[Quickos-general] Re: Questions & Suggestions
Status: Planning
Brought to you by:
azhobbes
From: Andreas Z. <azh...@gm...> - 2001-11-07 21:12:13
|
Aleksey Saushev wrote: > Well, I think it is better to suppose O.S. 386 CPU > dependent. Not to write "as portable as we can". Yes. 386sx...P4...K7/8/9/10 > About peripherial requirements. I asked if we need program > anything except FDD/HDD and console (at first & in future). VGA,HERC,CGA?,TIMER, my special hardware > Also, should I relate the "quickness" to computational > or to peripherial work. It think theres no problem for both ones. > I mean what to plan: > fast task switching (cooperative or preemptive) i prefer mix cooperative or preemptive - some task has to be cooperative( ... need so write 128 bytes within 250?sec... ). I want a my DOS back, 32bit advadages and multitasking. > or better disk caching policy. later. that not i think at this moment. if our kernel works, we could optimize in this way. > Suggestions. > 1. If we want really quick OS, we have to forget about > swapping, paging, protection. seldom if we need "hard"-realtime - therefor must forget swapping, paging. But most time we can go with preemptive and sometimes we can switch to cooperative ("soft-realtime"). I think if we want to produce a safe os, we must present a preemptive multitasking system with paging. if that works fine we can insert cooperative tasks. First a want a safe OS, i dont want to go the way which win3.1 win9x has gone. Only if you really need more control over you processor - then you can switch down protection. Most time we dont need. > Q: What do we want to throw away? > What do we want to see working? First I want a working preemptive-pageing-kernel. > 2. Booting from FAT. yes. or from ROMFS because i want to start this os diskless. just from ROM link network cards do > 3. FAT support. > Q: Do you have any doc/code showing I/O from > protected mode? YES. i have a working pmode-demo. > 4. Flat memory: no paging, no swapping, single address > space. its difficult to share data-memory among tasks (in a save way) without paging. for most tasks i prefer paging. > 5. Operating shell for interractive debugging. > (I think FORTH is the best way to organize it.) I dont believe that its the best way. But seeing is believing. :-) -- Mfg Andreas Ziermann b...@lo... -- http://www.loet.de |