[Ma-linux] partition weirdness

[Will Holcomb]

Thanks to everyone with their distribution suggestions in response to my
question last week. I figured out what was screwing up Fedora, but decided
to give Ubuntu a try anyhow.

The problem with Fedora makes no sense from my experience. There was an IDE
drive and a IDE cd on the first contoller and an identical IDE drive on the
secondary controller. For some reason the os though that both drives were on
the primary controller. I found this out when I was considering installing
Windows (to burn an install CD since I was booting Knoppix and couldn't
eject the DVD to use the drive). Windows listed both drives as on the
primary controller.

Well, I canceled the Windows install but not before it changed the partition
table on my second drive where I had all my data backed up.

I'm currently at a loss.

The file system seems ok. I tried deleting the windows partition table entry
and creating one for ext2, and I couldn't mount it. From opening the disk in
a hex editor, I can find the start of the filesystem by searching for the
ext2 label. On my current root partition the start is at 0x400. On the
deleted partition, it is at 0x6400. If I do:

dd if=/dev/hdc1 of=test.ext2 bs=1024 skip=$((0x6000/1024)) count=50000

I can then do:

mount -o loop test.ext2 test

I can't copy the whole filesystem though, it is 300gb and I only have one
other 300gb drive.

I tried creating the partition starting at cylinder 1, but it moved the
start 0xA1400 bytes. I noticed that fdisk had options for changing the disk
geometry. I was wondering what would happen if I changed that. Currently the
two disks which are the same model purchased at the same time have different
geometries:

Disk /dev/hda: 300.0 GB, 300069052416 bytes
255 heads, 63 sectors/track, 36481 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Disk /dev/hdc: 300.0 GB, 300069052416 bytes
86 heads, 15 sectors/track, 454319 cylinders
Units = cylinders of 1290 * 512 = 660480 bytes

I suppose worst case scenario is a clear off as much space as possible, copy
as much of the partition as possible, format the drive and the recovered
data back. This seems really silly though.

I suppose another option is to write a program to open the raw device and
try to shift everything back 0x6000 bytes. This also seems like more work
than ought to be needed. This disk worked before.

Anyone have a clue?

Will
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://calypso.tux.org/pipermail/ma-linux/attachments/20061221/7d79dfbc/attachment-0002.html