fdisk cannot partition correctly the Toshiba Mk2109 (2.5"
HD Drive ATA-4, 4200 cylinders, 16 heads, 63 sectors). I
boot with the arguments hda=4200,16,63 and fdisk gives the
correct info with p command.

I create partition (whatever their size) with fdisk without
problems BUT as soon as write the results and enters again
in fdisk I get an error (for a partition going starting
at cylinder x and ending at cylinder y):

Partition N does not end on cylinder boundary:
phys=(y-2,14,63) should be (y-2,15,63)

fdisk behave like if it badly manages the last head and
stops 2 cylinders before the end. Worse, if y > 1024 then
the previous message becomes whatever y is:

Partition N does not end on cylinder boundary:
phys=(1022,14,63) should be (1022,15,63)

I tried to change the heads number to 15 within the extended
mode of fdisk but it still gives the same error but at heads
13 instead of 14.

Even stranger when using the redhat installation
mount the partition and write to it as soon as
I reboot the computer, I completely loose the partition:
it is completely corrupted (not even recognized as an ext2
partition)

The problem is only for the Mk2109 as fdisk succeeds to
partition a Toshiba Mk 2104 2.5 " disk (4200 cylinders, 16
heads, 63 sectors) with the same hardware configuration.

Thanks for your help,