Very Computer

I have Linux installed on a 1.6G drive (WD 31600) (on /dev/hdc).  The
first partition is a 480M DOS partition.  After that are the Linux
partitions.  The BIOS and Linux fdisk correctly detected the "true"
CHS geometry (16 heads, 63 sectors, 3148 cylinders).  I have the BIOS
set for LBA mode and "Large DOS Compatibility" (which I think is the
INT13 CHS translation).  Everything has been working fine since I
installed it in September.

I'm going to replace the 1.6G with a 2.5G (WD 32500) (on /dev/hdb).
The BIOS again detected the geometry (16 heads, 63 sectors, 4960
cylinders).  I created a 480M partition under DOS as before (with DOS
fdisk).  When I went to create the Linux partitions (with Linux
fdisk), fdisk shows the geometry as 128 heads, 63, sectors, 620
cylinders (ie, three bits have been shifted from the cylinder field to
the head field).  I created the Linux partitions and copied all the
files over from the 1.6G (using cpio), and everything seems to work.
I can access the DOS partition, which I consider to be a big test.

Is the different geometry in Linux fdisk going to cause problems?
What'll happen if I create another DOS partition after the Linux ones?
I can force the geometry in fdisk of course, but should I?  Why is a
1.6G's CHS handled differently from a 2.5G?  Does this have to do with
the bug some BIOSes had regarding disks over 2.1G? (I got an update
that supposedly fixes my BIOS, BTW.)

I'm at the point where I can afford to*up the new disk, but I've
got to send the old one back to Western Digital in a couple weeks.
Although I'd prefer not having to repartition and recopy the files if
I can avoid it.

This is the version of fdisk that came with RedHat 3.0.3 (same kernel,
too; 1.2.13, I think).

Thanks.
--
John A. Breen         | I teleported home one night with Ron & Sid & Meg.

                      |                            - Douglas Adams

I just had a horrible thought.  Could this be due to something like
EZ-Drive having been previously installed on the disk?  Would that
persist even if partitions were completely recreated?

Also, here's another tidbit.  It's not just fdisk.  Here are the lines
from dmesg:

hdb: WDC AC32500H, 2441MB w/128KB Cache, LBA, CHS=620/128/63,
    MaxMult=16
hdc: WDC AC31600H, 1549MB w/128KB Cache, LBA, CHS=3148/16/63,
    MaxMult=16

--
John A. Breen         | I teleported home one night with Ron & Sid & Meg.

                      |                            - Douglas Adams

Quote:>hdb: WDC AC32500H, 2441MB w/128KB Cache, LBA, CHS=620/128/63,
>    MaxMult=16
>hdc: WDC AC31600H, 1549MB w/128KB Cache, LBA, CHS=3148/16/63,
>    MaxMult=16

I had another thought.  These two disks are on different IDE
interfaces.  I seem to recall something when I originally installed
the 1.6G about there being two drivers: hd.c and the newer ide.c.
Could it be that the older one is being used on the primary and the
newer one on the secondary?  Would this explain why hdb is doing CHS
translation and hdc isn't?  Will this cause me problems when I
upgrade to a newer kernel?
--
John A. Breen         | I teleported home one night with Ron & Sid & Meg.

                      |                            - Douglas Adams