[yoshiMember]

Follow up: Since that build, the snapshots don’t crash anymore.

There is a problem now with the WiFi interface (mlan0) on geexbox-devel-20150208-r4fbae6f.utilite.tar it is not listed at all.
and on geexbox-devel-20150304-r43066cc.utilite.tar, there is a really large delay (4 seconds and more) with pings from WiFi.

GeexBox 3.1 doesn’t have these issues.

[yoshiMember]

Quick update. XBMC is not shown (stuck on the GeexBox logo), not really sure what’s going on there.
Anyway, this is way closer to working than kernel crashing every second:)

[yoshiMember]

Sorry for not having to get around to it sooner!

Good news! This is the first time since 3.1 that Geexbox builds did a successful boot!
I don’t have time now to test it thoroughly, here’s a little output from the serial:

[ OK ] Started FTP Server.
[ OK ] Started Network Service.
[ OK ] Started Load/Save RF Kill Switch Status of rfkill0.
[ OK ] Started Bluetooth service.
[FAILED] Failed to start Create basic config for NetworkManager service.
See ‘systemctl status nm.service’ for details.
Starting Network Manager…
usb 1-1: reset high-speed USB device number 2 using ci_hdrc
usb 1-1: reset high-speed USB device number 2 using ci_hdrc

Welcome to GeeXboX arm devel-20141228-r9159724 (C) 2002-2014
geexbox login: usb 1-1: reset high-speed USB device number 2 using ci_hdrc
fec 2188000.ethernet eth0: Freescale FEC PHY driver [Generic PHY] (mii_bus:phy_a
geexboxusb 1-1: reset high-speed USB device number 2 using ci_hdrc 1-1: reset high-speed USB device number 2 using ci_hdrc t
Password: imx-hdmi-audio imx-hdmi-audio: HDMI Video is not ready!
imx-hdmi-audio imx-hdmi-audio: ASoC: can’t open platform imx-hdmi-audio: -22
imx-hdmi-audio imx-hdmi-audio: HDMI Video is not ready!
imx-hdmi-audio imx-hdmi-audio: ASoC: can’t open platform imx-hdmi-audio: -22

imx-hdmi-audio imx-hdmi-audio: HDMI Video is not ready!
imx-hdmi-audio imx-hdmi-audio: ASoC: can’t open platform imx-hdmi-audio: -22

BusyBox v1.23.0.git (2014-12-27 19:28:36 CET) built-in shell (ash)
Enter ‘help’ for a list of built-in commands.

usb 1-1: reset high-speed USB device number 2 using ci_hdrc
root@geexbox:~# usb 1-1: reset high-speed USB device number 2 using ci_hdrc
usb 1-1: reset high-speed USB device number 2 using ci_hdrc
usb 1-1: reset high-speed USB device number 2 using ci_hdrc
usb 1-1: reset high-speed USB device number 2 using ci_hdrc
usb 1-1: reset high-speed USB device number 2 using ci_hdrc
usb 1-1: reset high-speed USB device number 2 using ci_hdrc
…
(these messages are shown repeatedly every 1 second)

Anyway I think this is really great news!

[yoshiMember]

Every snapshot I’ve tried since July didn’t work.

Compulab just released a new image with a new kernel. Are there any plans to release an official build? use Compulab’s kernel?

Right now I’m still fixed to GeexBox 3.1 which has some problems and I suspect they can be resolved with a newer kernel / version.

[yoshiMember]

Any updates? I really want a new image.

I’ve encountered a new problem with GeeXBoX 3.1.
Although it is stable (no crashes so far) after a while it is powered on (happened twice consistently after 24 hours)
Video started stuttering and sound chipping – the video is unwatchable.
Restarting xbmc doesn’t solve the problem, and top shows no other processes hijacking the cpu.

I will mention the video is on an external hard-drive, but after reboot the video works smoothly..

Any ideas?

[yoshiMember]

Hmm.. Really interesting! Which snapshot did he try? I’ll try the same one..

I wonder if it’s related to Utilite Standard vs Pro (in theory they differ only in the number of cores, and of course the lack of SSD)

BTW – I wish there was some consistent crash. It just hangs most of the time near the Welcome, I guess it starts running services… (I can also differentiate the hangs, because input to console stops echoing back.. i.e. Enter “not responding”)

[yoshiMember]

Okay, so I’ll test your snapshot when its ready:)

[yoshiMember]

I’m not quite sure, but doesn’t this: https://github.com/tomlohave/utilite/commit/f34da00fba3369fd4beb10e608299b149af6966c
implies that the kernel is the same version as Compulab’s ? (http://utilite-computer.com/download/utilite/kernel/linux-image-imx6_3.0.35-cm-fx6-6.1_all.deb )
It says the commit was 15 days ago – does that means the snapshots I’ve tried (from the 16th) already had this kernel?

@tomlohave – Could you write down the output of “uname -a” from the latest snapshot?

[yoshiMember]

wolfgar is Stephan Rafin?

I would love to test the new Image that you’ll make. Can you verify that it uses the newest kernels from Compulab’s site?

[yoshiMember]

@tomlohave – Is it possible for you to confirm that with a newer U-Boot the snapshots do not work?
I would also love to actively try to find and patch the problem, but I’m not sure where to start (especially since there is no consistent crash I can trace back to)

[yoshiMember]

News!

I’ve managed to boot Compulab’s Ubuntu images – the problem was I forgot to apply the kernel update procedure (as specified here: http://www.utilite-computer.com/wiki/index.php/Utilite_Linux_Installation_and_Update )

Since the symptoms were very similar in both cases (Ubuntu and the geexbox snapshots), I think it is a combination of the kernel with my U-boot (@tomlohave – you said you have an older U-Boot)…

So all is left now is to identify the changes between the kernels and pinpoint the exact problem (and apply the patches).

[yoshiMember]

So I’m clueless to what the problem is, and how to debug it.
I guess I could try and boot up a kernel with a debugger, but:
1. Replacing the kernel could result in a successful boot or in additional non-related failures.
2. I’m not sure exactly how to do that (where to get / how to compile the kernel and upload it to the SD, and how to change the bootargs to allow the debugging).

[yoshiMember]

@rudi I haven’t compiled the kernel myself. only used the tarballs / images in the snapshots – So I guess they have the patches (as tomlohave said).

I’ve started a related thread in Compulab’s forum, regarding the Ubuntu images, which doesn’t seem to work either: http://utilite-computer.com/forum/viewtopic.php?f=64&t=1891

@tomlohave The fact that it works for you reinforces the possibility that it is some specific hardware issue (or maybe an issue with the specific U-Boot version installed on my Utilite). It is very weird though, that GeeXBoX 3.1 works.

[yoshiMember]

So I’ve tried again.. Now with the make-sdcard script and the latest snapshot (geexbox-devel-20140616-r72f5a68.utilite.tar.xz) and I got something a little different.
The kernel constantly crashes.. here are the last 3 crashes:

FEC Ethernet Driver
Unable to handle kernel NULL pointer dereference at virtual address 0000041a
pgd = c0004000
[0000041a] *pgd=00000000
***********************
FEC Ethernet Driver
Unable to handle kernel paging request at virtual address feffddfc
***********************
Unable to handle kernel paging request at virפual address 0cdd3000
pgd = c0004000
[0cdd3000] *pgd=00000000
Internal error: Oops: 5 [#1] SMP
Modules linked in:
CPU: 0 Not tainted (3.0.35 #1)
PC is at tick_program_event+0x2c/0x4c

I’ll trying playing with other snapshots to see whats up.

[yoshiMember]

For those who are interested, I wrote a solution that works and does exactly what I thought in the previous post, which means it doesn’t “suspend” the system, only halts xbmc and blanks the screen.
(This can also be considered an advantage because it is still accessible via ssh and can be to do other tasks while on “standy”)

1. keymap.xml that maps “power” key to a script which calls “systemctl stop xbmc” and blanks the screen (by writing 1 to /sys/class/graphics/fb0/blank)
2. Service that runs in the background waiting for “power” key and then turns on the screen and starts xbmc, by calling “systemctl start xbmc”.

The reason for the external service is that I wanted to kill xbmc, since it was overloading the CPU on idle.
And the reason the solution was not implemented only in that service is that xbmc opens the keyboard event file exclusively, which makes the service “blind” to any key presses when xbmc is running.

[Author]

Posts