Document HDMI-CEC feature disparity between full and fake KMS

[cihlarma]

With the Bullseye update, full KMS is now the default graphics driver as opposed to fake KMS in Buster.

From config.txt in 2022-09-22-raspios-bullseye-armhf.img:

# Enable DRM VC4 V3D driver
dtoverlay=vc4-kms-v3d
max_framebuffers=2

From config.txt in 2022-09-22-raspios-buster-armhf.img:

[pi4]
# Enable DRM VC4 V3D driver on top of the dispmanx display stack
dtoverlay=vc4-fkms-v3d
max_framebuffers=2

The full KMS driver seems to not have the same level of support for HDMI-CEC as the fake KMS which automatically turns the TV on and changes to the appropriate input source without any other configuration, something I discovered this the hard way after I upgraded to Bullseye. Nowhere in the official documentation was I able to find a single mention of this, what saved me was a user post on the forums:

This function is not supported when using the now default kms video driver. dtoverlay=vc4-kms-v3d
If you want that function as before then you need to change to the fake kms driver.
In config.txt change dtoverlay=vc4-kms-v3d to
dtoverlay=vc4-fkms-v3d

Please consider documenting this lack of automatic HDMI-CEC functionality in full KMS somewhere at least until this feature disparity is resolved, assuming such resolution is even technically possible; announcing this in a news item like you have done with the Widevine issue with the 64-bit OS version would be good.

[JamesH65]

Rather than changing the docs, it really should be fixed, so have you reported this as an error on the kernel repo?

[cihlarma]

By 'kernel repo' I assume you mean the 'linux' repository as opposed to the 'firmware' one? I have not reported it in either repo; since a regular forum user not denoted as 'engineer' with a badge as RPi engineers are in the forums was already aware of the issue, I assumed this was already known internally and was on the engineering's team to-do list as part of bringing the FOSS driver to feature parity with the proprietary firmware as they have been doing over the years. I don't pay much attention to RPi development, I had simply assumed this was general knowledge that had been passed around on forums and was forgotten to be added to official docs due to its ubiquity.

[JamesH65]

There is no guarantee that a forum post will make it to the issues repo. If its not an issue in the repo, then it could go missing. And yes, the Linux repo is the one.

[aallan]

Also related #2310.

[github-actions]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.