Hi, I was just wondering why the BIOS loads its own set of drivers in order to then load a boot-loader, then the OS loads its own 32bit drivers. Why doesn't the OS just use the BIOS's drivers; meaning, why doesn't the BIOS manufacturers work with OS manufacturer's to just write one 32bit bit driver for on-board/embedded devices i.e. NIC, video, keyboard, mouse etc (?) It seems like they are over-complicating the process by starting in 16bit mode and switching over.

I noticed this once when I could boot from a Network boot/PXE server in the POST process, but yet my fresh install of XP hadn't detected/installed my NIC in 32bit yet.

While this is a very good question it's not a Linux one. Moving along
to our General forum.


Cheers,
Tink

It's a PC related question. 'Nix/Windows, the OS isn't the focus.

Exactly my point. Thanks for emphasising that.

The bios does not have a protected mode interface (It should).

That was helpful...

Several reasons for keeping the BIOS isolated:

1) The BIOS is now just a minimal system to get the operating system going. If you were to implement everything the operating system required, it just means a larger BIOS (which adds development cost).

2) The BIOS is relatively hard/risky to update (compared to updating the operating system). It is better to keep the drivers on the disk where they can get timely security updates, bug fixes, and added features.

3) Since the BIOS must be mapped into RAM anyway (for speed reasons), there is no space advantage to using the BIOS drivers.

4) It would be difficult to decide on an API for BIOS supplied drivers, without matching some operating systems better than others.

5) Although there are some advantages to making drivers the responsibility of the manufacturer, the open pluggable nature of the PC hardware means that you cannot anticipate all the hardware that might be present. For example, a user might decide not to use the onboard network card.

Ideally the BIOS should be even smaller, though for historical reasons it does have an API and drivers (going back to the days when the BIOS really was the I/O software).

Thanks for the reply...Those are some very good points but why doesn't the BIOS manufacturers just increase the memory size and type of chip so that it can hold unix/linux/windows drivers and switch to a flashable ROM in case you would need to update drivers with usb key(?) Preloading the certified 32bit drivers from the POST the drivers would increase the load time slightly, but more importantly an installation of an OS would be much more simple and acquiring drivers generally wouldn't be an issue anymore.