cvs commit: src/sys/conf files.i386 src/sys/pci agp_ati.c agpreg.h src/sys/modules/agp Makefile

[John Baldwin]

On Saturday 17 September 2005 04:44 pm, Nate Lawson wrote:
> Eric Anholt wrote:
> > On Sat, 2005-09-17 at 12:04 -0700, Nate Lawson wrote:
> >>Mike Silbersack wrote:
> >>>I think I asked in irc a while ago, but I can't recall the answer.
> >>>
> >>>I don't seem to have agp support on my laptop, which has:
> >>>
> >>>(--) PCI:*(0:2:0) Intel Corporation 82852/855GM Integrated Graphics
> >>>Device rev 2, Mem @ 0xe8000000/27,
> >>>0xe0000000/19, I/O @ 0x1800/3
> >>>(--) PCI: (0:2:1) Intel Corporation 82852/855GM Integrated Graphics
> >>>Device rev 2, Mem @ 0xf0000000/27,
> >>>0xe0080000/19
> >>>
> >>>According to Xorg.  Was there something odd about the chipset, or is it
> >>>just a matter of adding device IDs?  This is with RELENG_6.
> >>
> >>855GM and GMS are supported by the latest beta Xorg.  So you just need
> >>to update your Xorg-server port.
> >
> > We clarified in private mail that the agp device isn't attaching.  I
> > think acpi_video is beating it out, confirming that we really need to be
> > able to somehow share devices well for video.  That's 3 drivers now
> > (agp, drm, and acpi_video) that would all like to play.
>
> Ah.  Yes, I think we'll have to do something like what I did for cpu
> devices.  Basically, cpu0, cpu1, etc. implement a pseudo bus and
> multiple drivers attach below it.  Do a devinfo to see this.  There is
> no real way in newbus to have multiple drivers attached to a device if
> it doesn't implement a bus interface.
>
> This will require some design thinking.

Yes, a vgapci(4) driver or some such.  I've started tinkering in an (old) p4 
branch (jhb_agp) with moving the various agp drivers to be children of 
hostb(4) so that one would be able to kldload agp as a first step.

-- 
John Baldwin <jhb at FreeBSD.org>  <><  http://www.FreeBSD.org/~jhb/
"Power Users Use the Power to Serve"  =  http://www.FreeBSD.org