Thank you for such a quick response. I have tried your suggestion in
comment 5 above. However, the shortened overlay name (vc4-pi5) did NOT
resolve the issue on my system. I can see on the display if I do a power
cycle that when it reads from the NVME and reads the config.txt file it
does use the shortened vc4-pi5.dtbo

Your bootloader logs show "Loaded overlay 'vc4-pi5'" at 7.27 seconds,
which is great - the overlay loaded from the bootloader's perspective.

Can you verify that you are using Ubuntu Core 24 and the snaps are
stable? Didn't know if maybe you were on an edge version or not

However, after my system boots with dtoverlay=vc4-pi5,cma-128:
- ALL device tree nodes still show status="disabled"
- /dev/dri/ does not exist
- vc4 module loads but never binds (usage count = 0)
- dmesg has ZERO vc4 messages
- ubuntu-frame still fails with "No DRM devices detected"

Questions:
1. On YOUR system with the vc4-pi5 overlay, what does this show?
```
   cat /sys/firmware/devicetree/base/soc/hdmi@7ef00700/status
   ls -la /dev/dri/
   lsmod | grep vc4
```

2. Does your system actually have working graphics/ubuntu-frame with
this workaround?

3. Is the vc4-kms-v3d-pi5.dtbo file in pi-kernel 1047 known to create
disabled device tree nodes? Should it set status="okay" by default?

Without vclog available on Ubuntu Core, I cannot verify if the
bootloader actually loaded the overlay on my system, but even if it did,
the nodes remain disabled.

Could this be a bug in the overlay file itself, rather than just the
path length issue?

-- 
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/2128161

Title:
  pi-kernel VC4-KMS not working with CM5

To manage notifications about this bug go to:
https://bugs.launchpad.net/canonical-kernel-snaps/+bug/2128161/+subscriptions


-- 
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs

Reply via email to