers-6.10.4-common
    • linux-headers-6.10.4-mnt-reform-arm64
    • linux-image-6.10.4-mnt-reform-arm64
    • linux-kbuild-6.10.4
1 Like
9

This indeed seems to confirm that this is 100% a kernel issue. Somebody needs to bisect this…