SSD on the Netrwinder

Ralph Siemsen ralphs at netwinder.org
Fri Mar 29 12:13:18 PDT 2019


Hi Jon,

Glad you got the jumper sorted out!

On Fri, Mar 29, 2019 at 05:54:48PM +0000, Jon Bradbury wrote:
>Another setback – the new partition is reporting as “wrong iNode size”. Must be something to do with the more recent formatting program I used (mkext2fs I think, but gparted calls it).

Yes, this is a bit of a common problem... the newer versions of mke2fs 
have new features enable, which the old kernel on the netwinder does not 
know how to handle. One such feature is "sparse_super", this was already 
a problem back in year 2000. I haven't checked, but there are certainly 
more such features added to ext2/3/4 since then.

One way around this is to use the old netwinder software to format the 
new drive. This would mean attaching both old and new drives (for which 
you would put the jumper on your new one back to "slave" mode). Then 
boot from the original drive, and use the mke2fs on there to format the 
new drive (which will be /dev/hdb then).

Or you can look at what options/arguments to pass to newer mke2fs to 
produce a filesystem that's compatible.

>So I have recreated it as a 1024Mb sized unformatted partition, then 
>used dd to copy the original boot partition to it. This gives me the 
>right format. Then, delete everything in the partition and unpack the 
>contents of the base-nw9.tar.gz archive onto it. This didn’t work, so I 
>went to the first partition (a working copy of the boot partition from 
>the old HDD) , did rm -rf * at the root and again unpacked 
>base-nw9.tar.gz onto it.

You might want to check the nettrom settings regarding where it boots 
from. There is one to tell it where to get the kernel, and another for 
the root filesystem. It sounds like they are both set to /dev/hda1, 
which would explain the behaviour. You can set them however you like, as 
long as it matches with the hard drive contents.

>This time it booted (with a picture of Tux at the top right hand corner 
>of the screen) and began checking the boot partition. Lots of errors, 
>but all fixed. The last error is /lost+found not found. CREATED.

If the old hard disk boots up without such errors (eg. no fsck) then the 
copy of it on the new drive should really behave the same.

>/dev/hda1: ***** REBOOT LINUX *****
>
>/dev/hda1: 32093/263160 files (5.5% non-contiguous), 245789/1048916 
>blocks

That's fairly normal when FSCK finds something... although it should 
generally reboot itself after that point.

>..and it has now locked up. There’s a flashing cursor though and it is echoing keyboard input but there’s no prompt.
>
>Any idea what it’s doing? I tried to reboot and it did all that all over again (including all the filesystem checks and fixes).

I'm wondering if the hard disk interface is actually working properly. 
Your earlier email, indicating hda timeouts being reported, are also a 
bit of a sign that something low-level may not be quite right. On the 
other hand, the fact that you can get as far as running fsck on boot 
suggests that things are working reasonably well.

Ralph



More information about the Netwinder mailing list