Posts by popcornmix

There is a test build here.

Place in /storage/.update and reboot to update. It's a Matrix build with some kernel debug options added.

By default it should behave the same as official release. But you can try some modifications by adding options to /flash/cmdline.txt

The default behaviour delays invalidating cec address when hotplug is deasserted for 1000ms (in case it gets reasserted).

That fixes the AVS off/on problem for me with a Panasonic TV and Onkyo reciever.

Try with vc4.inval_delay=5000 (added to end of /flash/cmdline.txt) to increase this delay further.

If that fails, then try another setting.

vc4.inval_delay=0 vc4.cec_debug=0x100

That will ignore the hotplug deasserted message completely. I'd be interested if either (or both) of these helps.

Is it failing to boot, or failing to display a picture? Can you ssh in?

That's the setting. Post a debug enable log when playing a problematic video.

Quote from tomstephens89

Logical next step is to bump the GPU frequency up a little, however if I even attempt to bump 'gpu_freq' up a mere 50Mhz to 550 then when I try to playback a h265 file, I will immediately lose video output but the OS doesn't crash and the Pi is still reachable via SSH and responds to commands so I can revert and reboot etc..

Disable "adjust refresh rate to match framerate" and I suspect you'll get a picture.

Your overclock is having the same effect as "hdmi_enable_4kp60=1" and something isn't supporting that.

Do you have another hdmi cable or 4k display to test with?

Quote from Baltasar

I have tried the additionof video=HDMI-A-1:D to the command line, and it kind of works

That suggests you don't have a working hotplug connection on HDMI. Try with a diffrent HDMI cable.

Quote from f3jeff

Thanks for the update. Regarding deinterlacing support on Pi4, do we need to enable it somewhere on the settings or it's already on?

No, it's automatic when playing an interlaced file.

Interesting. Nightly/master has moved to 5.15 kernel (compared to 5.10 kernel for Matrix) which is one possible explanation.

Probably a kernel display bug, but I'm currently confused.

I just noticed the file played fine for me initially (where I didn't have hdmi_enable_4kp60=1) then later on a different setup did fail.

Looks like the more common [email protected] video doesn't have this failure, but yours of [email protected] does.

I'm guessing the resolution is having some effect on clocks or timing.

I think this is a display issue rather than a video decode issue.

Azurewrath can you remove hdmi_enable_4kp60=1 from config.txt, reboot and try this file again?

Are you saying you get video out of both hdmi ports?

Or are you saying you can get video from one port and audio from the second hdmi port?

The original post read to me like they wanted video from both ports, which I don't believe can be done.

Video from one, and audio from the other can be done.

Never cut down a log. The most useful information in the log is at the start.

Sorry, that's not supported by kodi. You may be able to use a hdmi splitter to get the same result.

config.txt:
gpu_mem=384M
framebuffer_depth=32
otg_mode=1
hdmi_enable_4kp60
dtoverlay=hifiberry-amp100
MALLOC_ARENA_MAX=2
initramfs edid.cpio

gpu_mem=384M - Remove it. That memory is likely just being wasted.

framebuffer_depth=32 - Remove it - will have no effect when kodi is running

hdmi_enable_4kp60 - you need an =1 on end to do anything

MALLOC_ARENA_MAX=2 - Remove it. This will have no effect in config.txt

Have you added hdmi_enable_4kp60=1 to config.txt?

"dtparam=audio=on" added to config.txt will enable the headphone socket in audio settings.

It's happening in this tree: https://github.com/bavison/FFmpeg/commits/vc1_neon

Almost certainly the hard drive doesn't get enough power.

Externally powering the drive, or using a powered USB hub may fix this

If cpu usage is unexpectedly high, you can run "perf top", leave it to settle for a few seconds,

and the functions it identifies may give a clue to what it taking the cpu.