🚀 go-pugleaf

I have a problem with something I'm trying to do.  I haven't seen it
addressed anywhere; yet I would have thought that someone else would
have tried it by now.

We ran out of disk space on our Linux machine.  We got a larger disk: I
made larger partitions on it, copied everything over, and then tried to
make a LiLo boot floppy that works with the new disk.  So far, every-
thing I've tried has failed one way or another.

Details:

The system is actually a joint MS-DOS/Linux development machine, for
internal political reasons.  Because MS-DOS has less ability to under-
stand "different" partitioning schemes [or so I had thought], and so
that it boots when the boot floppy is removed, MS-DOS is on the first
partition.  The Linux boot partition is on the second partition, then
another Linux native partition, then a Linux swap partition.  A boot
floppy is used because (a) you can't install LiLo into a MS-DOS parti-
tion, and (b) if we take out the floppy, you can't boot into the Linux
side - this has some security or paranoia implications, I'm not always
sure which [;-)/2].

Linux is Slackware Professional 2.1.

Both old and new drives are IDE.  When we got the larger hard disk, I
made the old one MASTER and the new one SLAVE.  I booted MS-DOS; I
created a new, larger MS-DOS partition on the new drive; and I used
XCOPY to copy over all of the files.  This messed up attributes on some
files, and didn't copy over others; so I fixed those by hand.  As this
is an old FAT file system, and not the new Windows NT / Windows 2095
Unicode / Berkeley FFS clone file system, XCOPY worked fine for this
purpose.  I tested it by swapping the MASTER and SLAVE jumpers on both
disks, swapping the CMOS disk descriptions, and booting MS-DOS.  It
worked fine.  I then switched jumpers and descriptions back.

I booted Linux, set up Linux partitions on the new disk the same size
as the old ones, and set up one or two more.  Yes, this did mean that
partition 4 had to become "extended", and further partitions are
logical partitions over the extended partition.  I'm rather perturbed
that 'fdisk' complains about all non-logical partitions that have over
1024 cylinders.  It also prints out incorrect "begin" addresses for
partitions that start after 1024.  I had thought that I'd read that SWP
2.2 was the distribution with the bad 'fdisk'.  Anyway, I was able to
copy over all files properly.  I used 'find' and 'cpio', so that I
could do the copy one file system at a time.  This also worked fine.

Now I haven't figured out whether there is a way to generate a LiLo
floppy disk to boot from the newly formatted hard drive.  I have tried
alternate settings for "disk = ..." in the config file, but that seems
to upset the LiLo install program, since the configuration in the file
no longer matches the active configuration (on the old disk).  I've
also tried booting (using the old floppy) with the new drive as the
root disk; but this doesn't seem to help.  Everything I've tried seems
to either cause 'install' to abort, or creates a boot floppy that
believes in the disk partition sizes on the old drive.  In the latter
case, when trying to boot with the new drive as MASTER, I get just "LI"
... since the disk geometry on the new drive is different from that on
the old drive, and the LiLo primary loader can't find the root file
system, so it can't find the secondary loader.

I suspect that the originator of LiLo may have overlooked the possibi-
lity that the boot disk may have changed its name at boot time, from
what it is at LiLo install time.  I'm perusing the code to try to see
whether this is true, or if there's a way out.

Does anybody have any useful suggestions?  Thanks.

Joe Yao				jsdy@cais.com - Joseph S. D. Yao
-- 
Joe Yao				jsdy@cais.com - Joseph S. D. Yao