[Angstrom-devel] BeagleBoard (C3) won't boot

[Koen Kooi]

Op 29 mei 2011, om 14:18 heeft Gary Thomas het volgende geschreven:

> On 05/29/2011 05:46 AM, Koen Kooi wrote:
>> -----BEGIN PGP SIGNED MESSAGE-----
>> Hash: SHA1
>> 
>> On 29-05-11 13:34, Gary Thomas wrote:
>>> On 05/28/2011 07:00 AM, Gary Thomas wrote:
>>>> On 05/28/2011 05:45 AM, Koen Kooi wrote:
>>>>> (Adding beagle group to CC:)
>>>>> 
>>>>> Op 28 mei 2011, om 12:46 heeft Gary Thomas het volgende geschreven:
>>>>> 
>>>>>> On 05/28/2011 04:35 AM, Koen Kooi wrote:
>>>>>>> -----BEGIN PGP SIGNED MESSAGE-----
>>>>>>> Hash: SHA1
>>>>>>> 
>>>>>>> On 28-05-11 12:07, Gary Thomas wrote:
>>>>>>>> I just built Angstrom via oe-core for the BeagleBoard, rev C3 (not
>>>>>>>> xM)
>>>>>>>> Here were my steps:
>>>>>>>> $ git clone git://git.angstrom-distribution.org/setup-scripts
>>>>>>>> angstrom-bb
>>>>>>>> $ cd angstrom-bb
>>>>>>>> $ git checkout -b local-oe origin/oe-core
>>>>>>>> $ MACHINE=beagleboard ./oebb.sh config beagleboard
>>>>>>>> $ MACHINE=beagleboard ./oebb.sh bitbake console-image
>>>>>>>> 
>>>>>>>> It doesn't boot at all. I built a bootable SD from the deploy/images,
>>>>>>>> including MLO and U-Boot. Here's the boot sequence (I broke in to set
>>>>>>>> the console):
>>>>>>>> Texas Instruments X-Loader 1.4.4ss (Jan 29 2011 - 10:54:03)
>>>>>>> 
>>>>>>> That should say 1.5.0, but wouldn't cause boot problems.
>>>>>>> 
>>>>>>> For reference, this is what I get on my C3:
>>>>>> 
>>>>>> It seems like every major component (MLO, U-Boot, Linux kernel)
>>>>>> differs between
>>>>>> my build and yours.
>>>>>> 
>>>>>> So what did you do differently? Why is my build so different (and
>>>>>> not work)?
>>>>>> I used the stock conf/local.conf - did you make any changes to it?
>>>>> 
>>>>> You're right, my FAT partition was a mess, I recreated the card using
>>>>> OE-core built stuff, log is further below. The one change I have is
>>>>> this one:
>>>>> http://patches.openembedded.org/patch/4677/ which might explain the
>>>>> problems Gary is seeing, but not the problems I'm encountering.
>>>> 
>>>> I'm trying this now with that patch applied. So far, I can already see
>>>> signs
>>>> of change (fewer strange warnings while parsing recipes, etc). I'm
>>>> doing a
>>>> complete build from scratch, so I won't know the results for a few hours.
>>> 
>>> Sadly, I'm still getting the same results even with this patch.
>>> 
>>> I'm still curious why my components are different (MLO=1.4.4 vs 1.5 for
>>> you, etc)?
>> 
>> It's probably loading from nand, do a 'nand erase.chip' from inside uboot
> 
> I pressed the user button to force it to read from MMC and see the
> correct date.  Here are the package versions that were built:
>  u-boot-v2011.03+git1+b29fbb347698286935bfc401c08499a6f63479de-r3
>  x-load-1.42+r13+git1+6f3a26101303051e0f91b6213735b68ce804e94e-r13
>  linux-omap-2.6.39-r0
> 
> I am confused by this though.  For example, that version of x-load is
> coming from
>  sources/openembedded-core/meta/recipes-bsp/x-load/x-load_git.bb
> not (as I think yours is)
>  sources/meta-texasinstruments/recipes-bsp/x-load/x-load_git.bb
> It would seem to me that the build should come from the meta-texasinstruments
> layer which has priority=10?  The list of all packages built by the process
> I've described (above) is attached.  How does it line up with what you've
> built and why the difference(s)?

You might have an old copy of the setup-scripts, which revision are you using? I check on my 3 buildhosts and they all build the correct MLO, but I can imagine that using an old bblayers.conf would seriously mess that up.

regards,

Koen
regards,

Koen