Radeonhd and Acceleration. Compile problem

business_kid · 08-25-2009, 10:41 AM

I found what docs there are on the radeonhd driver.
http://wiki.x.org/wiki/radeonhd

Getting into this, I now have a 2.6.30.5 kernel and radeonhd-1.2.5
git clone git://anongit.freedesktop.org/mesa/drm
gave me a drm/ directory with a hacked mesa, I gather. Next they say to
cd drm/linux-core
git checkout -b r6xx-r7xx-support origin/r6xx-r7xx-support
make radeon.o drm.o
That ended this way:

bash-3.1$ make radeon.o drm.o
make DRM_MODULES=radeon.o modules
make[1]: Entering directory `/home/dec/download/drm/linux-core'
make -C /lib/modules/2.6.30.5-sd64/source SUBDIRS=`/bin/pwd` DRMSRCDIR=`/bin/pwd` modules
make[2]: Entering directory `/usr/src/linux-2.6.30.5'
CC [M] /home/dec/download/drm/linux-core/drm_auth.o
In file included from /home/dec/download/drm/linux-core/drmP.h:84,
from /home/dec/download/drm/linux-core/drm_auth.c:36:
/home/dec/download/drm/linux-core/drm_os_linux.h:36: error: conflicting types for ‘irqreturn_t’
include/linux/irqreturn.h:16: error: previous declaration of ‘irqreturn_t’ was here
make[3]: *** [/home/dec/download/drm/linux-core/drm_auth.o] Error 1
make[2]: *** [_module_/home/dec/download/drm/linux-core] Error 2
make[2]: Leaving directory `/usr/src/linux-2.6.30.5'
make[1]: *** [modules] Error 2
make[1]: Leaving directory `/home/dec/download/drm/linux-core'
make: *** [radeon.o] Error 2

Anyone spot the error? Is this a kernel headers conflict? I am a little shy of updating the headers again, as I started with headers from 2.6.27.x, went to 2.6.29.x, and now apparently need to update to 2.6.30.x. I fell it's only a matter of time before I go down ingloriously with crazy errors.

business_kid · 08-26-2009, 02:28 PM

Further on this:

/home/dec/download/drm/linux-core/drm_os_linux.h:36: error: conflicting types for ‘irqreturn_t’
include/linux/irqreturn.h:16: error: previous declaration of ‘irqreturn_t’ was here

One was typedef void, while the other was typedef enum. In short, I had the most recent kernel headers(2.6.30.5) and I altered the source, commenting out the 'typedef void'.

The compile staggered on and fell over before it was finished. I refused to chase the second (wildly different) error. This stuff obviously needs work.