Posts by HiassofT

    The log shows that the video driver read an invalid edid from your projector and therefore refused to output via HDMI.

    Did you have to use any hdmi_... config.txt settings with the older milhouse build?

    Can you run "pastekodi" on the milhouse build as well and also run "tvservice -d edid.bin" on it and upload the resulting edid.bin file?

    so long,

    Hias

    Doesn't anyone have an idea, not even the developer?

    Ah, this seems to be a quirk in the firmware handling/parsing of dt parameters - the dtoverlay scope of vc3-kms-v3d, which is loaded in distroconfig.txt on RPi2/3, seems to persist after the include in config.txt so the audio parameter applies to the overlay.

    The solution is quite simple, clear the overlay scope with "dtoverlay=" or move the dtparam=audio=on statement up, before the include distroconfig.txt.

    eg

    Code
    dtoverlay=
    dtparam=audio=on
    audio_pwm_mode=1

    so long,

    Hias

    One thing I did notice is that during the boot the red LED on the Pi was coming on and off which I havent seen before. It normally stays on permantently after power on.

    The red LED must be permanently on. If it's going off it means you have a power supply problem which you need to solve first - if the power isn't stable the RPi won't run stable and it could well be that the odd crashes you are seeing are caused by that.

    Disconnect all USB peripherals and everything else you might have attached (fans, HATs etc) and make sure everything's working fine. I also strongly recommend using the official RPi power supply as that's known to work just fine.

    so long,

    Hias

    Can you connect with ssh when connected to the projector? If yes run "pastekodi" and "modetest | pastebinit" and post the URLs.

    There's currently a bug in the video driver, if the HDMI cable isn't plugged in or the driver doesn't detect a display then RPi2/3 lock up during boot - this can also be caused by a broken (or not fully wired) HDMI cable, so testing with a different HDMI cable would be worth a try, too.

    so long,

    Hias

    ntfs-3g is built in (as in the official LE images).

    Better stay away from USB Y-cables, at best they won't help (RPi has a total limit it can draw from USB-C and provide to USB ports) - use the official RPi power supply and a powered USB HDD or a powered USB hub if your drive needs more power.

    so long,

    Hias

    for me , things seems to be more stable, still testing but so far it is solid, usuallly it crash in 20 minutes and so far it is holding up nicely

    by the way , are you going to keep making 64bit kernels from now on or this is just a quick test ?

    Thanks for testing and reporting back!

    It would be great if you - and other people with OOM issues - could do some extensive testing with this build, keeping an eye on issues (also seemingly unrelated ones), and tell us about the results.

    For now it's only a quick test, but if the OOM issues are indeed resolved by the 64bit kernel and LE runs fine with it then we will consider switching official LE RPi4 builds to it in the future.

    so long,

    Hias

    Modifying /flash/distroconfig.txt is the wrong approach - it will be replaced on every update. Commenting out the "include distroconfig.txt" in /flash/config.txt and replacing that with custom dtoverlay or other settings is even more dangerous as it could brick your installation on an upgrade - the stuff in distroconfig.txt is essential.

    If you need to force a video mode you should add a video= parameter to /flash/cmdline.txt, eg video=HDMI-A-1:640x480M@60 - see the kernel docs for more info modedb default video mode support — The Linux Kernel documentation

    It would help even more if you post why you need to change settings, which settings they are and provide pastekodi logs - then we would have a chance to look into it and if there's an issue we could fix it. If you don't report issue details and post logs they will never get fixed.

    so long,

    Hias

    Thanks for testing!

    I checked the driver code and noticed it only supports 180° rotation and x/y mirroring, but not 90/270° rotation. And I vaguely remember one of the RPi devs stating that 90/270° rotation was only possible with the help of videocore code in the firmware of the old, proprietary RPi0-3 driver.

    So this will be quite a bit more involved, kodi's GBM windowing code would need to be extended to support rotation, direct-to-plane rendering probably won't work but egl rendering should be possible.

    so long,

    Hias

    Display rotation should be possible by using the rotate and/or panel_orientation options of the video kernel command line parameter.

    eg add something like video=DSI-1:1920x1080@60,panel_orientation=left_side_up

    See the kernel documentation for infos about the video parameter: modedb default video mode support — The Linux Kernel documentation

    The RPi forums might be a better place to ask about DSI/DPI panels and KMS details (I never used DSI/DPI displays myself), I found eg this post there rotate display with DPI and vc4-kms-v3d - Raspberry Pi Forums

    so long,

    Hias

    The lockup seems to be HDMI related, it happens both on LE and RPiOS with latest rpi-update kernel/firmware kernel: drm/vc4: More CPU Stall Fixes · Hexxeh/rpi-firmware@9dbcd6d · GitHub (LE uses the same fw/kernel version as this rpi-update commit) when booting without a HDMI monitor connected.

    I already chatted with RPi devs, they'll look at it next week (popcornmix and I also confirmed it still happens with current rpi-5.10.y kernel which contains some fixes in that area).

    Concerning LE kernel config changes: it'd be best if you change the panel driver to "y", so it's available right after boot (otherwise you won't get splash and update notifications on screen).

    Also drop the gpu_mem change from config.txt (that's not needed and only wastes memory) and you can also drop the vc4-kms-v3d dtoverlay line (that's already in distroconfig.txt).

    Better wait with further testing until RPi devs have a kernel fix, serial console won't help too much ATM as you'll only see the kernel is locking up before the vc4 hdmi line should appear (other than that a serial console is really helpful).

    so long,

    Hias

    deltasjimmy can you test if the display is working on RPi OS with the latest rpi-update kernel (currently 5.10.46-v7+)?

    I built an LE image with the missing drivers included, but it just locks up on my RPi3B+ when I add the vc4-kms-dsi-lt070me05000 dtoverlay - and the same happens on RPiOS with latest kernel as well. I don't have any DSI displays here so not sure if that's the cause - the DSI driver is still under heavy development so it could be an issue in latest kernel, too.

    so long,

    Hias