Re: A snapshot of the LuxAsm developments

From: wolfgang kern (nowhere_at_nevernet.at)
Date: 12/19/04 

Date: Sun, 19 Dec 2004 17:38:35 +0100

"The Half" skrev:

[]
| > I like the parameter passing by registers or by pointers
| > instead of the brainless double-copy stack pollution.

| (Run away in shame)

:) why?

| > But when I look at the Linux API and the HW-libraries
| > coming with it, I feel very sad for the missed opportunty
| > to do somehow better than M$-windoze.

| Please explain. What oportunity is missed ????

I really don't know much about Linux, but what I saw so far,
it doesn't allow (beside other safety caused detours):
The unrestricted, undelayed direct I/O access, and
I mean it should always have IOPL3 for all rings.
After comparing a few drivers, it looks like some 'new' Linux
drivers are just translated from the windoze stuff.

| > Hardware-drivers are the main speed- and performance-factor
| > of an OS and it's tools. So I would take care about optimised
| > HW-drivers (incl. MEM.mgt) first, and not creating a tool
| > which just uses what's already there without checking on it.

| Once before we discussed theese hw.drivers. Why not use your hex talents
| at extracting the needed information from the drivers directly ? I am sure
| that is legal ? If you will not do it, then how the hell would I be able
| to? But if I could read code like you, thats what I would do.
|
| At least teororetically, should it be fully possible ?

Yes, but a heavy time-consuming story, and I may have done it too often.
And that's the point why KESYS doesn't support 'all' hardware around.
I can only code for known hardware, so I decided to not support all
thoose "buy and die" products which doesn't come along with detailed
hardware descriptions anymore.

__
wolfgang

Relevant Pages