Re: Award BIOS / Linux (XFree86) / Geode questions

> BIOS 1.21 or earlier does not exhibit this problem, *BUT* because of > event #2, any BIOS earlier than 1.23 will play back audio > approximately 10% too fast on currently shipping hardware. So we can't > just backlevel boards to a working BIOS version. > > We have alternative sources, and we're working to activate them, but > it means changing some mechanical bits and that has an annoying > leadtime. > > I've spent many hours comparing BIOSes using AwardMod and a > disassembler. All the BIOSes I have examined from 1.14 through 1.21 > inclusive have the same VSA code, vsa.vkd, binary identical. 1.22 and > later have *radically* different code (gx55306.bin) - subroutines > shuffled around, a lot of changes. Clearly, a hell of a lot more than > just a little serial Rx code was changed.

Just an idea - maybe applying the audio fix from 1.23 to the code of 1.21, means you only have to find the audio block in 1.23 ? Have you tried talking with the Award bios people, or is Advantechs a unique varaint ?

-jg

Hi Jim,

This is basically what I want to do. However it's apparently not just a matter of changing a timing constant somewhere. A lot of stuff has been shuffled around between 1.22 and 1.23. Swathes of code, jump tables and other data are different, even in unrelated areas.

The code is also very difficult to understand because it's a combination of special SMM-mode code and regular code (e.g. code that I thought was already running in SMM is writing to ports that are trapped for SMM access - I thought this kind of double fault was illegal).

Also my question is: is that ROM segment checksummed over and above the LHA compression integrity check? Can I hack with impunity?

Didn't think of it, to be honest. I'm pretty sure they'll either not respond or give me a cookie-cutter that says it's an OEM-customized piece of firmware and they have no idea. But I'll give it a try, thanks for the suggestion. Advantech claims it's a unique piece of code, BTW, which I'm sure it is for small values of "unique".