Igor

Random MAC issue in kernel was fixed some time ago. Wrote on mobile phone

U-boot. If u-boot doesn't properly bring up the Ethernet it won't work. I made one boot test with fresh SD card. In first try it failed to initialize but than I cut power (reset might not be enough) and boot again without a problem. Wheezy with kernel 4.1 from the post above. It's working for me. If you won't be able to boot this image we need to figure out why.

H3 is in the same position. There is no community support (yet) for kernel 3.4.x which means you need to use previously mentioned (Official) Allwinner SDK. Current result is poor so I don't want to build images using that kernel. Way too much bugs for a single person or a small group to fix and support. For mainline kernel (kernel.org) there are some patches for basic functionality but haven't got time to test. https://patchwork.ozlabs.org/project/devicetree-bindings/list/?submitter=65627

Kernel 4.1 for all A20 boards: http://mirror.igorpecovnik.com/kernel/4.1.0-bananapi-next.tar (tested quickly. temperature readings are false which is the only noticed bug) Uboot is only for Banana! Don't install on other boards. Kernel 3.4.108 for Banana & Orange http://mirror.igorpecovnik.com/kernel/3.4.108-bananapi-default.tar (tested quckly) Uboot is only for Banana! Don't install on other boards. Kernel 3.4.108 for Cubieboards, Lime1-2 http://mirror.igorpecovnik.com/kernel/3.4.108-cubietruck-default.tar (tested quckly) Uboot is only for Cubietruck! Don't install on other boards. I started today with long term stability stress test. Configuration: Cubietruck Wheezy 3.4.107 on NAND, Wifi connect Banana Jessie 3.4.108 Olimex Lime Jessie 4.1 Build today, using default configuration + some dummy stress scripts.

I never done exact scenario. In theory it should work. You have done (too) many steps where things can get wrong Especially avoid distribution upgrades. I made a quick Wheezy build with latest 4.1. Untested. https://mirror1.igorpecovnik.com/Cubietruck_Debian_4.0_wheezy_4.1.0.zip

This builds only basic desktop environment and it's not fine tuned. It's a good start but far from plug and play. On both. Server is minimum requirements. Yes One option is also to get an prebuild image an build a desktop of your wish on it. For hw acceleration you have to use kernel 3.4.x, Wheezy is recommended. Check here: https://github.com/igorpecovnik/lib/blob/next/boards.sh#L229

It's working on some A20 boards. I don't have the answer 'why' yet Wrote on mobile phone

It's possible to build such image with my script, also with HW acceleration. I made one (private) test build with Wheezy about month ago. Since this part is experimental it might need further fixing. I can't say for other images - which is cool, working, supported, HW acc. enabled. Don't use them.

http://www.denx.de/wiki/view/DULG/WarningBadCRCUsingDefaultEnvironment The whole problem (I think) is Ethernet initialization which is done in u-boot. This should be fixed in last edition but since there are still reports of problems I can conclude that we don't share exactly same hardware? It's possible in a quality manner. But anyway this should be fixable since it was working well all the time. I don't think it's related. Nothing has been change system wise. If your Ethernet is not up you can login via console eventually. I'll recheck again what's going on but as I said Ethernet is working o.k. on my board.

Amazing I'll look into and fix.

Mainline doesn't have script.bin any more but Device Tree which has similar role but completely different structure. Sources are located in /arch/arm/boot/dts/ ... compiled under /boot/dtb Documentation: http://events.linuxfoundation.org/sites/events/files/slides/petazzoni-device-tree-dummies.pdf

No rush, take your time. I don't know how complex this issue is but I assume it is. So don't be frustrated if it doesn't go at first ..10x try. I am planning to write a full manual during my holiday so it will be simpler to use & develop further.

Can you describe the exact scenario. Have you power off your board in between?

Unfortunately no. A31 has only one kernel which needs a lot of re-work to become standard which is necessary for my script. Even if this is done the end result will be poor. Kernel is old and buggy, A31 hardware is abandoned and not many people work on the kernel if any. There is basic support in mainline kernel (kernel.org, I think 3.19 and newer). I could boot it but not even SD card was operational ... so I abandon the project and maybe in the autumn I'll try again and rework the board configuration, together with H3 and A80.

And here are also working on this problem: http://forum.solid-run.com/viewtopic.php?f=8&t=1573&start=400 I am just about to leave office for some time so this fix will wait or if someone take this issue and prepare a pull request @Github. I'll have limited working ability for some time.

OK, I can test the DVB too but no rush. I won't be able to do much during summer time - a lot of "out of office" time

You can have 20A PSU but still it wont be enough. It's a design fail. Micro USB allows cca. 1.8A @5V and this is a bit tight. I only need to plug in my HDMI2VGA converter and HD doesn't spin, USB devices sometimes fail, ... Maybe I should pin this topic http://forum.armbian.com/index.php/topic/15-lamobo-r-1-sata-power/?p=168

You need to change wifi module to brcmfmac and you need to use different hostapd binary and configuration. More info: https://github.com/igorpecovnik/lib/blob/next/documentation/general-faq.md#how-to-setup-wireless-access-point

Wait that first run finish the job, than login - it takes few minutes. Wrote on mobile phone

You mentioned: 7 LEDs triggered by the system events So I thought you are using GPIO header. The numbers are different . Regarding power supply. There are some topic on this forum and else where this problem is worked into details. R1 has a design fail regarding power supply. If you want to use board in full scale than you need to fix powering. Micro USB has it's own power limits and they are less then what's needed for everything.

Added that to, thanks Horst. https://github.com/igorpecovnik/lib/commit/282fac50d704a02621f04471cc46c69f2612dc10

Yes and no. I do many compilation and sometimes it does not build up, than in few days - when things are fixed - everything is fine. Is this a problem of debootstrap, packets, ... I don't explore. Jessie is the problem winner, than Trusty and Wheezy which is somehow logical. I am among those few % who notice such things. Yes, Jessie become stable. With minor, probably irrelevant, glitches. But if you can't boot the system due to bug which is most likely from os ?

There is a hardware walk around for powering SATA. GPIOs. On new kernel you will need recalculate: http://linux-sunxi.org/GPIO

Short version of answer is: to avoid questions - "why do we have a high load without running anything?" More specific: http://www.cubieforums.com/index.php/topic,1084.0.html

GPT is supported, that's not a problem. What do you have if you type: