I'm new to the BB. I went through the BB validation procedure fine.
However I'm now attempting to boot the latest Angstrom demo on a C4
board. I've setup a dual boot SDHD card that appears to be working /
mounting, etc. From there I was able to flash to nand, u-boot and MLO.
On boot, it loads the Angstrom uImage and eventually produces a kernel
panic shortly after loading the root fs from mmcblk0p2. Apparently
something with freeing init memory? The sequence is shown below.
Try 5V supply, but I have had panics caused by a mismatch between
filesystem and kernel. The kernel uImage.bin you put in the boot
partition should be the one from /boot in the Angstrom filesystem.
Strange things can happen otherwise.
If you boot with USB OTG not connected, is it more reliable? I had
trouble with the gadget composite driver which gets loaded by default
on that image. Seems it was causing a hang - but not this addres 0x14
issue. I have not seen that one. I also have a rev C4 board.
Unfortunately it is unlikely you will find much extra support for the
Angstrom image. The developers appear to have moved on and users are
encouraged to get into OpenEmbedded and built their own filesystem.
My best suggestion is to try power from 5V supply with USB OTG
disconnected from the host.
Did I mention a directive? I'm not sure of the point of contention here.
I'm going by what I see on this list. Admittedly I don't know the
Angstrom developers from a bar of soap so I'll admit to inaccuracy on
the point which I assume you're arguing about (Narcissus v
OpenEmbedded). However when I said "developers" I was not thinking
Angstrom developers, but those who I have visibility of on this list.
I cannot know their affiliation.
On this list I have not seen anyone take ownership of the angstrom
demo image. Who did build it? Who maintains it? AFAICT no-one has put
their name to it.
I have not seen anyone update its README to warn people of the common
mistake (I assume) of booting the beagleboard validation kernel
against the angstrom image despite a reasonable number of people
having that problem (me included).
In response to booting issues, several responses I have seen pushed
users toward later versions. Looked to me like no-one was supporting
2010.3. I think I can stick by my claim that developers have move on.
Watching the list traffic I worry that there is a schizm between the
experienced developers and beagleboard beginners. Please don't forget
what it is like to be starting out on a new platform. The beagleboard
angstrom demo is an excellent resource but it should not fall into
disrepair. It helped get me started quite quickly.
Personnally I have no clue about Narcissus. I've never used it and I
have no idea why I should and I'm not aware of any developer
preference towards Narcissus (which I promise I will investigate).
Most of the interesting information my research turned up has pointed
to OpenEmbedded. Has it been superseded?
For beginners Google turns up the beagleboard demo image and it looks
perfect when you know nothing about the platform and just want a place
to start. If there's something wrong with the demo image then I think
it should be fixed as a community service.
You're thinking like someone who knows what they're doing.
This is how it happened for me - roughly. I assume similar for other
beginners because I still don't see any logic error in what I did.
1. I have a board. How do I make it work? Beagleboard manual (SRM)
points to validation page [1, 1a] with kernel plus ramdisk.gz. OK
kernel boots, looks normal, great!
2. Now what? I need to find a more complete filesystem that is known
to work. Validation page [1] mentions beagleboard angstrom demo [2].
Sounds perfect!
3. It's a tarball to unpack into the rootfs partition. OK done. Let's
boot.... kernel panic. WTF?! I followed the instructions!
4. Email beagleboard list.
Beginners need a progression that leads them with confidence to the
point of building their own generic filesystems without requiring
list/group consultation and also establishes the project relationships
and requisite knowledge. I hope you will agree that this is an outcome
we want to promote.
IMO some work needs to be done to provide a guided learning curve.
I'll put my hand up to contribute some of that material. After all, if
you want to make a difference you have to put in the work; right?
So here's my proposal: the page pointed to by the beagleboard SRM for
board validation [1] points users to the angstrom demo page [2]. How
about I write a hundred words or so for beginners, then you (Koen
Kooi) review for correctness and utility and when it's ready it
replaces the existing ftp-style page. AFAICT the existing page text
come from the accident that a README exists. I assume that providing a
index.html would allow the HTML to be served instead.
I hope this will serve as a public service for beginners as well as an
education for me.
since i helped with some of the content at the latest validation
page, give me a day or so and i'll see if i can't tidy things up so
that it just *works*. i'll go back through this thread but, just to
be clear, what the validation page was meant to do was lead one
through the process by which one could get their freshly-unpacked
beagleboard up and running. is that the goal we're after here?
in any event, feel free to post what you think is wrong or missing
on the validation page(s).
The one for rev C4 is actually RevCValidationv3. Confused yet?
Some appropriate cross-linking and prominent directions to appropriate
pages are required. That will take care of the mess at code.google.com.
On the angstrom side of things, there aren't sufficient instructions
on the /demo/beagleboard page to safely guide a beginner through the
steps of unpacking the demo image, booting and verifying operation. It
is this latter element I was going to work on.
i contributed specifically to the ...v3 page. are you saying that
those instructions somehow don't work? if not, i'll make sure the
instructions are fixed -- if i have time today, i'll hook up my C4
board and run through it.
but go ahead and post what seems to work for you and i'll add
to that page whatever's missing.
No not at all. The v3 page is accurate and the files worked for me.
What I'm saying is that there is a confusing mess of older pages. If
you have a rev C4 board it is hard to find the right page because the
others exist and they do not all have cross-references to the v3 page
for rev C4 boards. I'd like to see all the other pages prominently
point to the correct page for rev C4.
I also have this problem. The thing missing in the thread is what
solves the problem? Is it power or is it the image? Could someone
just post which image and kernel play nice with each other? Thanks!
You should be fine if you copy the uImage from the /boot directory in
the root filesystem into the boot partition, instead of using the
uImage from the online demo directory.
Correct. The validation kernel does not work with the Angstrom demo
filesystem. You must copy the uImage.bin out of the tarball (from
/boot) and put it in the boot partition, replacing the existing
uImage.bin.
You also need to rename ramdisk.gz to something else.