Please note, this is a bleeding-edge community OS for UDOO. It comes without any kind of warranty, so it is recommended only to experienced users. Using it, your board may be kidnapped, burnt or ate by dark kittens, or it may just produce magic smoke and flames. Use it at your own risk! Main features: * Based on Ubuntu 16.04 * Mate desktop environment * New virtual serial port between A9-M4 * Linux version is 4.1.15 * U-Boot version is 2016.09.1 * Integrated newer firmware and patches from NXP * Wifi and Bluetooth firmware updated * QT 5.5 with GPU acceleration * UDOO services ported to Systemd Known issues: * UDOO Bricks are using wrong permissions * Device Tree Editor has to be ported to Linux 4.1.x-2.x M4 core: mqx_upload_on_m4SoloX reports an error (the sketch is not running). The sketch should be running anyway, the uploader needs to be updated. Load the /dev/ttyRPMSG driver: sudo modprobe imx_rpmsg_tty The images are (almost fully) untested! In order to keep the forum clean, please ask questions about Ubuntu 16.04 for Quad/Dual only inside this thread. Keep in mind this image is not supported nor provided by the UDOO Team. Support tickets opened about this images will be discarded. Download: * 3.0 Beta 2 Previous versions: * 3.0 Beta 1 * 3.0 Alpha2
I just tried it, but it does not boot on my Neo Basic with this image I only get the pinguin logo, nothing else happens, only stable green light. Red light flashes shortly just after start but then stays off. Just installed Udoobuntu 2.1 on the same SD card and it booted without any issues. Do you want some extra data? What can I do to troubleshoot? Downloaded Image ZIP data: udoobuntu-udoo-neo-desktop_3.0alpha2.img Folders: 1 Files: 1 Size: 3487563776 SHA1 for data: 9BB8F9DD0ADF95F4FC4F4F423CE1D8D7F6838D7F SHA1 for data and names: DFBCFB2F720922F039AA050B0A22AF573D80EBBA
Please attach a serial-to-USB converter on the TX/RX pins on J7 (or connect the debug module) and post the boot log.
I managed to download the image now on my Linux server using the links browser. Would you mind if I post the link in case others have problems with google drive?
Here is the link: https://www.tiny-dev.com/udoo/udoobuntu-udoo-neo-desktop_3.0alpha2.img.zip MD5: d11bb7d0821312f7bd5176c67a371120 SHA1: ba3c77f223dc5d4c4e9069e78fc1e0cd22e10ef8
When I put the image on a 4GB SD card it works. When I put it on a 16GB it does not. When I put Udoobuntu 2.1 on the 16GB card it works. See the attached bootlog file of the alpha3 16GB. I suppose this is the interesting part of it: [ 1.359780] sdhci: Secure Digital Host Controller Interface driver [ 1.366011] sdhci: Copyright(c) Pierre Ossman [ 1.370385] sdhci-pltfm: SDHCI platform and OF driver helper [ 1.377252] /soc/aips-bus@02100000/usdhc@02194000: voltage-ranges unspecified [ 1.384471] sdhci-esdhc-imx 2194000.usdhc: Got CD GPIO [ 1.390764] sdhci-esdhc-imx 2194000.usdhc: No vqmmc regulator found [ 1.434989] mmc0: SDHCI controller on 2194000.usdhc [2194000.usdhc] using ADMA [ 1.450776] galcore: clk_get 2d core clock failed, disable 2d/vg! [ 1.460225] Galcore version 5.0.11.41671 [ 1.593626] mmc0: error -110 whilst initialising SD card
That's interesting! This errors come from https://github.com/fmntf/linux_kernel/blob/3.14-1.0.x-udoo/drivers/mmc/core/sd.c#L1252 If you want try to rebuild the kernel, you can add some `printk("i am here")` before each `goto err`, so you can see where it fails.
According Google error -110 should be a time out error. I don't understand why a different card would cause this. Perhaps the mmc driver off 3.10 is not compliant with 4? I am no Linux expert, never done any builds myself and have no ambition to become one also. Just tried your image to see if something changed dramatically. I tried the video performance, which is still the same (bad, 98% CPU while dragging a window). Is there a new video driver in the new kernel?
It's hard to say without looking the code carefully (and I did not do that). One simple thing is to do `meld linux-3.14/drivers/mmc/core/sd.c linux-4.1/drivers/mmc/core/sd.c` and see what changed. Maybe a meld on the whole `mmc` directory would be better, btw. The video driver is the latest available. Unfortunately the driver is not open source, so it cannot be improved. Consider that X11 is GPU-accelerated and videos can be played even in window-mode using only the GPU. But yes, moving the windows around the desktop will still use a lot of CPU. Something worth trying would be Etnaviv, which is an open source driver for Vivante. I had no time to test it yet.
Image boots fine on Neo extended, but there is no network interface whatsoever - ifconfig -a just shows a loopback, /dev doesn't have the corresponding devices - unlike the official 2.1 image. Boot log seems to be without relevant errors, it seems like there is no attempt to access the network interfaces. Is this expected? Am I doing something wrong?
That's not expected! I have got wifi and ethernet working. Please pastebin a full dmesg and systemd bootlog.
Okay, i've only checked the serial output, at least the systemd journal seems to have some grave errors. Here goes: http://pastebin.com/uZ9xE6mv http://pastebin.com/Ta7gfbH5 http://pastebin.com/Aad5fB1G Hope that helps!
That "wlan-en-regulator: disabling" looks really suspicious. We have to understand the reason the kernel is powering-off the device. Wlan is detected however ("mmc2: new high speed SDIO card at address 0001"). I have to compare your dmesg to mine.
Alright, but does this explain why no RNDIS daemon is run? Just that WiFi device being disabled because of... reasons is one thing, but why o why is there no trinity of mass storage/IP connect/serial console via USB either? :-S Anyhow, good luck!
Oh, I have forgotten to port RNDIS over USB to systemd! Try to run "/usr/sbin/g_multi.sh" on the tty console. Don't know if will work, btw.
Doesn't seem to work, goes to the endless loop at its end without any apparent changes happening. Haven't yet had the time to investigate, but without any network interfaces, an additional serial port is all I can hope for
Some insights after an hour of further investigation: Found out that wlan-en-regulator is a GPIO (WiFi on/off), then found the various GPIO export errors in the journal (which only happen after disabling wlan-en-regulator - but this seems to be normal at boot-up). Found a bug in /usr/bin/udoo-gpio-export, then saw this commit: https://github.com/ektor5/udoo-gpio-export/commit/4ae69557094b56c42232b59888eb1afe39e72bc5 Now, sadly, the fix doesn't change anything. I suspect a race condition or similar (triggered with my micro-SD card), maybe something with missing/wrong systemd unit dependencies and order?
