Re: Revisit Large Hard Drives
I've been playing with a couple of boxes with large drives on over the
weekend and here are some more data:
1. Dell box with 20GB Maxtor and Phoenix BIOS. The box came with a
single 98 partition which we resized with FIPS to 16GB Win/4GB
Linux. Neither the slink nor potatoes cfdisk liked the partition
table one bit (a), but slink's fdisk was able to change the
partition type to Linux and after that the system installed
happily. I couldn't get LILO to boot Linux (b), but the guy who
owned the machine liked the idea of a Linux bootdisk lest his
girlfriend booted into Linux by mistake.
(a) The error message was something drastic sounding; something like
`Unable to read valid partition table, perhaps it's a new drive,
shall I write a new one for you ?'. Cowardly I wanted to protect
the existing Win partition and declined this kind offer.
(b) I didn't try anything beyond running the menu option from the
slink installation disk.
2. A mongrel box with two 10GB IBM drives and originally an Intel
motherboard also using a Phoenix BIOS. In the past I've not managed
to see more the 8GB of the 10GB drives, but didn't ever try too
hard---in particular I didn't try specifying the disk geometry on
the kernel command line. Similarly LILO also proved intransigent,
probably because I didn't have a small boot partition at the start
of the disk. I swapped the motherboard for an ABIT BX6 with sports
the Award BIOS and everything started working: cfdisk now sees the
10GB drives as 10GB rather than 8, when I added an new 18GB drive
it saw that as 18GB too, LILO works flawlessly. The Award bios
offers three choices of hard drive mapping: NORMAL, LBA, and
LARGE. It suggested LBA, and without a good reason to do otherwise
I followed its advice.
The conclusions:
1. ABIT boards make large drives a doddle.
2. cfdisk with Phoenix BIOS/large disks doesn't work, but fdisk does.
3. There might be some scope to make things work better with the
Phoenix BIOS by specifying the drive geometry to the kernel or to
fdisk.
Cheers,
--
Martin Oldfield,
Reply to: