Hi,

Manuel Bouyer wrote:
> In theory maybe, but I got no ends of troubles with unaligned partitions.

Hmm, are you sure?  M$ fdisk happily starts the first partition on a
disk
at sector 1, head 1 of the disk (keep in mind here that while sectors
start counting at 1, heads start at 0), i.e. the first MBR partition
starts almost never on a cylinder boundary.

> It seems some BIOSes read the MBR themselves.

Yes, there are some BIOSes that do read the MBR partition table.  AFAICT
those try to deduce the geometry of the drive that was used when the
MBR partition table was created.  You get into trouble if the CHS values
for a partition in the MBR cannot be matched to the sector start/sector
count values for the same partition.  If you keep both of these values
consistent (well, if the disk size is below ~8GB; above that the CHS
values cannot match the other values anyway), you should be ok.

I've never seen problems when you keep this in mind (that's the reason
that I reduced the size of the fake MBR partition within our bootcode
from 5000 to 16 sectors, i.e. low enough so it will be within the first
track for any disk (hopefully) in order to avoid any conflict).

Ciao,
Wolfgang
-- 
ws@TooLs.DE     Wolfgang Solfrank, TooLs GmbH 	+49-228-985800