[Xorg] can't xorg detect on bootup which gfx card is installed?

[Adam Jackson]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Sunday 15 August 2004 17:59, Jakub Stachowski wrote:
> Dnia niedziela, 15 sierpnia 2004 21:45, Adam Jackson napisał:
> > The problem is that we have no way to bind a PCI ID to a driver before
> > the driver is loaded.  We don't have a database anywhere saying "this
> > chip id means i need this driver".  Clearly you as a human can look at
> > the pretty PCI name and know, but the server isn't that smart yet.  What
> > we'd want to do is put that list in the driver itself, so when we first
> > dlopen the driver we can ask it what PCI IDs it supports.
>
> So only PCI bus is important?

PCI encompasses AGP, PCI-X, and PCI Express as well.  You can also do sane 
device probing on Nubus, Sbus, MCA, in fact just about any bus besides ISA.

> > And in fact we have all the necessary information in the drivers already
> > - since they all have to check for supported chips somehow - it's just a
> > matter of factoring that out into a common data structure and teaching
> > the server to read that at module open time.
>
> It could be function - so for example fbdev could check if framebuffer
> device is available and report that it supports all devices (if fbdev
> available) or none (if not).

I think it makes more sense to consider kernel framebuffer devices as a bus 
type, and leave detection to the OS interface.

> > The problem here would be stupid buses like ISA, but I'm perfectly
> > willing to tell people insane enough to use ISA video cards that they
> > need to explicitly specify their driver or else be stuck with vga(4) or
> > vesa(4).  I think pretty much every other bus type (maybe excluding VLB)
> > is capable of sane self-identification.
>
> Some kind of 'priority' field would be useful - so only if specific (for
> given card) driver  is unavailable, server will fall back to less
> preferred, catch-all drivers as vga, vesa or fbdev.

This makes sense, particularly in the ISA case where we aren't really able to 
probe for the device.

- - ajax
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)

iD8DBQFBH+TJW4otUKDs0NMRAsTZAKDooC97L/ay7kW0b7YrttqtIa8HuwCgryqz
X0ASgq2qFR2VfTGlvXR6Ih8=
=DrBH
-----END PGP SIGNATURE-----