Posts by chewitt

I suggest upstream translators don't do a half-arsed job with their translations. It's the correct and darn sight easier approach

QED

flirc_util is packaged for more conventional distros. I'm far from being an expert package creator, but I'll have a look into what's required to repackage it into a binary add-on for LE use. This will take me "some time" though (lots of other priorities at the minute) so it's probably easiest if you find a Windows, macOS or other Linux box to reprogram the USB dongle from.

If we remove the word from our anti-spam list we get lots of nice chinese/korean people advertising their solution for preventing the mail loss of filament structures that come out of your head. The occasional user inconvenience is easier to live with than regular spam deleting.

Quote from izno69

I tried to search a folder "update" to paste the .tar but the folder not exist.

The folder is /storage/.update not /storage/update .. the . hides the folder unless you view hidden files.

FYI Add Wireguard support

From ConnMan devs:

"ConnMan resolves the FQDN on start of the VPN service and hands over the IP address to the WireGuard device. The WireGuard kernel code doesn't do any resolving. I am going to post a patch which reresolves the Endpoint address. If the address changes ConnMan will 'update' the WireGuard device. I've stollen this idea from reresolve-dns.sh « reresolve-dns « contrib - wireguard-tools - Required tools for WireGuard, such as wg(8) and wg-quick(8)"

There's also a second patch to remove "WireGuard.domain" from the config as "Domain" isn't really needed. NB: After this patch is included the internal service identifier will change needing you to correct any systemd files.

Patches => [PATCH 0/2] WireGuard/VPN small tweaks. - connman - Ml01.01.Org

If some of you are self-building images .. it would be good if you can test these patches. Please provide feedback to this thread and if there are positive reports on the DDNS improvement I will ack the changes on the list so things get merged upstream (and then added to the LE codebase).

linux-amlogic/gxbb_p200_1G_wetek_hub.dts at amlogic-3.14.y · LibreELEC/linux-amlogic · GitHub

^ dropping that from the device tree should/might do the trick. LE moved all development away from the legacy kernels to mainline so you won't find anything from us. Hopefully everyone being locked up at home for a while stimulates the ffmpeg work to be finished off..

I'm not reacting to length of cable, more to the fact that you changed cables.

50m cable runs are starting to get long and you're not connecting two powered switches; one end is a lowly Pi board. So what happens if you go back to using the old (shorter) 30m cable?

Copy/paste touch /storage/.config/safemode.disable on the command line .. done.

If you're seeing the same issue on two different OS images it looks like the device has a failed SDIO wifi module; although the BT section (which is on the same module) looks working which is unusual. I guess if you're only using Ethernet it's not a big deal.

Boot times on mainline are way lower and I find the entire OS to be lighter and more focussed. The only thing that's still "work in progress" for an S905 box is hardware video playback which still needs some rework after recent kernel-side changes. If it works for you, you're welcome to use the nightly images. If not, stick to 3.14 for a while longer. WP2 will be broadly similar, the only exception is that 4K might work as it has 2GB RAM so the currently insane CMA memory allocation needed for that is possible vs. the Hub which is 1GB and needs to wait for some memory use optimisations.

Quote from rabdallah

Are there any settings I can adjust to fix the built in wifi throughput?

As a rule software cannot fix a hardware problem. Most internal antennae are an afterthought. Connecting a proper external antenna will fix reception issues, but that probably requires a dremel and some handicraft on the NUC case

RPi3 remains the benchmark for 3D support. RPi4 has more oompf but a different SoC/drivers and many things with a higher priority for development effort than 3D.

It's hacked into meson-gx.dtsi so that all GX devices inherit the same config. See:

HACK: set meson-gx cma pool to 896MB · chewitt/linux@3c771f1 · GitHub

I've reported this to connman devs already .. the Jira is Intel staff only and hardly used. The mailing list is the other place to raise issues.

--- a/projects/NXP/devices/iMX6/linux/linux.arm.conf
+++ b/projects/NXP/devices/iMX6/linux/linux.arm.conf
@@ -861,7 +861,8 @@ CONFIG_IP_PNP_DHCP=y
 CONFIG_NET_IP_TUNNEL=y
 # CONFIG_SYN_COOKIES is not set
 # CONFIG_NET_IPVTI is not set
-# CONFIG_NET_FOU is not set
+CONFIG_NET_UDP_TUNNEL=y
+CONFIG_NET_FOU=m
 # CONFIG_NET_FOU_IP_TUNNELS is not set
 # CONFIG_INET_AH is not set
 # CONFIG_INET_ESP is not set

^ that probably fixes it