770Z using the full 14G drive

[Rob Mayoff]

+---------- On Jul 17, Mark Ball said:
| Regular fdisk was failing with "unable to read".  This turned out to be
| the sector read for the 3rd logical partition, the first one to start,
| end, and be entirely contained above cylinder 1024.  Take your pick
| at which one is the cause.  (Seemed to me to be the first, but I'm an
| idiot.)

The cause of the fdisk failure is not having the disk geometry set. I
know this because I recompiled fdisk with debugging and then stepped
through it in gdb when I got that error.

If you access the whole-disk device (/dev/hda) and try to seek past what
it thinks is the end of the device, you will get an error (or maybe
llseek just returns zero, I don't remember which). The extended/logical
part of the partition table is scattered across the disk. It's a linked
list with nodes stored at the beginning of each partition, and when
fdisk tried to seek out past 1024 cylinders to read the next node (the
one for your third logical partition), the seek failed.

However, fixing the geometry only lets fdisk work correctly. Disk
Druid **ALSO** has a problem, I believe, with partitions above the
1024-cylinder mark. It sounds from your and other posts like it only has
the problem with logical partitions.