Alan Coopersmith wrote:
> Cool - though I doubt we'll ever see a SPARC machine with Intel graphics
> on the motherboard, I understand that's the only DRM module we actually
> have checked into ON right now. 

It was meant to be an interim step in getting mach64, r128, radeon and 
ffb DRM kernel level support working, as described.

Besides: (At least few of) those Intel chipsets are not exclusively 
shipped on-board.
And any Non-AGP card could also be plugged into an U25 or U45.
And levaing PCI-E alone: PCI-X or legacy PCI2.2 cards would fit into any 
PCI based SPARC.
So in theory even an Intel frame buffer inside a SPARC would be possible 
and would *work* for X11-only (no console, when no FCode support).
I successfully got several different mainstream x86 cards to work on 
SPARC-Xorg, that's possible (with a few SIGSEGV's at first). Only that 
the screen remains black before Xorg is going to start, because those 
x86 cards normally expect to get initialized by a x86 BIOS (using native 
x86 machine language). However, if Sun had really wanted to get 
mainstream graphics support to SPARC they could have done, what 
Freescale has achieved on ppc: Emulating the x86 frame buffer 
initialization by IEEE1275 OpenFirmware !
Any generic card would then work and x86 VBIOS could be read 
out/initialized properly.


It is only. that the SB1500/2500 and U25/U45 ' s pci bridges are not yet 
supported by Xorg's bus scanning. As those 4 systems are totally 
unsupported by the Linux or BSD kernels (and OpenSolaris' PCI code for 
them is still closed source) it is practically impossible to extend 
sparcPci.c to make it work on them.
Correct me if I am wrong, but the last time I looked those pci bridges 
were still not available in src (Tomatillo is something else I believe, 
but not 100% sure, need to re-check).


> (Jay is actually investigating what it
> would take to just build the entire DRI/DRM tree on Solaris, so we may
> be getting more soon.) 

That sounds good.


%m

Reply via email to