Different geometry settings for identical drives

I'm having some IDE troubles on 2.4.21-rc3 on amd64 platform. The problem is I've got these two drives:

hda: WDC WD1200JB-75CRA0, ATA DISK drive
blk: queue ffffffff80442aa0, no I/O memory limit
hdc: WDC WD1200JB-75CRA0, ATA DISK drive
hdd: MATSHITA CR-177, ATAPI CD/DVD-ROM drive
blk: queue ffffffff80443238, no I/O memory limit
ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
ide1 at 0x170-0x177,0x376 on irq 15
hda: attached ide-disk driver.
hda: host protected area => 1
hda: setmax LBA 234441648, native 234375000
hda: 234375000 sectors (120000 MB) w/8192KiB Cache, CHS=232514/16/63, UDMA(100)
hdc: attached ide-disk driver.
hdc: host protected area => 1
hdc: setmax LBA 234441648, native 234375000
hdc: 234375000 sectors (120000 MB) w/8192KiB Cache, CHS=232514/16/63, UDMA(100)
hdd: attached ide-cdrom driver.
hdd: ATAPI 24X CD-ROM drive, 128kB Cache, UDMA(33)

which are configured identically in the BIOS and recognized identically at bootup, but the geometry is getting setup differently:

prime:~# hdparm /dev/hda /dev/hdc | grep geometry
 geometry     = 14589/255/63, sectors = 234375000, start = 0
 geometry     = 35906/16/63, sectors = 234375000, start = 0
prime:~# cat /proc/ide/ide0/hda/geometry 
physical     232514/16/63
logical      14589/255/63
prime:~# cat /proc/ide/ide1/hdc/geometry 
physical     232514/16/63
logical      232514/16/63

The result is that hda works fine but hdc doesn't. When I try to mke2fs on the latter I see:

hdc: dma_intr: status=0x51 { DriveReady SeekComplete Error }
hdc: dma_intr: error=0x10 { SectorIdNotFound }, LBAsect=234441583, sector=232343808

You can see that LBAsect (234441583) is higher than the "native" sectors quoted by the kernel (234375000, difference 66583 sectors). Why are these two disks being addressed differently?

IDE controller is AMD 8111.

Cheers,
jwb

A FAQ. Nothing wrong about that.
See http://www.win.tue.nl/~aeb/linux/Large-Disk-14.html#ss14.2
"Nonproblem: Identical disks have different geometry?"

All this output is rather nonsensical. Geometry does not exist, and various pieces of software print various meaningless numbers. It would be better if /proc/.../geometry were removed. Every microsecond spent looking at its contents is wasted.

I think you incorrectly blame your problems on disk geometry. But I don't know the actual cause.

Geometry does not influence addressing mode. Moreover, mke2fs is a user space utility. You mention 2.4.21-rc3, and the ide code including this stuff changes quite a lot from version to version. What happens under 2.4.N for N=18,19,20? Both geometries given multiply out to 234372285 and 234374112 sectors, both below 234375000.

Does mke2fs actually ask for an access of sector 234441583? What size does *fdisk think the disk has?

Can it be that this was an attempted 48-bit command? (Sorry - no time to read 2.4.21-rc3.)

Andries