Yves Godin wrote: > At 10:34 AM 4/15/02 -0600, you wrote: > >> On Mon, 15 Apr 2002, Yves Godin wrote: >> >> > For now I need to get my system fully working as kick as possible. >> I need >> > this proof of concept to show that linux on a Geode is able to do >> what we >> > need to do. Therefore I will use the kickest way for now. Later, I >> will be >> > able nd I will have the time to clean-up things. For example, I plan >> > re-write my code for the DOC Millenium Plus wich seems to be only half >> > supported in LinuxBIOS. >> >> the put the kernel you want to use in DoC. That will do it. >> >> What is missing in linuxbios for hte Millenium Plus? > > > I'am only talking for the Geode: > > 1. There is no IPL for the geode in the cvs tree. The code that > initialize the Geode memory in the cvs tree could only be executed > from the ROM. > 2. It seems that someone has work on the the Millenium Plus (there is > a lot of DOC_MIL_PLUS everywhee) but if you define this flag and try > to compile you will get a "doc_mil_plus.inc: No such file or directory". > 3. I think that the routines in the docmil_fill_inbuf do not work with > the Millenium Plus, I'll have to investagate why. I replace functions > in this file by some code that we have here. > We (SiS), requested by M-System Asia, did the porting to DoC Millenium Plus so that they can claim MPlus is LinuxBIOS compatable (TM). Unfortunately, M-System asked us to sign NDA to do them a favor !!! This is why you see a lot of #ifdef DoC_Plus stuff but no real code. This is pretty much like ADM K7 issue, just even worse.
P.S. No, we don't have the register spec, only thousands of lines of their garbage source code. I have to semi-reverse engineering. Fsck you M-System !!! Now I can use 256KB flash with IDE/Compact Flash. Ollie
