Hello,
I'm trying to launch a complete Angstrom Linux distro on my new rev B4
BeagleBoard.
I've followed very precisely and tightly the instruction for
partitioning my 2Gb SD, in order to have a FAT32 partition with the
kernel image, and a second ext3 partition with the whole filesystem.
Running the board and breaking at the boot within u-boot, I've seen
the existing u-boot ( 1.3.3-00035-gab55ae5-dirty ( June 16 2008 ) on
NAND contains already some intruction for having finding the root in
the right second partition ( which seems to be very good ), and the
bootcmd which does a mmcinit and loads the uImage kernel in a suitable
address.
When I run the whole stuff, the u-boot finds the uImage, loads it
telling it's a Angstrom/2.6.29/beagleboard. It then start the kernel,
which after the uncompression phase, it hangs on the "booting the
kernel" message.
I've tried many and many times, choosing different kernels, but this
always happens.
I've found on the net a post telling this sometimes happes when the
architecture machine nmber stored in the flashed u-boot on the board
doesn't match the machine version in the kernel.
My point is: could it be my board is quite old ( 2008 as you can see
from u-boot version ) while when I build the Angstrom distro the
created kernel is too new despite the u-boot?
I don't like using a too old kernel, so probably may you suggest me to
update the u-boot with the latest version, in order it matches the
kernel?
I dunno if this is the righ solution, it's just a hint. May you help
please?
Op 18 okt 2009, om 18:16 heeft riccardoventrella@yahoo.it het volgende geschreven:
Hello,
I'm trying to launch a complete Angstrom Linux distro on my new rev B4
BeagleBoard.
I've followed very precisely and tightly the instruction for
partitioning my 2Gb SD, in order to have a FAT32 partition with the
kernel image, and a second ext3 partition with the whole filesystem.
Running the board and breaking at the boot within u-boot, I've seen
the existing u-boot ( 1.3.3-00035-gab55ae5-dirty ( June 16 2008 )
That's waaaaay too old, please use the u-boot from angstrom.
That's waaaaay too old, please use the u-boot from angstrom.
You are absolutely right Koen, and I has noticed by myself
looking inside the scr files, I has simply to copy part of the
scr for overwriting that old boot, and now the kernel doesn't hang on
anymore.
However, I've still two problems:
The uImage is the latest one from angstrom live, i.e. the 2.6.29.
With this settings the kernel boots, but hangss when says "console
[ttyS2] enabled"
Sometimes, instead, the kernel panics.
The strange thing is if a break into the u-boot and doing that u-boot
command above
manually, all works correctly and the kernel boots until the login.
the same happens when I do a savenv of that commands on the very next
boot.
But from the second boot, the above hang appears again.
I thouoght it was a problem of the SD to be correctly recognized, so
I've increased
the rootdelay too up to 3secs, but nothing has changed, unfortunately
2) when the kernel boots correctly, it takes a very very long time to
reach the
complete boot up to the login, approximately 15/20 minutes.
Is it correct? I hope not. I've read the boot coudl be quite slow form
SD, but I've
never experience a problem like this before. If yes, are there any
workarounds?
I can recompile the kernel using less options ( now I even disabled X
window
from the angstrom live ), but I think the problem could persist.
Is there a way to put the kernel directly on NAND?