In the past I have tried (meego's kernel), but with no success

As I had entered it on the wrong bug capturing system, I re-entered it just now on the right system. I also updated the text of the bug with some extra info: basically both (tiny) kernel patches didn't solve the issue. Oh well, it was worth the try.

new bug created in Nemo Mobile bugzilla system : 92
https://bugs.nemomobile.org/show_bug.cgi?id=92

Some update

Ali1234, the creator of the 'official' Nemo Mobile uboot package was willing to have a look at the problem, on my plea, as he was very close to the code before. You can follow the discussion in the irc chat log, starting yesterday http://mer.bfst.de/logs/%23mer/%23me...02-08.log.html 23:47. It continued early this morning till just now. Feel free to read along and understand the background and issues to resolve this problem.

Right now he/we're a bit on a dead end as the publicised code is not behaving as the binaries so it's very hard to analyse and simulate the problem.

Pali, can you help ali1234 along, to get him to compile the proper code that behaves like ur published binaries ? I think it's always good to have a 2nd or 3rd pair of eyes looking at a complex issue.

irc, #mer is a good spot to meet for now, I guess

Thanks.

tl;dr

I have rebased my n900-v2011.06 branch on to v2011.12 which uses the omap_hsmmc driver instead of the omap3_mmc legacy driver. You can find this in the branch n900-v2011.12 in my github repo. It boots the nemo kernel correctly.

I also tested the u-boot binary in the first post of this thread, and it gives the -110 timeout errors as reported.

I want to ask: If i put for example nitdroid will it be flashed (like multiboot does) or will it be loaded by u-boot

If it were working, NITDroid kernel would be loaded by uBoot, no flashing required, but I haven't managed to get it going yet.

e-yes told me that we have to modify init scripts to use uboot and nitdroid

Pali, will you update u-boot-bootmenu to use the said new mmc driver?

which new mmc driver? and where? I do not know that there is some new mmc driver in kernel...

the -more complete- version of the story ...

There is no new mmc driver in the kernel but there was a new generic mmc driver in the 'officail' u-boot code introduced over the last months. Pali is already using that one as the basis for his work.

The latest 'official' u-boot code (+RX-51 patches, as ali1234 made/described above) from the developers is booting Nemo fine. TEsted by ali1234.

So the Pali introduction of the bootmenu style on top of the 'working' code is breaking the init of the mmc, not allowing Nemo kernel to boot through.

Either by not completely initializing/resetting the twl4030 power convertor to power on the mmc subsystem as well or by not restoring/resetting the mmc registers the way the kernel expects them.

btw, Nemo is not the only kernel having this issue. Maemo5 had to include patches to get it working. The Kubuntu Mobile kernel (only one specific patched version of 2.6.35) is working fine with the Beagleboard (v v v similar HW). Later, unpatched kernel versions don't work. Same issue. Plenty of references on the gmane lists on this.

This can be solved (by smart people) on the Kernel side or on the u-boot side. Unfortunately the Nemo kernel people are not really interested, as they say it's mainly a u-boot (Pali version) fault and Pali says it's the kernel fault. I guess it's somewhere in the middle ...

btw, my Nemo Mobile - N900 Adaptation bug has been accepted and received status low.

Unfortunately ali1234 doesn't have access to the real code as created by Pali to help pinpointing where the issue might be on the u-boot side. Kernel side is v hard, as lots of patches I've tried for v similar issues (Error -110, TimeOut while waiting for a mmc answer), are not fixing the issue so far.

We asked Pali for that code so ali1234 can have a look as well and debug his way through if need be.

To be continued ??