Michael Graff <explorer@flame.org> writes:

> IMHO, believing what the drive returns is just fine in most cases...
> 
> parameters extracted from in-core disklabel are:
> cylinders=3511 heads=11 sectors/track=108 (1188 sectors/cylinder)
In my particular case cylinders is just wrong, so you don't know
the correct number of cylinders for future sysinst/fdisk calculations.
You do know the capacity.

> 
> parameters to be used for BIOS calculations are:
> cylinders=259 heads=255 sectors/track=63 (16065 sectors/cylinder)
Large scsi disks and (often) large ide disks do this.  I'm not sure
that you must use these numbers if you are the only OS on the disk.
But clearly if you are not the first OS you must use them or you 
don't boot.
I have a fix to x86 disksubr.c that I'll commit soon that is
essentially for sysinst.  When it notices that the heads and sectors
are to be fudged, it also fudges the # of cylinders, so the disk
capacity appears consistent.  (If you have a real label, everything is
replaced by its contents, so the above change gets washed away.)