hard disk gives trouble on change of size

[hendrik@topoi.pooq.com]

I have a 160G hard disk.  I had installed it in a mobile rack, and used 
it on a vary old machine that couldn't talk to hard disks of more than 
131GB (I think that's a power of two).  I partitioned all it could 
see of the disk (only 131GB, of course) as a single partition and used 
it without trouble as a backup drive for years.

Being in a mobile rack, I also used it on a machine that does recognise 
larger drives.

Now one of my machines (called lovesong) currently boots from ab 80GB 
hard disk that is slowly failing.  Yesterday came the time to reorganise my 
deployment of hard disks.  I plan to replace the 80GB failing drive with 
the 160GB drive mentioned above.  The first step would seem to be to 
copy my existing sarge to it after appropriate partitioning:

This new machine with the failing drive has no problem reading and 
writing the existing 131Gb partition.  fdisk recognises it as being a 
16oGB drive.  fdisk happily created two new partitions at the end of 
the drive.  They passed a bad-block check and an ext3 file system.  I 
copied my existing / partition from the old, failing drive using rsync, 
adjusted lilo.conf to be able to boot the new sarge as well as the old, 
and adjusted the new drive's /etc/fstab to recognise itself as 
containing the / partition.

When I rebooted, I was astonished how *slowly* it booted.  During a lilo 
boot, it writes a series of dots on the screen.  I'm used to them 
appearing and flashing off the screen faster than I can quite see them.  
But when booting from the new drive at /dev/hdc4, they appeared at about 
one per second -- about the same speed I'd expect from a *very* slow 
floppy drive.

And after the dots finished appearing, it did nothing at all.
Presumably it had not succeeded in loading a working kernel.

Now I remember seeing one message that I ignored some time through the 
process -- a warning that the kernel's drive geometry differed from the 
BIOS's.  I'm not sure what system component produced this message, nor 
its exact text.  But I'm so accustomed to the artificiality of drive 
geometries that I ignored it.

fdisk says:

lovesong:/farhome/hendrik# fdisk /dev/hdc

The number of cylinders for this disk is set to 310101.
There is nothing wrong with that, but this is larger than 1024,
and could in certain setups cause problems with:
1) software that runs at boot time (e.g., old versions of LILO)
2) booting and partitioning software from other OSs
   (e.g., DOS FDISK, OS/2 FDISK)

Command (m for help): p

Disk /dev/hdc: 160.0 GB, 160041885696 bytes
16 heads, 63 sectors/track, 310101 cylinders
Units = cylinders of 1008 * 512 = 516096 bytes

   Device Boot      Start         End      Blocks   Id  System
/dev/hdc1               1      266305   134217688+  83  Linux
/dev/hdc3          266306      288203    11036592   83  Linux
/dev/hdc4          288204      310101    11036592   83  Linux

Command (m for help):


Could its use as a small drive have poisoned it for use as a large one?
I've never noticed any lack of speed in the past, certianly not a factor 
of a hundred or so slowdown over normal disk behaviour.

Is the sarge lilo so old that it cannot handle drives with more than 
131GB or boot partitions after the first 1024 cymlinders?

-- hendrik