No 3D acceleration: "DRI initialization failed!"

[Hugo Mills <hugo-fglrx at carfax dot org dot uk>]

   I'm having some problems getting the 3D acceleration to work.

Platform: AMD64, 2.6.12.5 kernel, X.org 6.8.2

   I've followed Flavio's excellent instructions; I found a patch
which allows the drivers to build against 2.6.12, and all seemed good
at that point.

   Now for the unpleasant bit.

   With "UseInternalAGPGART" "no" in my xorg.conf, I can get the
driver to load successfully (and start X) with either agpgart,
amd64_agp and fglrx loaded, or nothing at all. In either case, my X
log file reports:

(II) fglrx(0): Kernel Module Build Time Information:
(II) fglrx(0):     Build-Kernel UTS_RELEASE:        2.6.12.5
(II) fglrx(0):     Build-Kernel MODVERSIONS:        no
(II) fglrx(0):     Build-Kernel __SMP__:            no
(II) fglrx(0):     Build-Kernel PAGE_SIZE:          0x1000
(II) fglrx(0): [drm] register handle = 0xfd400000
(EE) fglrx(0): [agp] unable to acquire AGP, error "xf86_ENODEV"
(EE) fglrx(0): cannot init AGP
(II) fglrx(0): [drm] removed 1 reserved context for kernel
(II) fglrx(0): [drm] unmapping 8192 bytes of SAREA 0xffffc20000710000 at 0x2aaaaaad3000
(WW) fglrx(0): ***********************************************
(WW) fglrx(0): * DRI initialization failed!                  *
(WW) fglrx(0): * (maybe driver kernel module missing or bad) *
(WW) fglrx(0): * 2D acceleraton available (MMIO)             *
(WW) fglrx(0): * no 3D acceleration available                *
(WW) fglrx(0): ********************************************* *

   If I set "UseInternalAGPGART" "yes", and remove agpgart.ko and
amd64_agp.ko, the fglrx kernel module does not load (which I think is
the correct behaviour, if I understand which bits go where). If I
remove only amd64_agp.ko, then fglrx loads (bringing in agpgart), but
X hangs in uninterruptible sleep ("D state") when it starts. The last
text in the log file after the hang is:

(II) fglrx(0): [drm] DRM interface version 1.0
(II) fglrx(0): [drm] created "fglrx" driver at busid "PCI:1:0:0"
(II) fglrx(0): [drm] added 8192 byte SAREA at 0xffffc200007a3000
(II) fglrx(0): [drm] mapped SAREA 0xffffc200007a3000 to 0x2aaaaaad3000
(II) fglrx(0): [drm] framebuffer handle = 0xe0000000
(II) fglrx(0): [drm] added 1 reserved context for kernel
(II) fglrx(0): DRIScreenInit done
(II) fglrx(0): Kernel Module Version Information:
(II) fglrx(0):     Name: fglrx
(II) fglrx(0):     Version: 8.14.13
(II) fglrx(0):     Date: Jun  8 2005
(II) fglrx(0):     Desc: ATI FireGL DRM kernel module
(II) fglrx(0): Kernel Module version matches driver.
(II) fglrx(0): Kernel Module Build Time Information:
(II) fglrx(0):     Build-Kernel UTS_RELEASE:        2.6.12.5
(II) fglrx(0):     Build-Kernel MODVERSIONS:        no
(II) fglrx(0):     Build-Kernel __SMP__:            no
(II) fglrx(0):     Build-Kernel PAGE_SIZE:          0x1000
(II) fglrx(0): [drm] register handle = 0xfd400000

   The only other thing which I can see that looks "unusual" in the X
logs is:

(II) Loading extension GLX
(II) LoadModule: "dri"
(II) Reloading /usr/X11R6/lib/modules/extensions/libdri.a
(II) UnloadModule: "dri"
(EE) Failed to load module "dri" (once-only module, 0)
(II) LoadModule: "int10"

   It definitely recognises the chipset correctly (Radeon 9600 Pro),
but says:

(--) fglrx(0): (PciSubVendor = 0x1458, PciSubDevice = 0x4022)
(--) fglrx(0): board vendor info: third party graphics adapter - NOT original ATI
[...]
(WW) fglrx(0): board is an unknown third party board, chipset is supported

   Any suggestions for where this has gone wrong?

   Hugo.

-- 
=== Hugo Mills: hugo at  dot  dot  dot  carfax.org.uk | darksatanic.net | lug.org.uk ===
  PGP key: 1C335860 from wwwkeys.eu.pgp.net or http://www.carfax.org.uk
     --- Once is happenstance; twice is coincidence; three times ---     
                            is enemy action.