Upgrading the BIOS makes the hard disk unbootable: why?

[Jean Castonguay]

The motherboard is an ATC 2000 equipped with an Intel 430 HX chipset.

The BIOS is an Award Modular BIOS v4.51
(07/26/96-i420HX-2A59FA39C-00).
This BIOS does not «see» beyond the 8 GB limit.

The hard disk is a Western Digital Caviar WD800 (80 GB).

On it, I installed OS/2: the bootable partition (and several others)
is below the 8 GB limit. Several other partitions, above the 8 GB
limit, become accessible as soon as the OS/2 hard disk device driver
is loaded.

To have more flexibility, I upgraded the BIOS so that it «sees» hard
disks up to 128 GB.

When I restarted the machine, it recognized the whole hard disk but
OS/2 does not boot. The hard disk device driver crashes!

How can this be explained? Upgrading the BIOS does not move a single
bit on the hard disk. I use LBA addressing which means sectors are
indexed from 0 to N-1.

Thank you for your enlightenment.

 

[YKhan]

The previous BIOS used CHS addressing for the hard disk, while the new
one uses LBA. CHS and LBA don't map the sectors in the same order as
each other. The hard disk device drivers maintain the addressing scheme
handed to them by the BIOS, to maintain legacy support, even when they
are extending it beyond the BIOS's capacity.

Yousuf Khan

 

[Jean Castonguay]

The previous BIOS used CHS addressing for the hard disk, while the new
one uses LBA. CHS and LBA don't map the sectors in the same order as
each other. The hard disk device drivers maintain the addressing scheme
handed to them by the BIOS, to maintain legacy support, even when they
are extending it beyond the BIOS's capacity.

Yousuf Khan

The previous BIOS offered LBA addressing, and I asked for it. If it
used CHS addressing; it lied to me :-(

 

[James J. Weinkam]

The previous BIOS used CHS addressing for the hard disk, while the new
one uses LBA. CHS and LBA don't map the sectors in the same order as
each other. The hard disk device drivers maintain the addressing scheme
handed to them by the BIOS, to maintain legacy support, even when they
are extending it beyond the BIOS's capacity.

Yousuf Khan

The sectors are mapped in the same order but the cylinder boundaries are in
different places depending on the lies told.

 

[Yousuf Khan]

The previous BIOS offered LBA addressing, and I asked for it. If it
used CHS addressing; it lied to me :-(

More than likely it lied to you. But probably not a full-on bald-faced
lie, just a white lie. The early LBA BIOSes tried to make the LBA
sectors look like CHS sectors. I'm not entirely convinced that there was
ever a standard technique for mapping this conversion. Most operating
systems expected disks to be presented to them in CHS fashion, that's
why there was this conversion stage. It wasn't until later that
operating systems started understanding LBA directly, and therefore they
no longer required the conversion back to CHS format again.

Yousuf Khan