Posts by chewitt

mglae can you link to any upstream discussion on this?

The main two variables are wifi drivers and firmware. You can see if LE10 nightly images (with newer bits) are better than LE 9.2 images, but if not there is not much we can do. See: Index of / for images.

Andre Bollo BT in the GUI is managed by the LE settings add-on which is in the process of being rewritten for Python3; and BT is still one of the bits that's being worked on. If you SSH into the OS you can use "bluetoothctl" to pair devices etc. - until the GUI work is completed.

[ 41.084656] usb 1-1.2: USB disconnect, device number 3
[ 41.085934] sd 0:0:0:0: [sda] Synchronizing SCSI cache
[ 41.286978] sd 0:0:0:0: [sda] Synchronize Cache(10) failed: Result: hostbyte=0x07 driverbyte=0x00

^ I see this which triggers the "removal" notice that you see, but 10 seconds later the drive remounts and "mount" shows it as mounted read-write which indicates the filesystem is okay (else it would mount read-only or fail).

The only odd thing that I can see is this from blkid:

PTTYPE="atari"

but I'm not sure that's the reason..

The length of a passphrase string determines its ability to resist brute forcing. Once you've gone much beyond 16 chars the amount of compute time needed to guess the passphrase is so insanely huge that there's not much point in having a longer one. It's also largely irrelevant because attackers rarely brute-force, they focus on finding flaws in the crypto implementation; which is how WEP, WPA and WPA2 are normally beaten.

Anyway.. if you cannot change it, access via SSH and (from memory):

connmanctl

agent on

scan wifi

services

connect <identifier of service>

copy/paste the string when prompted

Purchase is done. Martin is rather amazingly persistent with Meson8 support. I'd liken the reverse engineering effort to working a crime scene where you already know whodunit but there are few clues. The older BSP kernels are full of magic values so there's a lot of detective work involved.

Ahh, ok, it will need to be in /storage/.config/firmware then not the mediatek subfolder. We need to replicate the bad location in the upstream repo.

Martin found one earlier .. and he will purchase (and we will refund the cost). It will still need some detectivework so I wouldn't expect quick results; the older Amlogic boards have more variance between devices than current boards which are usually a quick copy/paste exercise.

Err, my bad .. MT7601U .. I updated the URL in the above post. Those folders are overlayed on /lib/firmware at boot time so that firmware appears in the correct location.

Sources are in /storage/.kodi/userdata/sources.xml .. so SSH into the box and use the nano text editor to edit the file, the format is quite simple to work with and I find some copy/paste in nano is much easier than using a remote to set host/share info in the GUI.

Current plan for WeTek Core support is to find one on eBay somewhere in the EU and ship it to Martin. I also have a box, but 3.10 device trees are so radically different from what I'm used to (device tree was a very new invention back then) that I honestly don't quite know where to start

mkdir -p /storage/.config/firmware/mediatek
cd /storage/.config/firmware/mediatek
wget https://git.kernel.org/pub/scm/linux/kernel/git/firmware/linux-firmware.git/plain/mediatek/mt7601u.bin
reboot

^ SSH in and run these commands. The kernel has the MT6701U driver enabled but it complains about firmware. The error reads like it's complaining about the wrong firmware, however I can't see any sources for the firmware included in the RPi2 image so I have a hunch it's a badly written error and it's complaining about no firmware. Running the above commands should add it to an overlay folder so (on reboot) it appears to be in the right location for the driver to find it. No harm in trying..

The fix commit is in Linux 5.10.0 which means LE10 images (e.g. current nightlies) already have it included. There are currently no plans to make another LE9.2 release, but if that should change I've added it to a list of things that might need fixes backporting.

Since a while now Kodi defaults to SMBv2 which means Kodi on Linux does not support "browsing" and you must manually create a source using fixed host and path information. Kodi on Windows is different so any comparison is irrelevant.

new wiki creator says "about time someone other than him learned to do stuff in the new wiki"

documentation/.gitbook.yaml at master · LibreELEC/documentation · GitHub

^ stuff can be added there

Please share "dmesg | paste" and "lsmod | paste" URL(s) so we can see what audio hardware is seen by the kernel and what drivers are loaded.

Software problem.

LE can overlay files in specific dirs onto the read-only filesystem at boot time. Look in /storage/.config/pulse and pulse-daemon.conf.d.