> Followup to: <199712100251.VAA15925@jupiter.cs.uml.edu>
> By author: "Albert D. Cahalan" <acahalan@cs.uml.edu>
> In newsgroup: linux.dev.kernel
> >
> > > 1. Make the kernel know how to reset the video card to a known state.
> >
> > You can go from a known state to a known state.
> > You can go from any state to an unknown state -- useless.
> > Sorry, you can't get to a known state from an unknown state.
> >
>
> Sure you can. It's pretty common to do so. In fact, whenever you
> initialize hardware, this is usually exactly what you do.
That's what makes coding a driver for a secondary video card fun...
[BTW - Matrox's cards are nice about this AFAIK].
For the S3 cards you have to know: How much videoram you have, how fast
the videoram is (ram-timing) and what other timings are needed for that
particular manufacturer's card... BTW : this info is generally not all
that easy to get.... [though iirc most videoram is of common speed]
system BIOS knows how to tell video BIOS to initialize a videocard. This
is not true for secondary videocards... and asking the BIOS to initialize
those cards would crash as they'd want to be in VGA-mode.
Now I realize that not alot of people out there play with multiple
videocards.... so for them this point is moot.
[it's _VERY_ handy for CAD and VR though :]
Does this help explain some of it?
(I won't bother anyone about this on this list for another few
months... I don't really have anything more to offer for info..
personally even though I'm trying to work on GGI I haven't yet been sold
on why it's better than doing a usermode daemon - beyond the DMA and
IRQ-related stuff that is. Though libGGI is very handy as it'll run on
X as well as GGI/KGI [and soon svgalib I hope] without recompilation :).
G'day, eh? :)
- Teunis