Posts by kszaq

dodo95 cat /sys/class/amhdmitx/amhdmitx0/disp_cap > /storage/.kodi/userdata/disp_cap when your TV is connected directly to the box.

Quote from crubyape

I see reference to the Nougat kernel. Apologies if it's clear to some but is it correct to say the references are to Amlogic kernels used in Librelec for these builds because there is no mainline kernel support for these chips?

Mainline kernel is available but without hardware video decoding, so not very useful for us.

Quote from crubyape

I see discontinued at the top of this thread, is it in order to work on Librelec 9.0 Alpha builds? It's not clear.

This simply means that I will no longer work on these builds. I might release an update to 8.2.3 but without any additional Amlogic-specific updates or improvements.

I am currently working on improving Amlogic support in LibreELEC and Kodi master.

I will do this later today with a PR to LE master.

Edit:

GDPR-2 Done: qca9377-firmware-aml: change firmware version to 1.0.0-3 by kszaq · Pull Request #2420 · LibreELEC/LibreELEC.tv · GitHub

IIRC linode hosts the latest firmware binary which is 1.0.0-4. You may have compared it earlier and it matched but when they updated deb from 1.0.0-3 to 1.0.0-4, tar.gz was also updated without changing file name. I can repack 1.0.0-3 deb to tar.gz and put it in my sources so that LE master package.mk can be updated with 1.0.0-3.

The reason is that my builds use 1.0.0-3 firmware and build from GDPR-2 uses 1.0.0-4 which is known to cripple WiFi performance.

Compare:

LibreELEC.tv/package.mk at libreelec-8.2 · kszaq/LibreELEC.tv · GitHub

LibreELEC.tv/package.mk at master · LibreELEC-AML/LibreELEC.tv · GitHub

If you are using a build from GDPR-1 or me (arm userspace, not aarch64) restoring should work fine. In other cases you'd have to reinstall binary addons.

Or simpler explanation: try it, if it doesn't work, you can make a clean install on Odroid again.

sakos fw_setenv hdmimode 1080p60hz and reboot.

This is the main difference that makes 8.2.x builds shut down longer:

kodi: wait longer for Kodi to stop · kszaq/LibreELEC.tv@25ff796 · GitHub

This means that in 8.2 builds I extended a period of time in which Kodi has to shut down (or else the process is killed). This might affect CEC shutdown function - if Kodi is killed before it fully shuts down, it won't be able to send CEC shutdown command.

GDPR-2 You may want to pick that commit for testing.

Quote from Tidav7578

The [Discontinued] info is because Kszaq stopped working on the project? or is it because that branch is stopped or just a mistake?

I will not provide any updates to these builds as I don't have resources (disk, CPU and time) to maintain Krypton and work on Leia at the same time.

Shoog I used my build with device tree provided. But I had a list o pre-scanned channels configured in TVHeadend. I'm sorry I cannot help further, I have almost zero knowledge on DTV stuff.

vpeter I think the issue is with how big tmpfs can be. With ~120MB LibreELEC SYSTEM image this is not an issue but when you add emulation stuff it's over 500MB - it looks like it's not allowed to make tmpfs that big, especially that Amlogic devices reserve ca. 200MB for video decoders and that leaves you with 800MB free RAM for everything else.

Shoog Yes, it's the same and I think I have run DVB-T successfully with that board using the driver developed by afl1.

straylit Device tree overlays do not work with legacy Amlogic u-boot and kernel as far as I know.

What you can try is edit boot.ini and add this line right before "bootm prep": fdt set /spicc status okay

This should enable SPICC node and load the driver. Whether it works and how - you tell me.

mxqproject You can change cmdline in kernel config. Look for CONFIG_CMDLINE parameter.

algdu83 Amlogic 4.9 kernel doesn't provide any video decoder interface that we can use for Kodi on Linux.

ADB = Android Debug Bridge, it doesn't work with LE. SSH is a better console than ADB in my opinion.

Brissieboy Turn of automatic updates in LE settings and downgrade to 8.2.2.1 to see if it helps.

thomas output_rgb is to force RGB colorspace instead of YPrBr, not to switch from limited to full range. Before you ask: I don't know how to do the full range switch.