Cannot boot from NAND after reflash?

[moshiach]

Plugit, reboot and let it fail.  Scroll up a bit and near where it boots you will see the arguments passed to the kernel.  That's possibly where your problem lies.  Snip it and let's see what it shows.

[plugit]

This  bit?

Kernel command line: console=ttyS0,115200 mtdparts=nand_mtd:0x400000@0x100000(uImage),0x1fb00000@0x500000(rootfs) rw root=/dev/mtdblock1 rw ip=192.168.0.3:192.168.0.4:255.255.255.0:192.168.0.1:DB88FXX81:eth0:none

[plugit]

Interestingly, it does boot with the filesystem that shipped on the CD. Just not with the current one on the website. What's up with that?

I do get this, though:


FATAL: Could not load /lib/modules/2.6.22.18/modules.dep: No such file or directory
FATAL: Could not load /lib/modules/2.6.22.18/modules.dep: No such file or directory
 * Starting early crypto disks...                                        [ OK ]
FATAL: Could not load /lib/modules/2.6.22.18/modules.dep: No such file or directory
FATAL: Could not load /lib/modules/2.6.22.18/modules.dep: No such file or directory


(there's nothing in /lib/modules)

[moshiach]

Yes that bit.   One thing I notice right off the bat is the space between (rootfs) and rw.  That's probably not the problem but the kernel would look at that as two different config commands:  mtdparts and rw. 

Also your ip config statement for the kernel is incorrect, although since you are using DHCP it doesn't matter in this case,  but it  should be of the form

ip=<client-ip>:<server-ip>:<gw-ip>:<netmask>:<hostname>:<device>:<autoconf>

You have the gateway and netmask reversed and are missing a host name.  Something more like this:

ip=192.168.0.3:192.168.0.5:192.168.0.1:255.255.255.0:DB88FXX81:etho:none

But that also shouldn't be an issue here.   
Check that the kernel option for mtdparts is being set.  A little further down from the kernel commands should be a group of lines showing your partitions.  Make sure the name is nand_mtd and not orion_nand.  When I built my kernel the name changed on me.  You can tell it's reading it correctly if it doesn't say using static partitions.

Other than that, nothing jumps out at me.  Your bootargs are being passed to the kernel, albeit with some errors in them.  The real part that matters is the root= config and that looks good so long as it refers to a partition that mtdparts= is setting correctly.

As for the included image, the reason you get the fatal errors (which aren't really fatal) is as you've discovered the lack of /lib/modules.  You can safely ignore that since you will probably be building your own kernel and installing modules with make modules modules_install.

[plugit]

Ah! Thanks bunches for the info - I'll review this in the morning when my brain is less weird. I thought perhaps I may have been a bit off with my syntax (although interestingly, some of that syntax is as it was out-of-the-box, the space between the "rw" and the previous statement in particular).

There's definitely something different about the root image on the CD and the one on the website. They are different images (diff says so), and the CD one works for me and the website one doesn't. I've gone back and forth a couple of times without changing the uBoot parameters and the root image is the deciding factor in my current configuration.

Have you tried the website one?

Thanks again!

[moshiach]

I've tried so many different images my poor flash is all worn out.  Good thing I have two of them

Point me to a link for the image you are having trouble with.  There are too many floating around right now  to be sure.

[plugit]

This is the sneaky one:

http://www.marvell.com/files/products/embedded_processors/developer/kirkwood/SheevaPlug_FileSystem.zip

The bad blocks everyone's getting is a little bizarre. Hope we can figure out how to boot completely from SD.

Edit: Ah! The IP configuration doesn't complain anymore, since you fixed my syntax! OK, I'm off to bed.

Thanks for the help, everyone!

[Raúl Porcel]

The bad blocks thing is normal. I was told it was due to U-boot and the kernel or something. But it doesn't matter, you really don't want to touch the U-boot partition

[alexbez]

That should be ok. 

You should rewite the uImage and the file_system.jff2 make sure you use

flash_eraseall -j /dev/mtd1
flash_eraseall -j /dev/mtd2

nandwrite -pm /dev/mtd1 uImage.sheeva.20090319
nandwrite -pm /dev/mtd2  ubuntu-9.0.5.Release.jffs2

and make sure You use the same mtparts definition in all you different bootargs

Mark

Indeed, I did flash_erase and flash_eraseall before doing nandwrite and I made sure that mtd partition definitions are the same.
Now I am 99% sure that the ubuntu-9.0.5.Release.jffs2 image, taken from the "Download" section of the WEB site, is broken.
Today I am going to check the image from the companion CD (I should have tried that before! But the rule "Online image is always more up-to-date than on a supplied CD" played a bad joke on me, I guess).

On a side note: I tried to mount the older, 9.0.4 jffs image (using mtd emulation on a block device), taken from the site - and it mounted allright and contained all the necessary stuff inside. So the next experiment will be to flash that one and to see if it works.

Will keep you posted on the outcome of these experiments.

[moshiach]

I flashed the rootfs plugit linked to and came up with a ton of errors as well.  Unless someone can verify it's working, I would say use the older rootfs.

[Blazer]

You have the gateway and netmask reversed and are missing a host name.  Something more like this:

ip=192.168.0.3:192.168.0.5:192.168.0.1:255.255.255.0:DB88FXX81:etho:none

That "etho" should be "eth0". I'm certain that you know that but just making sure as some folks just cut and paste

[plugit]

I flashed the rootfs plugit linked to and came up with a ton of errors as well.  Unless someone can verify it's working, I would say use the older rootfs.

Thanks for giving it a go! My goodness, I spent probably 8+ hours on that yesterday, convinced that it couldn't be a problem with the rootfs.

Never mind! Onward and upward!