ERROR: line 125, Function intelInitDriver, File intel_screen.c

[Albert Vest]

On Wed, 07 Dec 2005 12:27:17 +0300
Denis Shaposhnikov <dsh at vlink.ru> wrote:

> >>>>> "Eric" == Eric Anholt <eta at lclark.edu> writes:
> 
>  Eric> You might want to try plain graphics/dri.  If that works and
>  Eric> dri-devel doesn't, it would be great if someone told me so I
>  Eric> could spread the word.
> 
> Ok, direct rendering work on my system with x11-server/xorg-server
> and graphics/dri-devel only. It doesn't work with any other
> combinations.

Denis,
Hearing your success I had to try that myself (ATI RS300 IGP 9100).
DRI loaded without blanking my screen and deactivating my keyboard, so I started tweaking xorg.conf.
After I started getting complaints about "PCI GART" I decided to try setting option "BusType" "PCI" (even though the agp0 device clearly activates, drm recognizes it, etc.).  Surprisingly it works.

Now I get "direct rendering: Yes", and glxgears reports well over 600 FPS, though it looks like they're turning more slowly than before (probably fewer skipped frames now?).

I also modified some files under dri-devel/work/* which may or may not have helped (such as adding CHIP_R200| to the definitions of the RS300 chips in drm_pciids.txt)

Full files/details will be provided on request to anyone interested.

$ dmesg | grep drm
drm0: <ATI Radeon RS300 IGP> port 0xc000-0xc0ff mem 0xd0000000-0xdfffffff,0xe4020000-0xe402ffff irq 16 at device 5.0 on pci1
info: [drm] AGP at 0xe0000000 64MB
info: [drm] Initialized radeon 1.16.0 20050311 on minor 0
info: [drm] Loading R200 Microcode
drm0: [MPSAFE]

Thanks,

-- 
Albert Vest, al vest at earth link dot net