Mele A2000 - can't boot anything but Android

Quote

uboot-multi-allwinner is forked from uboot-allwinner,

No. You are wrong. It is absolutely NOT forked from uboot-allwinner!

Where the hell did you get that idea? Have you even read the uboot-multi-allwinner wiki?

uboot-multi-allwinner is forked from lichee. It has absolutely no code from uboot-allwinner.

Unfortunately, due to problems integrating the NAND and MMC code, the SPL does not yet build. The SPL only is from a mirror of uboot-allwinner, and I will not touch the uboot-allwinner code since it a mirror that has nothing to do with the uboot-multi-allwinner project.

If there is an issue with the uboot-allwinner SPL then it is separate from uboot-multi-allwinner and must be patched in uboot-allwinner.

The a10linux / uboot-multi-allwinner github page states it as being forked from hno/uboot-allwinner so it's pretty easy to see how someone could get that impression.

Quote

patched multi u-boot with patched spl : is working

What I need results for is unpatched uboot-multi-allwinner with patched uboot-allwinner SPL.

. . . also please try out unpatched uboot-multi-allwinner NAND branch, which does not use an SPL since it is NAND boot. Just back up the existing u-boot.bin in /boot/linux to u-boot.bak. Use default environment and do not save the env or load anything from conf.d. That will keep your existing env so you can then uninstall it after you test.

If neither uboot-multi versions work then it is uboot-multi-allwinner which should be patched. If both work then it is uboot-allwinner code that must be patched (unless you or someone else can get uboot-multi-allwinner SPL working) and I can't do anything except update the mirror once uboot-allwinner is patched. I can't do anything with the SPL branch as it is a mirror.

Quote
gnexus
No. You are wrong. It is absolutely NOT forked from uboot-allwinner!

Where the hell did you get that idea? Have you even read the uboot-multi-allwinner wiki?

uboot-multi-allwinner is forked from lichee. It has absolutely no code from uboot-allwinner.

Dude, you need to chill out or take a nap, i'm trying to help here.
I have burn a week nights to learn and find the u-boot and sata bugs.
I was just reading your github:
a10linux / uboot-multi-allwinner forked from hno/uboot-allwinner
hno / uboot-allwinner forked from amery/uboot-allwinner

I will do the test on Monday if your post didn't convince anyone to do the test before me...

Regards.

Sorry. I know you're trying to help.

But, considering it took a very extensive amount of work for me to try and get the NAND working in uboot-allwinner before I finally gave up, I don't want people to think I'm using any of that code, and I will get highly upset, as you noticed, if people do.

The only uboot-allwinner code I'm using is the SPL, and I'm trying hard to get that eliminated too. The other code is only the original Allwinner Lichee code. In my opinion that is what should be worked with. The NAND code in uboot-allwinner is totally missing, and after all the changes to uboot-allwinner it can no longer be easily merged with the original Lichee NAND code. Unless people only want to boot from MMC, and not have SATA, USB, NET or NAND support, we need to be moving away from the current uboot-allwinner MMC code.

I was hoping that by creating the git repository that people would help to patch the code, and get the SPL working, and also help with native PHY, SATA and USB support like kirkwood has. To be frank, the device manufacturers should have already implemented it, and I really do not appreciate needing to do their work. But I want my device working the way it should. I am more than happy to share the changes to help others. But I only have a limited amount of time and only two A10 devices.

So far the community support is not happening. That makes the frustration level double what it would otherwise be. If people want uboot-multi-allwinner to continue to improve they need to help - even if testing is the only thing they do.

Ok, it's an early christmas gnexus.
I had to go to my office... long story, so i made the test.

My test package here: http://db.tt/yoScNhuc
I have generated both uboot and uboot-multi in vanilla and patched version.

And the results are (in the makefile):

t1: splvanilla ubootvanilla
@FAILED

t2: splvanilla ubootmvanilla
@FAILED

t3: splvanilla ubootpatched
@SUCCESS

t4: splvanilla ubootmpatched
@SUCCESS

t5: splpatched ubootvanilla
@FAILED

t6: splpatched ubootmvanilla
@FAILED

t7: splpatched ubootpatched
@SUCCESS

t8: splpatched ubootmpatched
@SUCCESS


it's in u-boot.bin and not in the spl.
The patch should be pushed everywhere, and you have to regenerate your u-boot-multi package. Sorry.

For the record, i used first a new other v1.7... working like the v1.3 does and passing all the t1/t2 successfully...
I verified first all my compilation. no error.

I took another one and bing... it was failling.
They have exactly the same chips on them: Nand, Ram, etc..
It could be a lot of things like:
- the A10 more or less pure/error-free a10...
- a power regulator in the board more or less accurate.

The patch correct this issue and it's the most important.

Quote

The patch should be pushed everywhere, and you have to regenerate your u-boot-multi package. Sorry.

No need to say sorry. . . :)

It's not that I did not want to patch multi. That will need to be done a lot anyway. It is that I want to know exactly what changes produce what results. That way I can better understand the allwinner u-boot when going forward with my own changes. That helps everyone. Thus your testing helps everyone.

I also don't want to apply unneeded patches. That just complicates further updates. Also, as stated before, I can't patch the SPL branch. That must be done by updating the mirror. So good to know the SPL is not too crtitical.

Thanks for your extensive testing. I will try to get the patch implemented ASAP, along with the upcoming Linux on NAND support.

First, I wanted to say thanks to Guillaume for all his hard work. I'm using his packaged boot files and it's the first time I am getting consistent successful boots on my A2000 v1.7. Before I found this thread I spent a whole week trying out different setups and only got a few successful boots that I couldn't reproduce.

Now I need SATA support...

To gnexus - thanks for your work on uboot-multi-allwinner, too... really looks like that's where support for booting from SATA will come from.

You said you only had two A10 units and neither is a v1.7 Mele... so that made me think, since those things are really not very expensive, how about some community support in the form of a donated A1000/A2000?

Hello itavor,

Nice to hear it's helping. Not a big deal, i got the trick from Tom Cubie for the uboot.

And SATA is here:
http://forum.doozan.com/read.php?6,9597

and the discussion around the patch here (modified PLL6 clock or disabled USE_PLL6M_REPLACE_PLL4, both works)
https://github.com/amery/linux-allwinner/issues/40

gnexus is going to include the uboot and sata patch in his next release.

Regards,

Thankyou so much guillaume and gnexus, that help us a lot :)

Quote

gnexus is going to include the uboot and sata patch in his next release.

The NAND version has a preliminary patch for the issue using guillaume's/hipboi's fix. The next MMC version, and the final NAND version, will be patched using hno's code. I need tester's for all of the u-boot muti versions so that the fixes can be incorporated into updated SD images on their next release.

You can also use hno's git. His latest code has fixes for this issue. But I would ask that users also test the latest u-boot multi. It has additional features that many people want, including booting from NAND, SATA, and USB.

Hi Guys,

How can I tell my Mele is V1.7 or not?
I got 6 Mele A2000, one of the have a sticker on PCB board: HMC-g32A01 V1.1, and all others are HMC-G32A01

Thanks,
Uceegoor what

HMC-g32A01 V1.1 is version 1.1. The others may be 1.0. I think 1.7 is marked thus, but only Guillame could say for sure.

Try to boot Debian. If it works then it does not matter what version it is. If it does not boot then use use linux-sunxi uboot SPL, or even better use the NAND uboot from here, and then please post your results. In all cases you should have a serial console connected since that, and SATA, is the main reason for buying a Mele.

gnexus Wrote:
-------------------------------------------------------
> HMC-g32A01 V1.1 is version 1.1. The others may be
> 1.0. I think 1.7 is marked thus, but only Guillame
> could say for sure.

My two A2000s are both marked 1.7 with white stickers on the main board..

-- Inge

All versions should work nicely now.

The story around v1.7 is quite simple.
It does not seem related to the sticker but, by side effect, to the version of the A10 (VersA,B,C).
Latest version (VersC) has a problem on one of the clock making the uboot and sata corrupting the kernel stack.
It was generating a kernel Oops for the uboot and was making the sata unable to discuss with the disk.

All these problems are now gone.
I have v1.7 all around working perfectly (and the lastest lastest version with the antenna soldered on the board, working too).

I want the schematics of the AB10-G32-V1.20-2 WLBM821-2400120-92 2011-12-20 ,could any one help for me ,please any one knows the schematic s please mail me.

thanking you