DRI on iBook G4

[Magnus Henoch <mange%freemail.hu@localhost>]

I wanted to see if the DRI support in the NetBSD kernel (from netbsd-5
branch) is good enough for macppc.  So I added:

radeondrm* at radeonfb?
options DRM_NO_MTRR
options DRM_NO_AGP

to my kernel configuration.  I also had to patch an i386-ism:

--- dev/drm/drm_drv.c.~1.19.~   2008-08-19 20:06:41.000000000 +0100
+++ dev/drm/drm_drv.c   2008-11-30 00:48:34.000000000 +0000
@@ -409,7 +409,13 @@
 
        for(i = 0; i<DRM_MAX_PCI_RESOURCE; i++) {
                if (dev->pci_map_data[i].mapped > 1) {
-                       bus_space_unmap(dev->pci_map_data[i].maptype,
+                       /* pci_map_data is pci_map_data_t.  maptype is int.
+                          on x86, bus_space_tag_t is typedefed to int,
+                          but on powerpc, it is const struct powerpc_bus_space 
*.
+                          thus, powerpc's bus_space_unmap doesn't like
+                          being passed an int.  So we try what the other
+                          bus_space_unmap call did. */
+                       bus_space_unmap(dev->pa.pa_memt,
                                        dev->pci_map_data[i].bsh,
                                        dev->pci_map_data[i].size);
                        dev->pci_map_data[i].mapped = 0;


I recompiled modular-xorg-server (yes, my X is all pkgsrc) with the dri
option, created /dev/dri/card0, and restarted.  glxinfo still says that
I don't have direct rendering though.

From dmesg:

radeondrm0 at radeonfb0: ATI Radeon RV280 Mobility (unit 0)
radeondrm0: Initialized radeon 1.26.0 20060524

From Xorg.0.log:

drmOpenDevice: node name is /dev/dri/card0
drmOpenDevice: open result is 8, (OK)
[drm] failed to load kernel module "radeon"
(EE) RADEON(0): [dri] RADEONDRIGetVersion failed to open the DRM
[dri] Disabling DRI.

Any ideas?  Of course, the log messages give some clue, but I'll take a
break now, and hope that someone magically solves the problem until I
get back :)

Magnus