[Bug]: No HDMI signal with SKIPR and any 6.x kernels on non-standard resolutions (e.g.1024x600) #33
Comments
|
The screenshot was taken from So questions:
|
|
Hi, first of all, thanks for caring! 1: they were hooked before powering on Additional info: The screen(s) get activated when the board powers on, then once more during the boot process, but they display no image and go to power saving very soon. I have another skipr unopened in its box I can try on that to minimize the chance of hardware malfunction, also several of the 7" touchscreen, but the fact that it works flawlessly with 5.x kernels gives quite a low probability. Also, the same display works fine on my Sovol SV08 (BTT-CB1 based electronics) with the image build from main branch of this repo ( I cannot decide what logs or other info might be relevant, please specify and I will post them. Slightly related: The SPI-based touchscreen works fine on 6.x |
|
It's weird, let's try to realize what's going on here.
Basically I would try to understand how system see this HDMI device. BTW, I do believe running cutting edge image with 6.x kernel is not absolutely necessary (for 3D printer). My machine still works on pretty old image and I have never experienced any problem. And in worst case we may try to build an image with latest Armbian but 5.x kernel. |
|
1: I have 2 boards, both 1.0 I tested with my second board and another 7" touch and the results seemed to be the same until I removed the power from the screen (powered on USB) then connected it again. The console appeared in a few seconds. Then I tested (after reboot and a blank screen) if disconnecting/reconnecting the HDMI cable does the trick, but unfortunately, it does not. Requested info from 6.6.59-current-rockchip64, immediately after boot, screen is blank but ... to be continued... |
|
After power-cycling the touchscreen: (what isn't mentioned, didn't change) ... to be continued ... |
|
Sounds strange. Logs shows that the board recognizes monitor and able to read EDID info. May be issue in non standard "1024x600" resolution.... or this may point to luck or power (if display powered by USB). Could you try to connect it via USB3 port with all other USB devices disconnected (especially if you have WIFI dongle or WEB cam here)? Or try to power it from external source. |
|
On Armbian_23.05.0-trunk_Mkspi_jammy_current_5.15.111: |
|
On my laptop: |
|
Might be a bit more relevant: Armbian on RPi4 with the same display |
What it also strange, that Laptop and Armbian with 5.x kernel see 1920x1080 resolution and looks like uses it as default. I am not sure why 6.x kernel does not works in this case, and RPi4 does not use this resolution too. Could you try to use BTW, there are new release with the new kernels. And if it does not works, you always may use latest release with 5.x one |
attilabody
changed the title
|
|
OK, a short update: I found a screen where I can reproduce it. I have a display module with 1024x600 resolution which does not works property on the latest images. Now I am trying to realize why and how to fix it... |
|
@attilabody could you try 0.4.4-25.2.0-trunk release (one of current images)? There are some HDMI related adjustments. |
Sorry, I did not notice. I will do it this evening. |
|
feel free to use latest 1.x images, they should have the same patches |
|
Tested with the jammy image. There is picture, initially in 1920x1080 resolution (which this display "emulates", so the console is kind of readable, but image quality is not the best. After power-cycling or reconnecting, the signal seems to switch to the the native (1024x600) resolution (as far as I can see by naked eye), but the console still being drawn onto a 1920x1080 "virtual" screen. Output of "fbset -i" does not change. After boot: After reconnect: |
|
The same behavior here, which looks like one step forward :) What I'v realized so far: |
2 tasks
|
Isn't it possible that the root cause is quite simple, like a missing modeline from the device tree? |
|
Kind of... I BTW, this is fragmen from kernel debug logs, which shows that EDID data was decoded but kernel module just rejected custom modes with "NOCLOCK" resolution: |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
What happened?
I haven't been able to get a screen on HDMI with the latest (0.4.0-24.11.0-trunk) current (or edge), so I bisected. Results:
The most recent release which works is 0.4.0-24.11.0-trunk with 'current' (5.15.111) kernel. None of the releases show any image on HDMI using the 'edge' kernel (which was already 6.x-based in the first release)
How to reproduce?
Download any release with 6.x kernel (I tested with jammy because that is the most supported, but it should not matter), connect only an HDMI monitor to an MKS SKIPR board and boot.
Branch
main (main development branch)
On which host OS are you running the build script and observing this problem?
Irrelevant, I used the image from the releases
Relevant log URL
No response
Code of Conduct
The text was updated successfully, but these errors were encountered: