Extended BIOS translation (worth disabling?)

[Mike Isely]

On Thu, 5 Feb 1998, Peter Holzer wrote:

> On Thu, Feb 05, 1998 at 10:39:20AM -0500, Reinhold J. Gerharz wrote:
> > At 16:10 02/05/1998 +0100, Peter Holzer wrote:
> > 
> >         ...
> > 
> > >Yup. For some reason Linux fdisk (like DOS fdisk) only lets you start
> > >partitions at cylinder boundaries.
> 
> > This is probably to make the partitions compatible with LILO as well
> > as DOS. LILO's limitation is that it uses BIOS calls to actually load
> > the kernel, therefore inheriting the BIOS limitations. The advantage
> > of this is maximum and immediate compatibility with any disk drive
> > supported by the BIOS.
> 
> The BIOS handles arbitrary partition boundaries fine (all it gets are
> "read n sectors starting at cyl a, head b, sector c" requests. It
> doesn't even know about partitions). MS-DOS doesn't, but AFAIR it starts
> partitions on head 1 (instead head 0) of a cylinder, so either Linux
> fdisk cannot create DOS compatible partitions or it already has some
> special code for them.

I think this can be summed up by saying, yes, there are differences - with
extended translation on, you don't have to go through so many hoops to get
fine-grained partitioning of your disk.  But as far as I'm concerned, this
point is very small because why should one care about this granularity? 
We're talking about saving perhaps a few tens of MB on a disk with at
least 1024MB on it.  This is a 1% effect. 

The issue of bigger concern is the booting problem.  Without extended
translation on, everything that the PC BIOS touches and everything which
uses the BIOS to touch the disk, must be below 1GB.  That includes almost
every boot loader.


                        |         Mike Isely          |     PGP fingerprint
    POSITIVELY NO       |                             | 03 54 43 4D 75 E5 CC 92
 UNSOLICITED JUNK MAIL! |   isely @ pobox (dot) com   | 71 16 01 E2 B5 F5 C1 E8
                        |   (spam-foiling  address)   |