Shuttle kd20 - bricked

[Mijzelf]

Hm. Re-reading the wiki and some forum articles I think it's perfectly possible that the '1st-32MiB file' doesn't contain a partition table. For the EMC imager stick the box only needs to boot. The bootscript in the (original) initrd re-creates the partition table. So there doesn't need to be a partition table to start with.
You can just create one, if you first save the MBR code area. That is needed by the rather simple bootloader in the SoC:

Code: Select all

dd if=/dev/sdX of=/path/to/MBRCodeArea bs=446 count=1
fdisk /dev/sdX #(or parted /dev/sdX, if you prefer)
dd if=/path/to/MBRCodeArea of=/dev/sdX

[christiaan]

I've tried your code but it didn't have any effect.. Looking at the wiki page I tried the following different approach. It at least seems to leave the partition table intact, I'm however not so sure if the bootstuff is still perfectly fine.

Code: Select all

dd of=/dev/sda if=Iomega-HMNHD-CE-1st-32M.img bs=512 seek=34 skip=3

Edit: well that seems to keep the partition table intact but it does break the rest.
I've also tried booting it without a partition and just the magic stuff on it. The NAS hangs; so that confirms that it indeed tries to boot from the HDD!

[Mijzelf]

I've tried your code but it didn't have any effect..

No effect? THe idea is that you create a partition table with fdisk. How can that have no effect?

Code: Select all

dd of=/dev/sda if=Iomega-HMNHD-CE-1st-32M.img bs=512 seek=34 skip=3

Edit: well that seems to keep the partition table intact but it does break the rest.

Shouldn't the value of seek and skip be the same? An you still need the code area. No idea if that will be preserved by a partition editor.

[christiaan]

I've been going about this completely wrong. Rereading your answer apparently the MBR area contains code needed to boot.

Should I create a normal ms-dos partition table or a GPT table, because the Wiki also speaks of a GPT area.

I'll try your exact instructions this afternoon instead of misinterpreting it myself thanks!

Edit: My collegue found this topic about booting Arch Linux on the 782x entirely from Sata. The sector offsets etc all seem to be the same so I'll be trying that first.

[Mijzelf]

Should I create a normal ms-dos partition table or a GPT table, because the Wiki also speaks of a GPT area.

It doesn't matter. The data in the CodeArea makes the SoC to find the 1st stage bootloader on sector 34. the 1st stage bootloader finds u-boot @154, and so on.

The partition table is only important when Linux wants to read it. And Linux can read both GPT and MBR.

But of course, make sure the 1st partition starts at 32MiB+.

[wicked]

Hello,

Ive done:

Code: Select all

 zcat Iomega-HMNHD-CE-1st-32M.img.gz | dd of=/dev/sdb bs=1M
0+1024 records in
0+1024 records out
33554432 bytes (34 MB) copied, 2.95846 s, 11.3 MB/s
dd if=uImage of=/dev/sdb bs=512 seek=1290
5989+1 records in
5989+1 records out
3066408 bytes (3.1 MB) copied, 0.54459 s, 5.6 MB/s

and then I followed these instructions

rdimg.gz is actually a gzipped ext2 filesystem, with an uImage header. You can extract the filesystem:
Code:
tail -c +65 rdimg.gz | gzip -d >initrd
and mount it
Code:
mkdir -p initrd.d
mount -o loop initrd initrd.d
Now you can edit the bootscript (/rclinux) to setup the network, and start a telnet daemon....

and edited /etc/rc (added telnet daemon)
but I'm stuck (as a newbie) on how to make proper rdimg.gz from the mounted initrd.d. Can you pls poin me (does regular gzip will suffice?)

After that I have a question - if I'll manage to login via telnet, how to invoke firmware flashing ?

[christiaan]

but I'm stuck (as a newbie) on how to make proper rdimg.gz from the mounted initrd.d. Can you pls poin me (does regular gzip will suffice?)

I did it like this

Code: Select all

umount initrd.d
gzip initrd

mkimage -A ARM -O Linux -T RAMDisk -C gzip -a 61000000 -e 61000000 -d initrd.gz rdimg.gz.new

It seemed to work, mkimage gave the same header information on the old rdimg.gz file and this newly created one.

[Mijzelf]

After that I have a question - if I'll manage to login via telnet, how to invoke firmware flashing ?

If you manage to get telnet access, the script /rclinux contains the code to flash the files from KD20.zip. I think you can suffice with the ubi fs, as that is the only partition which should be editable from the webinterface.

[wicked]

but I'm stuck (as a newbie) on how to make proper rdimg.gz from the mounted initrd.d. Can you pls poin me (does regular gzip will suffice?)

I did it like this

Code: Select all

umount initrd.d
gzip initrd

mkimage -A ARM -O Linux -T RAMDisk -C gzip -a 61000000 -e 61000000 -d initrd.gz rdimg.gz.new

It seemed to work, mkimage gave the same header information on the old rdimg.gz file and this newly created one.

done that however Iwas expecting that NAS would obtain ip from dhcp and nothing happened.
I dod not do partition for logs as the MBR was faulty (as discussed earlier)

[christiaan]

Hm. Re-reading the wiki and some forum articles I think it's perfectly possible that the '1st-32MiB file' doesn't contain a partition table. For the EMC imager stick the box only needs to boot. The bootscript in the (original) initrd re-creates the partition table. So there doesn't need to be a partition table to start with.
You can just create one, if you first save the MBR code area. That is needed by the rather simple bootloader in the SoC:

Code: Select all

dd if=/dev/sdX of=/path/to/MBRCodeArea bs=446 count=1
fdisk /dev/sdX #(or parted /dev/sdX, if you prefer)
dd if=/path/to/MBRCodeArea of=/dev/sdX

I can confirm that this works. I've got a magic drive with a single partition on it as well now.

[wicked]

I can confirm that this works. I've got a magic drive with a single partition on it as well now.

are you using Shuttle KD20 or other oxnas based device?

[christiaan]

The KD20

[wicked]

so to sum up, my steps were:

Code: Select all

#make gpt partition table
zcat Iomega-HMNHD-CE-1st-32M.img.gz | dd of=/dev/sdb bs=1M
dd if=uImage of=/dev/sdb bs=512 seek=1290
#after whole mounting init.rd.gz, modifying /rc startup script:
umount initrd.d
gzip initrd
mkimage -A ARM -O Linux -T RAMDisk -C gzip -a 61000000 -e 61000000 -d initrd.gz rdimg.gz.new
dd if=rdimg.gz.new of=/dev/sdb bs=512 seek=16674

was your procedure any different ?

[christiaan]

so to sum up, my steps were:

Code: Select all

zcat Iomega-HMNHD-CE-1st-32M.img.gz | dd of=/dev/sdb bs=1M
dd if=uImage of=/dev/sdb bs=512 seek=1290
#after whole mounting init.rd.gz, modifying /rc startup script:
umount initrd.d
gzip initrd
mkimage -A ARM -O Linux -T RAMDisk -C gzip -a 61000000 -e 61000000 -d initrd.gz rdimg.gz.new
dd if=rdimg.gz.new of=/dev/sdb bs=512 seek=16674

was your procedure any different ?

Except for the fact that afterwards I did the

Code: Select all

dd if=/dev/sdX of=/path/to/MBRCodeArea bs=446 count=1
fdisk /dev/sdX #(or parted /dev/sdX, if you prefer)
dd if=/path/to/MBRCodeArea of=/dev/sdX

to create a partition on the disk. FYI the only thing I've got going on right now is the KD20 not booting up and hanging with just a blue light. And when I remove the disk it boots from the nand.
So I still have a long way ahead I guess, I've had trouble opening up the case otherwise I might have the stuff laying around to fix a serial cable.

[wicked]

...to create a partition on the disk. FYI the only thing I've got going on right now is the KD20 not booting up and hanging with just a blue light. And when I remove the disk it boots from the nand.
So I still have a long way ahead I guess, I've had trouble opening up the case otherwise I might have the stuff laying around to fix a serial cable.

and your disk has ms-dos or gpt partition table ?