This is one of the advanced topics that were the main reason about why I wanted to learn x86 Assembly. However, due to lack of patiente and motivation, I simply got stuck and didn't tried harder anymore, though I'm still interesed in looking at these advanced topics but not from the raw code standpoint. So that is what this is about.

As many know, BIOS Source Code is propietary, so making the smallest change to it would requiere dissamble the binary file and an expert guy with an smart brain to understand it. However, some people started to develop they own BIOSes from scratch. That is where LinuxBIOS ( http://en.wikipedia.org/wiki/Linuxbios ) comes from, that seems to be in good shape and gaining some ground. It got interesing concepts like the fact that the BIOS itself runs in 32 Bits mode and that it uses Cache-as-RAM, that I suppose that could allow machines with LinuxBIOS to boot with no RAM installed (Interesing for debug purposes, it may make easy to detect an uncompatible or defective Memory Module if you don't have another one to test). Another interesing fact is that the beggining part of the BIOS is done with a custom compiler that uses only Registers and not RAM space until it initilizes the Cache-as-RAM, then it uses regular C code.
Such types of projects looks interesing, but I didn't saw a custom project about a BIOS from scratch (Though they are few and far bewthem) with FASM and I am interesed in know if someone is planning to do this or there is more data to read about the matter.

*BUMP*

Seriously, no one got anything to comment? The BIOS is one of the few things that are pratically for Assembly programmers only and one of the most interesing targets for projects and research.

zir_blazer - I applaud you on your inquizitive mind and spirt. You certainly have chosen the path less traveled.

Have you considered investigating developing for the emerging BIOS standard known as UEFI (Unified Extensible Firmware Interface)? To me, UEFI holds much promise and some VERY interesting capabilities.
If I weren't so busy with this damn hypervisor, I'd be writing some UEFI code.

Here are some links for you to explore:

http://www.intel.com/cd/software/products/asmo-na/eng/219678.htm
http://www.uefi.org/events/EBC_Summit_invite_Final_2_.pdf
http://www.uefi.org/events/
http://blogs.phoenix.com/
http://www.insydesw.com.tw/en/Ps.asp?id=178

I would also like to add, that DexOS is like a 32bit BIOS, alot of people call it a 32bit Dos (including me), but the more you understand Dos, the more you see that most of the work is done with BIOS functions.
How the BIOS co, let M$ get away with all the $$$$$ from Dos, is beond me.

So to answer your ?, in a way Team DexOS is working on a 32bit BIOS.

I have readed many things about EFI, however, I don't place my bet in it.

http://en.wikipedia.org/wiki/Extensible_Firmware_Interface




For this sole reason, forget it. The potential of modernization of today BIOS is nice, but what we risk to lose is not. A open source developed BIOS with mainstream support (Well, LinuxBIOS is basically that) is needed.

Besides, even with EFI around, it would be needed a way to upgrade current Motherboards BIOSes to support useful features or correct bugs in them, not just something that comes with a new platform (This would requiere tons of reverse engineering for compatibility though). For example, considering that many BIOSes bring a FLASH chip with 2 or 4 MB capacity, I don't know myself why they can't bring useful tools in them like a BIOS version of Memtest, or a partition manager that can format them itself. Botting an OS or flashing the BIOS from an USB Pendrive is extremely nice also, so we can get rid once and for all of the 3 1/2 Diskettes Drive. Or even a small OS in the BIOS chip itself. There are many great things than the BIOS may be capable of doing if we give it a chance.

this sure is a cool combination.

there is AM2 Gigabyte mainborad support now in LinuxBIOS which allows assembly routines in the source code.

It would be useful to defeat any malware running in system management mode for some reason (DRM or so).

OS in ROM (meuetOS) is a poss, too. Or even Pacifica support.

I think that the next BIOS standard should support the old BIOS..so that it is backwards compatable, and also have way to check if the new bios thingy is supported..

I think it would be good to have a switch (jump) and have two chips (old & new), even better if they can be remove esay.

On my xbox chip i had 4 switchs so i could use 4 differant BIOS.

maybe a thing or two from these articles will give you some hints with the current BIOS code:

http://www.geocities.com/mamanzip/Articles/Award_Bios_RE/Award_Bios_RE_guide.html

http://www.geocities.com/mamanzip/Articles/award_bios_patching/award_bios_patching.html

Funny, I'm working on one of my own BIOS substitute, GreenOmega 64 AIOS (Advanced Input / Output System) - this one isn't really a BIOS - may be qualified as a 64-bit BIOS.

Although about the OS, it would basically shut Long Mode down before loading the boot sector onto Athlon 64's L2 cache to boot it up (in Real Mode session, to be exact)

I also plan to put in Blu-ray Disc boot support, and a 3D GUI too.
(Also, don't forget the complete USB emulation for DOS. Like port 209h for
detected USB gamepad, for example.)

Any source code would be appreciated, since I'm not sure about how the BIOS boots up. (Although I'm glad that I have Biostar K8T80-A7 motherboard, since it has all open-source chipsets, like VIA K8T800. I'm planning to flash it in this motherboard.)