Posts by chewitt

Also, it is very easy to loose track of what works or not when people start posting about multiple devices in these threads. Please keep threads focussed on a single box.

If you want all 4x (S905, S905X/D) or 8x (S912) CPU cores to work then 1.5GHz is the maximum achievable stable CPU speed. If you reduce the number of active CPU cores you can (over)clock the remaining CPUs higher. Amlogic's 2.0GHz marketing can only be achieved with a single CPU core active, but this gives worse performance than all cores active at a stable 1.5GHz. Most vendor boot firmware lies and shows 2.0GHz at all times. Feel free to complain to your box vendor.

There are silicon bugs in the mmc controller on all Amlogic boards that impact on stability. There are workarounds implemented in the kernel, but the upstream kernel has rules on code quality so we aren't able to hack around problems as eggregiously as the vendor kernel sources do. SD card speed is currently set to 50MHz but can probably be raised to 100MHz reliably, maybe 150MHz. There are also additional SD card modes that can be experimented with. The vendor codebase uses 200MHz but that will not (100% guaranteed) be reliable.

The process for creating custom-remote configs documented in the wiki is still valid but the file format changed from .conf to .toml in a semi-recent v4l-utils update and the wiki needs to be updated to reflect that. Have a look at e.g. /usr/lib/udev/rc_keymaps/tanix_tx3mini.toml to see an example of the very lightly changed format.

SSH into the device and run "pastekodi" .. and then share the URL generated here

connmanctl
agent on
scan wifi
services
connect wifi_<macaddress>_hidden_psk

From fuzzy memory ^ use connmanctl and the hidden network will be listed with a name something like ^ that, and when you connect it will ask for the SSID and passphrase to use.

The log shows the WiFi card is probed and bin/txt files are loaded (there are errors for device-specific files but it then falls back to the generic ones which is normal). So there is a wlan0 device, but there are lots of reference to tethering. Have you enabled tethering in LE settings?

NB: The "BCM4330B1.hcd" file needs to be in the same firmware/brcm/ directory for BT to work.

In textmode make /flash writeable then dump the boot log to USB with "dmesg > /flash/dmesg.log" and shutdown and copy/share the log via pastebin so we can see if the WiFi card is probed or not.

ConnMan supports browsing and connecting to visible networks. If the network is hidden you will need to configure a manual connection by providing the SSID and passphrase to connect to. This is nothing new in LE11, the authors of ConnMan regard hidden SSIDs as pointless and even though it's pretty trivial to detect them Connman has intentionally never supported listing them.

elonesna please update to https://chewitt.libreelec.tv/testing/LibreE….arm-11.0.0.tar and see if this changes anything with the USB ports? (share pastekodi again pls).

If cold boot works and warm (re)boot does not work, it's either a PMIC (power management chip) or SD card voltage-on-reset issue.

ping knaerzche

Please share the "pastekodi" URL from the working USB install.

Same problem, different bus.

If you use a decent SD card not some prehistoric 2MB one the runtime experience will improve. There is also lots of scope to increase the SD card device speed via device tree if faster speeds prove reliable. That's always a gamble with Android boxes, which is why the card device in the currently experimental device-tree is using slower-but-safe defaults.

NB: I'd rather have dental work done without anaesthetic than engage in supporting emmc installs on Android boxes that we have no boot sources for. Under the legacy kernel it's a major pain in the arse to support but technically do-able. On an upstream kernel where the kernel cannot understand the partition scheme or see any of the partitions on emmc it's an entirely different game.

Quote from ajbathe

I changed KERNEL and SYSTEM of LE11 on raspi4 with the old kernel 5.x, but saw the same behaviour, so I don't guess it's kernel related. Maybe the reason can be in the fact that wpa_supplicant was kicked out and iw is used now? As long as you can't establish a connection via console commands (like above), the way over GUI won't do it, wouldn't it?

If you switched both KERNEL and SYSTEM you swapped Linux 6.1 + iwd (LE11) for Linux 5.10 + wpa_supplicant (LE10) .. which would imply the issue has nothing to do with the entire software stack.

It would be useful it people would state what network card make/model exhibits the problem. Also what router make/model you are using.

The LE11 log shows the wl driver is loaded which should indicate something was seen and probed on the USB bus, but no interfaces for the device show up. The LE10 log has rotated so we don't see driver load, but the wlan0 interface is clearly present.

Google searching doesn't reveal anything for wl under Linux 6.1 other than yet-another-patch but we have the same patch collection as other major distros are using. In short; the driver compiles and is present, but there's some other issue.

Run "pastekodi" and share the URL please

PhoenixBG please update to https://chewitt.libreelec.tv/testing/LibreE….arm-11.0.0.tar and see if that resolves USB ports? - and focus on the t95z-plus dtb not Q200/Q201 comparisons. To get the remote working you need to share a working remote conf or follow the instructions in https://wiki.libreelec.tv/configuration/…ration-advanced and share a listing of keycodes and notes on what keycode represents which remote key; from that I can create a keymap driver and preconfigure it.

I've seen this on my own board, and I think the cause is noise on the UART pins which is interpreted as a keypress (hence it enters u-boot). If you remove the UART cable does it boot normally?