OSDev.org

is there a way to crank both processor FSB and memory timings (with or without voltage alterations) in a protected mode enviroment?

The way I understand it, the BIOS is the only part that can make those decisions.

and if so, could this be implemented into an actual functional overclocking feature later in kernel dev? when I say functional, I dont mean like the software utilities for win/lin that can 'soft-clock' your system, I mean something far lower-level that actually interacts directly with the appropriate pins.

Hi,

This is a feature that more and more mobo manufacturers are adding to their systems. Some of the Windows drivers for the appropriate boards can actually fully overclock timings and voltage settings (I am unsure of specific boards, but I believe there are Gigabyte boards that do this?). But yes, you can certainly interact with the board in protected mode.

Here's the catch: As with any other device, you need to write a driver to do it. If the specifications for the device are closed-source, you're basically stumped. And I guess you wouldn't want to try reverse-engineering and experimenting with stuff that could basically fry your CPU, mobo and RAM

Cheers,
Adam

I have MSI 845E Max and it has program called Fuzzy Logic 4, that program can overclock CPU FSB, change the cpu multiplier (if mobo supports it), and tweak the voltage of CPU, DDR and AGP. It has some drivers like rushtop.vxd, interesting Anyone for dissasembling? By the way, its a windows program, so what? It can oclock and change voltages, display the CPU temperature and can auto-overclock (increase the CPU FSB, simple graphics test and all over again until PC freezes and reboots). After this auto overclock thing you can set it fixed by Windows startup or in BIOS. I prefer in BIOS though

Or download BIOS from manufacter's site and disassemble it for your personal use. Not sure if it's legal, but if you aren't selling your OS I think it's OK. Correct me if I'm wrong.

Regards
inflater

On average, reverse engineering (the correct way, not by copying blocks of disassembled code) is legal unless you had to agree to a license that instructed you not to. Even if you did, some countries' laws may allow you to do so anyway. You may even charge money for your efforts.

Just tried to download my motherboard BIOS and disassemble the 256kB thing. And I'm sure I would never do it again - There was some "out dx,al" but I didn't knew what the hell is in DX and AL. I think DX=0, because there was no opcode with DX (at the first lines of code). Ranging from 0x0 - 0x1D86, I saw only "db 0xFF". Something about "256 kB alignment", this reminds me of my kernel - I must say, disgusting code

If you find some code that can change FSB on MSI mobo's, please post it here. Thank you.

Regards,
inflater

@inflater, just so you know.. several BIOS manufactures compress their updated firmware, It may contain several different files as well.. (Award BIOS is famous for that..)

So disassembling the firmware "probably" will require extracting the correct components.. (I think the LZH algorithm may be common.. just remember that the archive may include multiple files & archives - EPA logo?.. ACPI tables.. etc)

It will take time to figure out the format of the firmware.. but when and if you figure it out.. try documenting some of what it does, you could then release unofficial updates adding new functionality etc lol.. be prepared for CRC checks etc though..

Here is a nice article I found, Award BIOS related though..
http://www.geocities.com/mamanzip/Artic ... _guide.htm
(Disclaimer.. If this is illegal in your country... yada yada yada.. just don't click.. )

Brynet-Inc wrote:http://www.geocities.com/mamanzip/Articles/Award_Bios_RE/Award_Bios_RE_guide.htm

I get a file not found error on that

Apologies, I'm incompetent..
http://www.geocities.com/mamanzip/Artic ... guide.html

Have fun..

Of course when opening the 256kB file in Notepad I didn't saw any human-readable ASCII string except the BIOS manufacturing date. So it must be compressed, and even looking at "mov si,0x9eee out dx,al"... ehh...
LZH compression you say?
BTW it's AMI BIOS. Oh, and its newer that my mobo has , would I try flashing? Ahh, everything is working normally and I do not use S3/STR mode

Brynet-Inc wrote:(Disclaimer.. If this is illegal in your country... yada yada yada.. just don't click.. )

You thought I care?! But anyways, I'm not that skilled to glaze at a gross and big assembly code

inflater

Yes LZH/LHA, and as a Windows user you should be able to extract it under "WinZip" or "WinRAR".

But there may actually be multiple archives, so it may require manual separation

(This is all Award specific, I have no idea how your BIOS is structured..)

100% sure this version of MSI AMI BIOS isn't LZH-compressed (?). Neither Winrar or Total Commander couldn't open it, nor extract it. Didn't contain a valid signature I think (the first 7,3 kBs are "empty" - wonder what's doing the 0xFF there). Or is it? Maybe in different offsets

Regards
inflater

Yes, As I said - there may be multiple archives.. the firmware may also be mixed as well..

Clearly it'll take time to reverse engineer it.. (Google may have some related articles, perhaps someone else has succeeded?)

even if it is compressed, it might not contain the headers needed for an archiver to recognize the format. It means you should decode it manually to see what it spits out. Essentially you're only sure its not archived in a way winrar is familiar with.

since it is exactly 256k, it can just as well be just uncompressed. Try looking for raw video memory pieces that represent text (i.e. character, attribute, character, attribute, character, attribute, ...) instead of normal strings

I agree completely with you two. May be compressed on different offsets, may contain video ram pieces... Is it worth time to disassemble BIOS? If we just ask MSI/Gigabyte/ASUS/ECS /PC Chip /SiS/VIA/... to tell us how they are changing CPU FSB and claiming we aren't their competetive company (:D) but we are "students, who are just learning to program in pmode" - no one knows, maybe they will answer that to us - Who wants to be first?

Regards
inflater

The reason I asked this is because of the older-style mobo's that had the dip switches that could control your FSB and the fsb/mem mhz ratios. I'm not a big fan of soft-clocking, but I would be much more open if the soft-clocking worked just as good as bios OC'ing and such, and if it is possible, I would rather implement this feature now while my kernel is still young, than wait later and have to weave through the protected rings and ect...